The basic form of multi use of to divide up a reason for their use is222 © cuartens MULTIPLE ACCESS Guard band evi] [exe] ems |/[ene] “——— Transponder bandwiath Frequency Las (Guard time cht | |on2] [ona |/| cha Frame period => Time. TDMA jaiion AL FIGURE 6.1. Multipte access techniques: FOMA, TOMA, avy CDMA. Note that in te aa = . sequence form of COMA shay, aT Frequncy bere. all he channels overapy ‘ CDMA both time and frequency, : A = capacity is maximized, bandwidth is used efficiently flexibility is maintained, and cosy : the wser is minimized while revenue to the operator is maximized. The multiple acco system should also allow for changing patterns of trafic over the 10 or 15 years of he F expected lifetime of the satellite. Usually, all of these requirements cannot be satisfied OZ the same time and some may have to be traded off against others, ACCE There are three basic multiple access techniques, illustrated in Figure 6.1. In fg uency division multiple access (FDMA) all users share the satellite at the same time, bu ach user transmits at a unique allocated frequency. This approach to sharing the frequency spectrum is familiar to us all, as itis the way that radio broadcasting has always shared ‘the air waves, Each radio station is allocated a frequency and a bandwidth, and transmits iis Signals within that piece of the frequency spectrum, FDMA can be used with analog or digital signals. In time division multiple access (TDMA) each user is allocated a unique time slot atthe satelite so that signals pass through the transponder sequentially. Because IMA caus in transmissio® itis used only with digital signals. In code division ‘multiple access (CDMA) all users transmit to the satellite on the saitie frequency and at the same time, The earth stations transmit orthogonally coded spread spectrum signals that can be separated athe receiving earth station by correlation withthe transmitted code. ‘CDMA inherently a digital technique. In each of the multiple access techniques, some unique property of the signal (frequency, time, or code) is used to label the transmission such that the wanted signal can be recovered at the receiving terminal in the presence of all other signals. ‘The distinction between multiplexing and multiple access is sometimes blurred _Mul- tiplexing applies to signals that are generated_at one location, whereas multiple access refers to signals from a number of different geographical locations. For exaifiple, an carth Station might use time division ‘multiplexing (TDM) to create a high-speed digital data stream from many digital speech channels delivered to that earth station, and then mad- ulate the data stream onto an RF carrier and transmit the cartier to the satellite. At the satellite, the carrier can share a transponder using time division multiple access (TDMA) or frequency division multiple access (FDMA) with other carriers from earth stations any- where within the satellite's coverage zone. The resulting signal is called ‘TDM-TDMA or TDM-FDMA. Note the distinction between TDM and TDMA: signals at one earth sta- tion are combined by multiplexing, and then share a satellite transponder with signals from other earth stations by multiple access.2 group of earth stations may access part of the bandwidth A: while other TDMA groups of earth stations share differ. snaPonder bandwidth, This approach has been used in both VSAT Systems. (See Chapter 9.) Demand assignment can also be used with number of signals in the j{ansPonder at any one time, The Globalstar with demand assignment’, TDMA fechniques)are sometimes called EDM “(MF-TDMA). In the sections \é CDMA as fixed assignment schemes, le access, Ng ePhOnY: All signals were analog, and analog multi- £2 combine large numbers of telephone channels into a be modulated onto a single RE cartier. Individual tele- Frequency from baseband to a higher frequency 50 that * channels using frequency division multiplexing (FDM). vg Channels to the frequency range224 CHAPTER 6 MULTIPLE ACCESS ‘Transponder 1 <—— Transponder bandwidth 36 MH2————» 40 10 10 USA USA Chile T T T 3714 3728 3738 Carrier frequency in MHz ‘Transponder 2 Transponder bandwidth 36 MHz ———= 3759 77 Carrier frequency in MHz FIGURE 6.2 Frequency plan for two C-band transponders using FDMA. The triangles are symbols representing the bandwidth occupied by the signals, not power spectral densities ‘The places and figures within the triangles are the transmitting station location and carrier RF bandwidth, Frequencies shown are for the downlink from the satelite. communications began, and microwave filters were used in earth stations to separate the FDMA signals within a given transponder. In a fixed assignment system, each trans- mitting earth station was allocated a frequency and bandwidth for each group of sig- nals it wished to send. | Figure 6.2 shows a typical fixed assignment FDMA plan fortwo C-band transponders, ‘The triangles represent RF carriers with the transmitting earth station and RF bandwidth | shown inside the triangle, Frequencies shown are for the downlink from the satellite. Within | cach transmission, signals (primarily telephone channels) for different destinations are mul- i tiplexed using FDM. Typical Intelsat FDM carriers with a bandwidth of {0 MHz carried 132 to 252 telephone channels. If a small group of channels is intended for a given earth i station, the entire carrier must be received and demultiplexed to recover those channels (i Channels sent by the same carrier but intended for other earth stations are discarded. The 36 MHz transponder bandwidth can be used to send one or two television signals instead of hundreds of telephone channels. The use of microwave filters to separate channels made the fixed assignment ap- proach to FDMA very inflexible, Changing the frequency assignment or bandwidth of any ne transmitting earth station required retuning of the microwave filters at several receiving earth stations, The fixed assignment FDM-FM-FDMA scheme illustrated in Figure 6.2 also makes inefficient use of transponder bandwidth and satellite capacity, As an example, suppose an earth station in the west of the United States uses a Pa- cific Ocean GEO satellite to send telephone channels to earth stations in Korea, Japan, and Chile. The time difference between North America and the Pacific Rim countries ‘means that the channels will be busy for only a few hours per day, and at a different time of day than the U.S.-Chile links. With fixed assignment, the frequencies and satellite c2- i pacity cannot be reallocated between routes, so much of the satellite capacity remains idle. | Estimates of average loading of Intelsat satellites using fixed assignment are typically around 15%. It is not possible to achieve 100% loading of satellites used for internationalangles are | densities, 1nd carrier ral receiving, n Figure 6.2 2s uses a Pa ‘orea, Japan, im counties 62 FREQUENCY DIVISION MULTIPLE ACCESS (FMA) 225 = Transponder bandwidth 54 MHz (fo leaeh Me 8 40 kHz digital speech channels 10 kHz guard bands FIGURE 6.3 Illustration of a Ku-band transponder bandwidth filled with a large number of FDMA-SCPC digital speech channels. RF bandwidth of each channel is 40 kHz with 10 kHz ‘quard bands between channels. traffic, or even for domestic traffic in many cases. Demand assignment and single channel per carrier (SCPC) techniques allow higher loadings and therefore give satellite operators increased revenue. There has been a steady move away from fixed assignment systems as a result Every earth station that operates in an FDMA network must have a separate IF receiver for each of the carriers that it wishes to receive. SCPC systems can have a very large number of carriers in one transponder, as a result, FDMA earth stations tend to have a very large number of IF receivers and demultiplexers which select individ- ual carriers using narrowband IF filters. Figure 6.3 shows how the intermediate frequency bandwidth of a receiving earth station could be configured to receive 1000 digital speech channels, each with a bandwidth of 40 kHz from a 54 MHz wide Ku- band transponder. The 10-kHz frequency spaces between the channels are called guard band: bands are essential in DMA systems to allow the filters in the receivers to seleet individual channels without excessive interference from adjacent channels. ‘All filters have a roll-off characteristic, which describes how rapidly a filter can change from near zero attenuation in its pass band to high attenuation in the stop band. Typ- ically, guard bands of 10 to 25% of the channel bandwidth are needed to minimize adjacent channel interference. FDM-FM-FDMA was a telephone transmission technique well suited to analog telephone signals. Telephony has largely become digital, and frequency division multiplexing has been replaced by time division multiplexing. Digital speech is now used throughout telephone systems, so multiple telephone channels are always trans- mitted as a high-speed digital signal. The TI or DS-I carriers are examples of lower speed digital multiplexed carriers. Optical networks carry OC-48 digital signals at rates up to 2,7 Gbps and beyond. Appendix B discusses the techniques used in FDM-FM- FDMA systems, and the calculation of system capacity using this multiple access scheme. Apart from the analog nature of FDM-FM-FDMA which has rendered it ob- solete, it is a rather inflexible way to allocate satellite transponder capacity and is not easily adapted to demand access. FDMA is widely used as a method of sharing the bandwidth of satellite transpon- ders, Ideally, a satellite would carry a very large number of transponders, each of which could be allocated to a single RF carrier. In the case of telephony, each transponder would have a bandwidth exactly matched to the RF spectrum of the transmitted telephone chan- nel, with tight filtering to ensure that each signal can be separated from adjacent signals. ‘This approach is impractical: thousands of transponders would be needed and the satel- lite could be used only for telephony. The builders and operators of satellites have226 © cHAPTER6 MULTIPLE ACCESS historically shown a strong preference for wideband transponders that can carry any y, of traffic—the bent pipe transponder that can carry voice, video, or data as the may” place demands. As a result, transponders have always had wide bandwidths, with bang widths of 36, 54, and 72 MHz commonly employed. When an earth station has a carrie, that occupies less than the transponder bandwidth, FDMA can be used to allow that arn to share the transponder with other carriers. Allocating a wideband transponder to a single narrow bandwidth signal is clear wasteful, so FDMA is a widely used technique. When an earth station sends one signa, carrier, the FDMA access technique is called single channel per carrier (SCPC), (ee asystem in whjch a large number of small earth stations; suci as MObITe telephaiie, -cess a single transponder using FDMA is called a single channel per carrier frequency division multiple access ee surprisingly, this lengthy descriptor is abbreviate to SCPC-FDMA. Hybrid multiple access schemes can use time division multiplexing of baseband channels which are then modulated onto a single carrier. A number of earth sta tions can share a transponder using frequency division multiple access, giving a system known as TDM-SCPC-FDMA. Note that the sequence of abbreviations is baseband mul- tiplexing technique first, then multiple access technique next. TDM-SCPC-FDMA is used by VSAT networks in which the earth stations carry more than one baseband signal FDMA has a disadvantage in satellite communications systems when the satellite transponder has a nonlinear characteristic. Most satellite transponders use high-power am- plifiers which are driven close to saturation, causing nonlinear operation. A transponder using a traveling wave tube amplifier (TWTA) is more prone to nonlinearity than one with a solid state high-power amplifier (SSHPA). Equalization at the transmitting station, in the form of predistortion of the transmitted signal, can sometimes be employed to linearize the transponder when fixed assignment is used. Linearization of solid-state and TWT HPAs ‘on the satellite is also possible. Nonlinearity of the transponder HPA causes a reduction in the overall (C/N)o ratio at the receiving earth station when FDMA is used because in- termodulation (IM) products are generated in the transponder. Some of the IM products will be within the transponder bandwidth and will cause interference. The IM products are treated as though they were thermal noise, adding to the total noise in the receiver of the receiving earth station. intermodulation Intermodulation products are generated whenever more than one signal is carried by 3 nonlinear devic). Sometimes filtering can be used to remove the IM products, but if they are within the “bandwidth of the transponder they cannot be filtered out. The saturation characteristic of a transponder can be modeled by a cubic curve to illustrate the genera- tion of third-order intermodulation. Third-order IM is important because third-order 1M products often have frequencies close to the signals that generate the intermodulation, and are therefore likely to be within the transponder bandwidth. To illustrate the generation of third-order intermodulation products, we will model the nonlinear characteristic of the transponder HPA with a cubic voltage relationship and apply two unmodulated carriers at frequencies f, and f, at the input of the amplifier Vou = AVin + B(Vin)® (6.1) where A > b. The amplifier input signal is V, cosa + V coswatried by a ut if they saturation e genera: order IM tion, and ill model sship and ifier (6.1) 6.2 FREQUENCY DIVISION MULTIPLE ACCESS DMA) 227 ‘The amplifier output signal is Vou = AVig + (Vin)? = AV, cosa! + AV, cosunt + B(V, cost + V, coswyt)? (6.3) Fiver tem ‘entice ‘The linear term simply amplifies the input signal by a voltage gain A. The cubic term, which will be denoted as V;,., can be expanded as Vou = (V, coswt + V;coswt)? = B[V} cos'wt + Vj costae + 23 cos*wat X Vz coswat + 2V}cos*wnt X Vi cose!) (6.4) ‘The first two terms contain frequencies f;, f,, 3fi, and 3f2. The triple frequency compo- nents can be removed from the amplifier output with band-pass filters. The second two terms generate the third-order IM frequency components. We can expand the cosine squared terms using the trig identity cos? x = 4[cos2x + 1). Hence the IM terms of interest become Vy = BV} X Va[coswr X (cos2w,r + 1)] + BV} X Vi[coswt X (cos2wer + 1)] = BV? X V,[coswzt cos2wt + coswt} + BV} X Vi[cosw,t cos2wst + cosy] (6.5) ‘The terms at frequencies f, and f, add to the wanted output of the amplifier, so the third- order intermodulation products are generated by the f; X 2f, and f-X 2f; terms. Using another trig identity cosx cosy = cos(x + y) + cos(x — y) ‘The output of the amplifier contains IM frequency components given by Viw = BV? X Vafeos(2my + wzt) + cos(2ayt — wat)]} + BYE X Vi{cos (2a, + wyt) + cos(2wt — w1)) (6.6) We can filter out the sum terms in Eq, (6.6), but the difference terms, with frequencies 2, — fy and 2f, ~ f, may fall within the transponder bandwidth. These two terms are known as the third-order intermodulation products of the high-power amplifier, because they are the only ones likely to be present at the output of a transponder which incorpo- rates a narrow bandpass filter at its output, Thus the third-order intermodulation products that are of concern are given by Vay where V2V, cos(2at — at) + bVZV, cos(2wyt — ct) (6.7) "The magnitude of the IM products depends on the parameter b, which describes the nonlinearity of the transponder, and the magnitude of the signals. The wanted signals at the transponder output, at frequencies f, and f;, have magnitudes AV, and AV, The wanted output from the amplifier is Van = Voy = AV, coset + AV; coswzt The total powerof the wanted output from the HPA, referenced to a 1 ohm load, is therefore Pog = AA'VE + 2A7V3 = AMP, + P2)W (6.8) where P, and P, are the power levels of the wanted signals. The power of the IM products at the output of the HPA is Pra = PVE + EbVS) = BP} + PW (69)228 CHAPTER 6 MULTIPLE ACCESS It can be seen that IM products increase in proportion to the cubes of the signa with power levels that depend on the ratio (b/A)°, The greater the nontineayiyg amplifier (larger b/A ratio), the larger the IM products. Y OF intermodulation Example Consider the case of a 36-MHz bandwidth C-band transponder which has an outpy trum for downlink signals in the frequency range 3705-3741 MHz. The tranponq. ries two unmodulated carriers at 3718 and 3728 MHz with equal magnitudes at the. I the in to the HPA. Using Eq, (6.7), the output of the HPA will contain additional frequengs. ponents at frequencies fa = (2 X 3718 ~ 3728) = 3708 MHz. fig = (2 X 3728 — 3718) = 3738 MHz Both of the IM frequencies are within the transponder bandwidth and will there be present in an earth station receiver that is set to the frequency of this transponder. he magnitude of the IM products will depend on the ratio b/A, a measure of the nonlincaciy of the HPA, and on the actual level of the two signals in the transponder. Now consider the case where the two signals carry modulation which spreads ih signal energy into a bandwidth of 8 MHz around each carrier. Carrier | has frequencies 3714 to 3722 MHz and carrier 2 has frequencies 3726 to 3734 MHz. Denoting the band of frequencies occupied by the signals as fj. 10 fy, the intermodulation products cover the frequency bands (2fue ~ fini) © (2 ~ Saw) and (2 fale ~ fins) 10 (2.Fani ~ Fi The IM products are spread over bandwidths (2B, + By) and (2B, + By). Hence the third-order IM products for this example cover these frequencies 3706 ~ 3730 MHz and 3716 ~ 3740 MHz with bandwidths of 24 MHz. The location of the 8 MHz wide signals and 24 MHz wide IM products is illustrated in Figure 6.4. The intermodulation products now interfere with both signals, and also cover the empty frequency space in the transponder. Third-order IM products grow rapidly as the output of the transponder increases o ward saturation. Equation (6.9) shows that IM power increases as the cube of signal power in decibel units, every 10 dB increase in signal power causes a 30 dB increase in IM prod- uct power. Consequently, the easiest way to reduce IM problems is to reduce the level of the signals in the HPA. The output power of an operating transponder is related to its ‘Transponder bandwidth 36 MHz 3705 aris 3722 3726 3734 grat Frequency in MHz FIGURE 6.4 Intermodulation between two C-band carriers in a transponder with third-order nonlinearity.the band kts cover (62 FREQUENCY DIVISION MULTIPLE ACCESS (FOMA) 229, saturated output power by output backoff. Backoff is measured in decibel units, so a transponder with a $0 W rated (saturated) output power operating with an output power of 25 W has output backoff of [7 dBW ~ 14 dBW = 3 4B. Intermodulation products are reduced by 9 dB when 3 dB backoff is applied, so any nonlinear transponder carrying ‘more than one signal will usually have some backoff applied, Since a transponder is an amplifier, the output power level is controlled by the input power, and there is a saturated input power level corresponding to the saturated output level, When the transponder is op- erated with output backoff, the power level at its input is reduced by the input backoff. Because the transponder characteristics are not linear, input backoff is always larger than output backoff. Figure 6.5 illustrates the operating point and input and output backoff for 4 transponder with a nonlinear TWTA. The nonlinearity of the transponder causes the in- put and output backoff values to be unequal. In the example shown in Figure 6.5, the transponder saturates at an input power of —100 dBW. The transponder is operated at an input power of ~102.2 dBW, giving an input backoff of —2.2 dB. The corresponding out- put backoff is 1.0 4B, giving an output power of 16 dBW (40 W), 10 W below the saturated ‘output power of $0 W (17 dBW), Note that the TWTA has slightly different characteristics when operated witha single carrier and multiple carriers. The generation of intermodulation products when multiple carriers are present robs the wanted output of some of the transponder output power. For the nonlinearity shown in Figure 6.5, the reduction in output power is 0.6 dB at saturation. In the example above, both carriers had equal power. If the powers are unequal, the weaker signal may be swamped by intermod products from the stronger carrier. This can be seen from Eq, (6.9); the IM products that tend to affect Carrier 1 have voltages proportional to the square of the voltage of Carrier 2. Outback backott Single Multicarrier Operating ra aint Transponder | [n* backott power BW | Input | saturation 10 L = 105 100 Fn ‘Transponder input power dBW FIGURE 6.5 Typical input-ouput characteristic of a transponder using a traveling wave tube amplifier (TWTA). To maintain quasi-linear operation and minimize intermodulation Problems with multiple carriers, the transponder must be operated below its saturated ©utput power level. in this example, the saturated output power from the transponder is 50 W (17 dBW) with a single carrier. The saturated output power is slightly lower with ‘multiple carriers because some output power is converted to intermodulation products.230 CHAPTER 6 MULTIPLE ACCESS Operation of a nonlinear transponder with multiple carriers requires car ancing of the power levels of each carrier so that intermodulation products a spread across the transponder's bandwidth, Judicious spacing of the carriers can he to place the highest intermods in gaps between carriers. The provess is known as iy. the transponder. Sophisticated computer programs are used by satellite operators 1.” mize the backoff level of a transponder such that intermodulation is minimized output power is maximized, When a very large number of carriers access a transpuy using FDMA, as might happen with a network of VSAT stations or a transponder 4. with mobile satellite telephones, the transponder must operate in a quasi-linear region.) its characteristics. Quasi-linear means almost linear, either by equalization or by application of a large output backoff. Earth station HPAs can also cause intermodulation if they carry multiple cari, and operate close to saturation, In large earth stations where multiple carriers are mye likely to be transmitted, the HPA is often rated at a much higher level than the expecisy transmit power. This allows substantial backoff to be used, keeping the amplifier in iy linear region. In the above analysis of third-order intermodulation, only two carriers were con sidered. If there are three (or more) carriers present in a nonlinear transponder, intr modulation products at frequencies such as f + f; ~ f, can be generated that are likely to be within the transponder bandwidth. When many carriers are present, as with y transponder carrying narrowband SCPC signals, there will be a very large number of 1M products, making quasi-linear operation essential. et ig MO, om whi Calculation of C/N with Intermodulation Jntermodulation between carriers in a nonlinear transponder adds unwanted products into the transponder bandwidth that are treated as though the interference were_Gaussian noise, For wideband carriers, the behavior of the IM products will be noiselike; with narrowband carriers, ih€ assumption may not be accurate, but is applied because of the difficulty of determining the exact nature of the IM products. ‘The output backoff of a transponder reduces the output power level of all carriers, which therefore reduces the (C/N) ratio in the transponder. The transponder C/N ratio appears as (C/N)qp in the calculation of the overall (C/N)o ratio in the earth station receiver. IM noise in the transponder is defined by another C/N ratio, (C/N), which enters the ‘overall (C/N)o ratio through the reciprocal formula (using linear C/N power ratios) (C/N )o = W/LM(C/N yp + LAC/N)an + (C/N) i) (6.10) ‘Techniques for the calculation of (C/N) are beyond the scope of this text. Full knowl edge of the transponder nonlinearity and the signals carried by the transponders required to permit (C/N)pq to be calculated**, There is an optimum output backoff for any nonlinear transponder operating in FDMA mode. Figure 6.6 illustrates the effect of the HPA operating point on each C/N ratio in Eq, (6.10) when the operating point is set by the power transmitted by the earth station, The uplink (C/N)yp ratio increases linearly as the transponder input power is increased, leading to a corresponding nonlinear increase in transponder output p {As the nonlinear region of the transponder is reached, the downlink (C/N),p ratio in less rapidly than (C/N) because the nonlinear transponder is going into saturation Intermodulation products start to appear as the nonlinear region is approached, incrcas- ing rapidly as saturation is reached. With a third-order model for nonlinearity, the int modulation products increase in power at three times the rate at which the input power to(62 FREQUENCY DIVISION MULTIPLE ACCESS FDMa) 231 n FIGURE 6.6 C/N ratios for a link using the nonlinear transponder illus- (Nine trated in Figure 65. Overall (C/N, at the receiving earth station is the com- bination of the three C/N ratios shown (Me in this figure. As the power level at the input of the transponder is increased, : {C/Nigg in the transponder increases linearly, but (C/Nlay in the earth station 7 receiver increases less rapidly as the transponder saturates. Third-order in- termodulation products are generated in the transponder as it saturates, ‘causing overall (C/N), to peak at an com, input level of ~ 104 dBW. This is the optimum operating point for this soe transponder. The dashed lines show =110 ~105 100 Ph C/N ratios for a transponder that does ‘Transponder input power dBW not saturate. CIN ratio in dB (Cen, the transponder is increases, causing (C/N) to decrease rapidly as saturation is ap- proached. When all three C/N ratios are combined through Eq, (6.10), the overall (C/N)o ratio in the receiving earth station receiver has a maximum value at an input power level of ~ 104 dBW in the example in Figure 6.6. This is the optimum operating point for the transponder. The optimum operating point may be many decibels below the saturated output level of the transponder under some conditions* VSAT networks and mobile satellite telephones often use single channel per carrier (SCPC) FDMA to share transponder bandwidth. Because the carriers are narrowband, in the 10 to 128 kHz range typically, a 36 or 54 MHz transponder may carry many hundreds of carriers simultancously. The balance between the power levels of the carriers may not be maintained, especially in a system with mobile transmitters that can be subject to fad- ing. The transponder must operate in a linear mode for such systems to be feasible, either by the use of a linear transponder or by applying large output backoff to force operation of the transponder into its linear region. EXAMPLE 6.2.1 Power Sharing in FOMA ‘Three identical large earth stations with 500 W saturated output power transmitters access a 36 MH2_ bandwidth transponder using FDMA. The transponder saturated output power is 40 W and itis op- erated with 3 dB output backoff when FDMA is used. The gain of the transponder is 105 dB in its linear range. The bandwidths of the earth station signals are y the uplink : Station A: 15 MHz input powet Staion B: 10 MHz utput power Station: 5 MHz vi increases Find the power level at the output of the transponder, and at the input to the transponde ‘BW, for cach earth station signal, assuming that the transponder is operating in its linear region with 3 dB output backoff. Each earth station must transmit 250 W to achieve an output power of 20 W from the transponder. Find the transmit power for each earth station aed the transponder is ‘operated with FDMA by the three earth stations,232 HAPTER 6 MULTIPLE ACCESS ‘The output power of the transponder must be shared between the three signals in to their bandwidths. The output backoff of 3 dB means that the output power from the trans in ‘ is P, where : a P, = 10 logy 40 ~ 3 = 16 — 3 dBW 13. dBW = 20W. ‘The total bandwidth used is 15 + 10 + 5 = 30 MHz. The output power must be shared in Proportion to bandwidth used, so the transponder output power allocated to each earth station, signal is Station A: B= ISMHz P, = 15/30 X 20 W = 10.0W = 10 dBW StationB: B= JOMHz P, = 10/30 x 20 W = 6.67 W = 8.24BW StationC: B= SMHz P, = 15/30 x 20 W = 3.33 W = 52dBW ‘The transponder gain is 105 dB, in its linear range, so for linear operation the transponder input power for each earth station signal is Station A: Pj, = 10.0 ~ 105 = -95.0 BW Station B: Pi, = 8.2 ~ 105 = ~96.8 dBW Station: P, = 5.2 ~ 105 = -99.8 dBW ‘The EIRP at each earth station must be set to give the correct input power at the input to the ‘cansponder. A single earth station must transmit 250 W = 24 dBW to achieve a transponder out. pat of 20 W, For the transponder output power levels of each signal calculated above, the earth sta- tion transmitter powers are, Station A: 14.0 - 3.0 = 21.0 BW = 126 W Station B: 240 — 48 = 19.2 dBW = 83 W StationC: Py = 240 ~ 7.8 = 16.2 dBW = 42 W s EXAMPLE 6.2.2 Channel Capacity with Demand Access FOMA A large number of satellite telephones can access a single transponder on an LEO satellite using FDMA-DA. Data transmitted from the satellite on initial access by the telephone is used to set the transmit frequency and output power of the satellite telephone. The telephones transmit BPSK sig- nals in L band with an occupied bandwidth of 12 KHz and an output power level between 005 and 0.5 W, such that the power level at the input to the transponder is always —144 dBW for any up: link signal, The resulting C/N ratio in clear air conditions for any one signal in the transponder is 16 dB. ‘The transponder has a bandwidth of 1.0 MHz, a gain of 134 dB, and a maximum permitted ‘output power of 5 W. The center frequencies of the telephone transmitters are spaced 16 kHz apart to provide a 4 kHz guard band between each signal Determine the maximum number of satellite telephones which can simultaneously access the transponder. Is the transponder power or bandwidth limited” If the transponder is power limited, ‘what change could be made to increase the number of signals the transponder carries? What effect ‘would the change have on overall (C/N), for the link? If the transponder is bandwidth limited, the maximum number of signals, Nay, that it could carry is the available bandwidth divided by the signal bandwidth plus the guard band width Nox = 1000 kH2/16 kHz = 62 The value Of Nqgy Must be rounded down to the next lowest integer because we cannot send fractional signals. ; ‘The power level of each signal at the input to the transponder is ~ 144 dBW, The gain of the transponder is 134 dB, so the output power for each signal is P, = ~144 + 134 = ~104BW = Ww$9 TIME DIVISION MULTIPLE ACCESS (Toma) 233, If we have 62 signals, each at a power level of 0.1 W, the transponder is 62 W. This exceeds the maximum permitted output asset at 5.0 W. Hence the maximum numberof satelite telepho the transponder is 90, and the transponder is power limited, ‘We can increase the numberof signal inthe transponder o 62, which isthe maximum os: eee eecaieenones that can share the transponder atthe same time because ofthe bee. width init by eeducing the input power level by 10 log (62/50) = 09 4B. Then the cura power from the transponder, per signal, is {otal power at the output of the Power of the transponder, which mes that can simultaneously access P,= 1448 + 134 = ~10.9 dBW = 0.081 W OdBW ‘We can now transmit 62 signals from the transponder with a total ‘output power level of 62 x 2aBW 0.081 = 5.0W, which meets the power limitation for the transponder 20BW ‘The C/N ratio in the transponder will be reduced by 0.9 dB because the input signal is 0.9 4B weaker. Hence (C/N)ag = 16 ~ 0.9 = 15.1 dB. The transponder now transmits O09 a Jess ower Pet Slgnal which will reduce the (C/N), rato atthe receiving earth station by 0.9 gB: Hens the overall (C/N) ratio forthe link willbe reduced by 09 dB when the number of satelite tele- Phones sharing the transponder is increased from 50 10 62. TIME DIVISION MULTIPLE ACCESS (TDMA) transponder oy. a MA D satelite using with FDMAy However, using all of the transponder bi tion to transmit at a high Bit rate, which requires hig used to set the not well suited to_narrov nd_signals from small earth stations, , Nonlinearity ithe smit BPSK sig. trunspondet can cause an increas” in intersymbol interference with. digit carriers; tween 005 and caualizers can be used at the receiving earth stations to mitigate the effect, Many of the concepts developed in Chapter 5 for time division multiplex 1g (TDM) transponder is ako apply to TDMA. The difference between TDM and TDMA is that TDM jn pew band technique used at one location (for example, a transmitting earth station) to. multi- i apart | 4 ex sever dial bil streams int a single hi taken from each of the bit streams and formed Speed digital signal, Groups of bits are baseband packets or frames that also iy aca contain synchronization and identification bits, At a the high-speed ? Wine bit stream must first be recovered using the techniques discussed in Chapter 5, which re- Guires demodulation of the RF carver, generation ofa bit clock, sampling ofthe recenned tha it could [areform, and recovery ofthe bits. The synchronization bits or words in the packets og width ames must then be found so that the high-speed bitstream can be split into ae original lower speed signals. The clock frequency for the bit stream is fixed. andthe fae length is usually constant, Packet lengths can vary, however, whichis the main difference oor cannot send een frames and packets, The entire process requires considerable storage of bits so tha = the original signals can be rebuilt, leading to delays in transmission, In @ GEO serene € gain o Pea ne trBest delay is always the transmission time to the satelitéand back io earth, “typically 240 ms. The transmission delay is unavoidable, but any additional delays should be minimized234 CHAPTER 6 MULTIPLE AccESS Satelite TOMA stream from satelite Bursts from each ‘earth station Incoming bit streams Earth terminals FIGURE 6.7 Illustration of TDMA with three earth stations. Transmitting earth stations ‘must time their burst transmissions so that they arrive at the satellite in the correct se- quence. The signal transmitted by the satellite is a continuous sequence of bursts separated by short guard times. Mé le inique that allows a single transponder to be in time between RF carriers from different earth stations. Ina TDMA system, the ier from each earth station sharing a transponder is sent as a burst at a specific time. At the satellite, bursts from different earth stations arrive sequetiially, so the transpon- der carries a near continuous signal made up of a sequence Of short bursts coming from different earth stations. The principle of TDMA is illustrated in Figure 6.7. ‘The burst transmission is assembled at a transmitting earth station so that it will cor- recily fit into the TDMA frame a the satellite. The frame has a length from 125 jis to many milliseconds, and the burst from the earth station must be ‘transmitted at the correct time to arrive at the satellite in the correct position within the TDMA frame. This requires syn- chronization of all the earth stations in a TDMA network, adding considerable complexity to the equipment at the transmitting station. Each station must know exactly when to trans- mit, typically within a microsecond, so that the RF bursts arriving at the satellite from dif- ferent earth stations do not overlap. (A time overlap of two RF signals is called a collision and results in data in both signals being lost. Collisions must not be allowed to occur ina TDMA system.) A receiving earth station must synchronize its receiver to each of the sequential bursts in the TDMA signal and recover the transmission from each uplink earth station, The uplink transmissions are then broken down to extract the data bits, which are stored and reassembled into their original bit streams for onward transmission, The individual transmissions from different uplink earth stations are usually sent using BPSK or QPSK, and will inevitably have small differences in carrier and clock frequencies, and different carrier phases. The receiving earth station must synchronize its PSK demodulator to each burst of signal within a few microseconds, and then synchronize its bit clock in the next few microseconds so that a bit stream can be recovered. In high-speed TDMA systems, operating at 120 Mbps, for example, these are demanding requirements. Bits, Symbols, and Channels A potential source of confusion in the discussion of TDMA systems is that QPSK (or pos- sibly QAM) modulation is typically used by transmitting earth stations, and data rates can6.3 TIME DIVISION MULTIPLE ACCESS (TOMA) 235 then be described either by bit rate or symbol rate. Both bit rates and symbol rates need to be used in the discussion of digital radio transmission and TDMA systems, 60 the reader ‘must be clear on the distinction between a bit and a symbol, A bit is the fundamental unit in digital transmission. Data are generated by termi- nals (¢-8. @ personal computer) as bits, or by conversion of an analog speech or video signal to digital form as a serial bit stream. The bit stream is described by its bit rate, in bits per second, bps, thousands of bits per second, kbps, ot millions of bits per second, Mbps. (Note that the k and M prefixes are in units of 10° and 10%, not the binary digital version of 1024 and 1,048,576.) The bit stream must be modulated onto an RF carrier for transmission to the satellite. Phase shift keying is invariably used as the modulation tech- nique. In binary phase shift keying, BPSK, the logical data states of the bits, I and 0, are converted into two opposite phase states of the RF carrier, say O° and 180°, In quadrar- ure phase shift keying, QPSK, two bits at atime are converted into one of four phase states of the RF carrier (see Chapter 5 for details of QPSK), The state of the RF carrier is called a symbol, and the symbol rate is in units of bauds, or symbols per second. For BPSK, bit rate and baud rate are the same. For QPSK, the baud rate (symbol rate) is one half the bit rate. The importance of symbol rate in any digital radio system is that it is the symbol rate, not the bit rate, that determines the bandwidth of the RF signal, and consequently the bandwidth of the filters in the receiver. Occasionally, QAM modulation may be used on a satelite link. In QAM, cartier symbols are generated from the four phase states of QPSK, but can also have different amplitudes, This allows one symbol to convey more than 2 bits, which reduces the RF bandwidth required for a given bit rate. However, an increased (C/N)q ratio is required in the receiver to recover bits from QAM signals, so QAM can only be used in satellite links with higher than usual (C/N)q ratios. TDMA Frame Structure ATDMA frame contains the signals transmitted by all of the earth station in a TDMA network. It has a fixed length, and is built up from the burst transmissions of each earth station, with guard times between each burst. The frame exists only inthe satellite transpon- er and on the downlinks from the satellite to the receiving earth stations. Figure 68 shows a simplified diagram of a TDMA frame for four transmitting earth stations. Each station transmits a preamble that contains synchronization and other data essential to the opera- tion of the network before sending data. The earth station’s transmission is followed by a guard time to avoid possible overlap of the following transmission. In GEO satellite sys- ‘tems, frame lengths of 125 j1s up to 20 ms have been used, although 2 ms has been widely used by stations using Intelsat satellites. Earth stations must be able to join the network, add their bursts to the TDMA frame in the correct time sequence, and leave the network FIGURE 6.8 TDMA frame with four transmit. Trafic: Nits Preamble ting earth stations.CHAPTER 6 MULTIPLE ACCESS without disrupting its operation. They must also be able to track changes in the timing g, the frame caused by motion of the satellite toward or away from the earth station, Bach earth station must also be able to extract the data bits and other information from burg, transmissions of other earth stations in the TDMA network. The transmitted bursts mus contain synchronization and identification information that help receiving earth stations to extract the required information without error. ‘These goals are achieved by dividing TDMA transmissions into two parts: a pre, amble containing all the synchronization and identification data, and a group of trafic bis, ‘Synchronization of the TDMA network is achieved with the portion of the preamble trans. mitted by each earth station that contains carrier and bit clock synchronization waveforms, In some systems, a separate reference burst may be transmitted by one of the stations, d. station, A reference burst is a preamble followed by no traffic -affic bits are the revenue producing portion of each frame, and the preamble and reference bursts represent overhead. The smaller the overhead, the more efficient the ‘TDMA system, but the gréater the difficulty of acquiring and maintaining network syn. chronization. The preamble of each station’s burst transmission requires a fixed trans. mission timd, A longer frame contains proportional dst fremibieYia)than a short frame, so mofé fevenue producing data bits can be carrie in'a long frame. Early TOMX systems were designed around 125,1s frames, to match the sample rate of digital speech in telephone systems, in exactly the same way that T] 24-channel systems operate. A dig. ital telephone channel generates one 8-bit digital word every 125 xs (8 kHz sampling rate), so a 125-1s frame transmits one word from each speech channel. However, it is more ef- ficient to lengthen the frame to 2 ms or longer so that the proportion of overhead to mes- sage transmission time is reduced, It must be remembered that a longer frame requires multiple &-bit words when transmitting digital speech. For example, in a time period of 2 ms, a digital terrestrial channel will deliver sixteen 8-bit words, so a 2-ms TDMA frame requires sixteen 8-bit words from each terrestrial channel in each transmitted burst. Figure 69 shows a typical TDMA frame with 2.0 ms duration used by some earth stations operating in TDMA through Intelsat satellites. All of the blocks at the start of the cata | uw [rey] se VOW | VOW} Digital speech channels CEPLEEEL aa +]2|8| 4] s]6|7|8| 9] 10] 11}12]19] 4115} Sixteen 8-bit samples form each satelite channel FIGURE 6.9. Structure of an Intelsat traffic data burst. A satellite channel is a block of six- teen 8-bit samples from one terrestrial speech channel. Other blocks in the traffic burst are used to synchronize the PSK demodulator, the bit clock, and the frame clock in the receiver (CBTR, UW) and to provide communication links between earth stations (TTY, SC, and VOW). CBTR, carrier and bit timing recovery; UW, unique word; TTY, teletype; SC, satellite channel: VOW, voice order wire. —__46 TIME DIVISION MULTIPLE ACCESS (TOMAl 237 frame, labeled CBTR through VOW, are preamble. Speech channe! data transmission be- ins with satelite channel | and continues as a serial bit stream through channel M. A Jatellite channel is made up of the number of bits delivered to the earth station by a sin- fle digital specch channel during one frame period. The frame can equally well send dig- Hal data of any form as a serial stream of bits occupying the space taken up by M satellite I channels. For the specific case of digital speech channels using serial transmission at a rate ri the number of speech channels, m, that can be transmitted in a TDMA frame shared Gaially by N earth stations can be calculated from the duration of the frame. Trane in sec- nds, the guard time and preamble length, f, and fq, in seconds, and the transmitted bit cee of the TDMA system, Ry The ti, 7, available in each station burs for transmission rate oF system, Ry, The time, 7,, available in ench station x transmission _ of data ' Ts = (Tiame ~ Mlty + fne)I/N seconds (6.11) In 1 5, the total number of bits, Cy, transmitted by each earth station is Cy = [Trane ~ Nlty + bye)] X R/T (6.12) Since each digital speech channel requires a continuous bit rate of rg bps, the number of | speech channels that can be carried by each earth station is given by » where | Ry Tae Xe (6.13) : = [Thame ~ Mtg + toe)] EXAMPLE 6.3.1 TOMA in a Fixed Station Network A TDMA network of iv earth stations shares a single transponder equally. The frame duration is 0 ms, the preamble time per station is 20.us, and guard bands of 5 js are used between bursts Transmission bursts are QPSK at 30 Mbaud. ‘Calculate the number of 64 kbps voice channels that cach TDMA earth station can . transmit. If the earth stations send data rather than digital speech, what is the transmission rate ‘of each carth station in Mbps? What is the efficiency of the TDMA system expressed as Eitciency = 100% X Message bits sent/Maximum possible number of bits that could be sent? ‘Using Eq. (6.11) we can find the data burst duration for each earth station, Ta, in microseconds T, = (2000 ~ 5 x (5 + 20)/5 = 375 us ‘A burst transmission rate of 30 Mbaud is 30 million symbols per second, and QPSK symbols carry 2 bits, Hence the transmitted bit rat i €ach burst is Ry = 2 30 Mbps = 60 Mbps. “The capacity of each earth station in bits per second is C where | C= 375 X 60 X 10/2000 = 11.25 Mbps bursts each second, the number of message bits per burst is Since these bits are sent as 11.25 x 10/500 = 225,000 kbit ‘The number of 64 kbps digital speech channels that can be carried by one earth station is n= 11,250,000/64,000 = 175(.781) We have to discard the fractional channel since we can send only whole channels. The 0.781 fraction of a speech channel represents 50,000 bits per second that eannot be transmitted by each earth station, or 100 bits per burst. Ata bitrate of 60 Mbps, 100 bits is equivalent 0 167 pis, 30 the guard time per burst would increase from 5 to 6.67 js when 175 speech chan nels are sent. a238 CHAPTER 6 MULTIPLE ACCESS 1 is always a good idea to check the answer 10 any problem against re ‘ Value. Ifthe earth stations transmitted without any guard times or preambles, the Matin number of 64-kbps speech channels that could be sent by each of the five earth tions using a transmission burst rate of 60 Mbps would be n = Ry/(N Xx Y= 6), : 10° x 64 X 10") = 187.5. The calculated capacity per station of 175 channels ig : little lower than the maximum number of 187 channels; the difference-—12 channey this case—is the number of speech channels lost by the need to include guara times ig preambles in the TDMA frame structure. If we send data rather than digital speech, each earth station can transmi Mbps. Overhead accounts for the equivalent of 0.75 Mbps, leading to an efficien it Nas ey of Efficiency = 100% x 1125/120 = 93.75% Example 6.3.1 shows that there are diminishing returns to be obtained by ince, ing the duration of the frame beyond 2.0 ms in this system. There is a 6.25% loss of in tential data transmission time when the 2.0-ms frame length of Example 6.3.1 is used vip the other parameters of this TDMA system. Doubling the frame length to 4.0 ms wo Feduce the loss t0 about 38, but would add additional delay, and inerease the compl iy of the earth station equipment. In this case, a 2.0 ms frame period appears tobe agen) compromise between delay, complexity, and efficiency. Reference Burst and Preamble Figure 6.9 shows the baseband content of a typical TDMA burst. The segments marked CBTR and UW contain the carrier recovery waveform, bit clock synchronization, unique word, and station identifier. The remaining blocks are part of the preambs and will be described below. CBTR is the portion of the burst from a given earth sy, tion that enables a receiving earth station to recover the remainder of the burst. CBTR Stands for carrier and bit timing recovery. Cartier recovery is required at the receiver of any radio link in which coherent phase shift keying is the modulation technique. A local carrier must be generated in the IF portion of the receiver from the received BPSK or QPSK signal, This is typically achieved with some form of phase locked loop } (PLL), which must lock up quickly in a TDMA system. The local cartier drives the rnultipliers in the demodulator to achieve coherent demodulation of the PSK signal i (See Chapter 5 for details of the modulation and demodulation processin PSK systems) Once carrier phase lock is achieved, the demodulator will produce a baseband wave- form corresponding tothe bits that were modulated onto the carrier atthe transmiting i earth station. | In a TDMA system, the bursts of RF signal received sequentially from different | carth stations do not have the same carrier frequency, phase, or bit rate. The differences will be small, but sufficient to require the receiver to relock to each new carrier, and to ' resynchronize the bit clock. The CBTR contains a sequence of predetermined signals tha ensure rapid lock to the carrier and fast synchronization of the bit clock. The carrier re- covery portion of the CBTR sequence may consist of unmodulated cartier followed by & ‘number of symbols that follow a specific pattern, or the entire CBTR burst may be mod- ulated. The first part of the CBTR burst is used to obtain lockup of the PLL, and the r ‘maining portion is used for bit clock synchronization. In Figure 6.9, the CBTR burst has 3 duration of 176 symbols with a transmission rate of 30 Msps, giving a burst duration of i 5.86 uss. Within this very short time period the carrier recovery circuit must achieve pre: cise lock on the received signal and the bit clock must be brought into synchronization Carrier phase lock and bit synchronization must be achieved within the CBTR burst timeStreams is inverted. 4 kno ity resolution. The pattern allow the receiver to chy (, Q, or both) if ambi the correct UW sequen: correct time within Signal bury it the correct position (i.e, at the Uigger the threshold detector, ist) is receives carrier PLL, and the bit coc SSM Oso MuTPLE Access fOMA) 238 gen when the cation pj employing half rate forwarg error eye i, eth Typical TDMA systems im burst as all ones or all zeroes op ff QPSK, cd te is “ or 50 symbols ofthe core i i Of the earrer, which corres ments The core nt a by S422 Semece of nes and vere ints chan- which serves several purposes, acts ap tie Word (UW) of typically 20 to 48 bits or burst marker, and ier vc Atransmit gine frame (SOF) 38-8 cartier phase ambien po identifier, a stat of frame PS &s low as 5 dB in VSAT links ‘ove the BER of the received data, Unique Word The received bit sequence at relator which looks for the 4 illustrates a simplified unique ‘ween the incomin f correct UW sequence iants in which ed SeeNeNCES, corresponding to the inverted due to phase ambiguity in the ich the 7 result in ambiguity if the Possible in many QPSK dey demodhlae Cutbat is continuously run through a cor- | 7 ‘unique word in the bit stream. Figure 6.10 18 it pattern ang etor. The Correlator is looking for a match be- and three ya, four trey ‘i and @ bits of the sequence are = cuit. QPSK cartier recovery - | horn kup inthe incorest phase, which m vm ae ben tis Happens, one or both of the / and Q bit : OF ones and zergee feared in the received signal for ambigu- | eck for phage T° i the CBTR sequence and the unique word | iuity is foung) Ty mbistty ang invert the appropriate bit stream Ce is resent in ype elt diagram show in Figure 6.10, when the incomin, carrier TE Tecovery ci | correlator output will maximize and | I thon Ss he cariee Fecovery circuit locks up the local TOnizes to the bj ‘hen a new RF burst | it rate of the received signal. Output to threshold creuit : » Modula 2 multipters (exclusive OR, gates) ver cata Sele sre wae 01° chown hors has 46m ‘unique wors oa ca en en ‘ i the thee Od the ouant® serially. When the contents shold, marking the ena cr 1 unigan Of the summer is a maximum al inder of the earth station's transmisgign WU® Word. This Provides a time marker240° cuaPrens muttwte access Bits then begin to flow into the correlator, which detects one of the four POSsible fo, of the UW and sets logical inverters that invert the appropriate bite it necessary, Fesulting bit stream after the end of the UW is then output correctly and can be detected UW is known as a miss, and the Probability that a miss occurs can be caley Jated from the bit error probability (BER) of the recovered bit stream and the lengy in ¢rror while still ensuring that correct timing of the bit Sequence is achieved. The prob. ability that the specific bit sequence of a UW appears within the bit stream of traffic data must also be low, so that the probability of a false alarm is small. A false alarm Occurs when a UW is inadvertently detected in the traffie data, causing the timing of the burst to be reset. There are several ways that false alarms can be prevented, but the AS an example of the probability of a missed detection, a 24 bit UW with a bit error rate of 10° can be recovered with a miss Probability of 10-"° when three of the bits in the UW are allowed to be in error, corresponding to a detection threshold of 21/24”, 75 Hs Pet station. At a bit rate of 60 Mbps there are 225500 inire bits in each burst With a probability of missing the UW of 10°" the average BER of the traffic data re- sulting from one missed UW is 225 x 10% This BER is too high to be acceptable in re pencations, so a longer UW would have to be used to lower the probability of a the cagieteetion. For further details of unique word detection and TD¥e burst design, the reader is referred to one of the references™"". ‘There must be as many UW correlators as there are TDMA uplink stations in the meimork when the UW is also used as a station identifier. If there ic large number of in the earth station can route the data accordingly, ‘The remaining segments of the preamble provide the receiving carth station with bit eins that are used in the management of the TDMA system, Referring to Figure 6.9, there are groups of bits identtied as TTY, SC, VOW, Sinien bite (eight QPSK symbols) 'n each burst are allocated to a teletype link (TTY) between the earth stations, and aha bit nree of | eshold ight, a system | ime of burst ata re- able in y of a lesign, | ber of | he bit lexers ith bit | © 6.9, 1bols) and a 6.3 TIME DIVISION MULTIPLE access (Toma) 241 further 16 bits to a service channel (SC). There are two 64-bit voice order wire (VOW) segments that are used in digital voice links between the earth stations. The TDMA burst shown in Figure 6.9 is designed for use in a network of large earth stations in which there are personnel at the earth stations or network control centers. The teletype and voice channels within the preamble of each burst provide a closed communication network be- tween the control stations that is used to manage the TDMA system. In a VSAT network using TDMA, for example, there would be minimal communication requirement between terminals for management purposes, so a VSAT TDMA burst would omit most of these segments. The final part of the TDMA burst carries the traffic data, In Example 6.3.1 which used a 2.0 ms frame, speech data could be carried in satellite channels as sixteen 8-bit words from each terrestrial channel, giving a satellite channe! 128 bits in each burst. ‘The bits associated with each satellite channel are demultiplexed by counting the 128 bits of each satellite channel, beginning at the end of the preamble. Once again, the bit timing established by the unique word becomes all important in extracting the satellite channels correctly. There are many different formats for preambles in TDMA systems, depending on the design of the particular system. Figure 6.9 illustrates a TDMA preamble structure designed for a large fixed network with high-speed bit streams. A network of mobile earth terminals using TDMA would have different requirements and would require a different preamble structure. However, the earlier segments that control carrier and bit timing recovery, phase ambiguity removal, and station identification must always be present. A large earth station carrying high-speed data must link into a terrestrial data net- work to deliver received bits to customers, and to receive incoming data for transmission over the satellite link. The satellite link connects two earth stations which may be thou- sands of kilometers apart, each interconnecting to its own high-speed terrestrial network, ‘The individual terrestrial networks are not synchronized, and will therefore inevitably run at slightly different rates. This makes the interconnection process difficult, because a data stream delivering bits at 1.00001 Mbps cannot be connected directly to another data stream running at 0.99999 Mbps. Twenty bits would have to be discarded every second if a direct con- nection were made, which upsets the customer who is paying (o transfer the data. A mechanism must be developed that allows for a difference in bit rates at the two ends of the link. The usual solution is to run the bit clock of the satellite link slightly faster than the fastest of the terrestrial link clocks, and to allow additional bit slots for stuffing bits. Stuffing bits are inserted when there are no data bits available from the source because of the difference in bit rates. At the receiving end of the link, the stuffing bits are removed and the received data stream is retimed to match the outgoing terrestrial data channels. Guard Times Guard times must be provided between bursts from each earth station so that collisions are avoided. Earth stations must transmit their bursts at precisely the correct instant so that the burst arrives at the satellite in the correct position within the TDMA frame. This requires burst transmission timing to microsecond accuracy and tracking of the position of the burst within the TDMA frame by the transmitting earth station, Long guard times make it easier for the earth stations to avoid collisions, but waste time that could be used to send revenue-earning traffic data, Typical guard times in high-speed satellite networks242 CHAPTER 6 MULTIPLE ACCESS appear to be in the range | to 5 ws. The transmission time between an earth station ang a GEO satellite is about 120 ms, If a 2.ms frame time is used, there are typically 6 bursts between the earth station and the satellite at any time. The bursts must arrive at the correct time to mesh between the bursts that arrive from other earth stations, If the satellite range from the earth station increases by 300 m, due to eccentricity or incting. tion of a GEO satellite's orbit, o B~W drift in the orbital plane, the transmission dela, increases by 1 js. Thus earth stations must monitor the guard times before and after their bursts to ensure that transmission timing is correct. In LEO satellite networks that use ‘TDMA, range to the satellite is changing continuously and much larger guard times are allowed, Synchronization in TDMA Networks Earth stations operating in a TDMA network must transmit their RF bursts at precisely controlled times such that bursts from each of the earth stations arrive at the satelite in the correct sequence’. This poses two problems: how to start up a new earth station that is joining the TDMA network, and how to maintain the correct burst timing. If the satel- lite is in low earth orbit, or if it is a GEO satellite with a rapidly changing range, each earth station will perceive a different carrier frequency and frame rate, and even a different frame length, It is usual for the bit rate of transmitted bursts to be an integer multiple of the frame rate, which means that different earth stations must transmit at slightly different bit rates. ‘Maintaining synchronization with the TDMA frame is easier than initial synchro- nization when an earth station joins a TDMA network. One station is typically desig- nated as the master station, and may generate a reference burst to mark the start of the frame, Each of the stations within the network has a time slot within the frame, and must maintain its transmissions within that time slot. There are guard times at each end of cach station’s burst, which define the accuracy that the burst timing must achieve. If the guard times are 2 ys, each earth station in the network must keep its bursts timed to within 1 ps. This is usually done by monitoring the TDMA frame at the transmitting station and adjusting the burst timing to keep the transmitted burst in the correct time slot in the frame. ‘The start of the reference burst, or the start of the master station’s preamble, marks the Start of transmit frame, SOTE, which is the master timing mark for all transmissions. All earth stations in the TDMA network synchronize their clock timing with the SOTF marker. When an earth station monitors its own transmissions to maintain the correct burst tim- ing, this is called satellite loop-back synchronization. The TDMA frame is established at the satellite, so an earth station receiving the frame must subtract the transmission delay from the satellite to the earth station to obtain the SOTF timing at the satellite. It must then transmit its bursts ahead of the SOTF by the same delay time so that the bursts arrive at the correct instant at the satellite. Knowledge of the range of the satellite from the earth station is crucial in calculating delay times. The range can be calculated from the orbital elements of the satellite, which can be determined by a control station that repeatedly measures range to the satellite. ‘There are several ways that earth stations can enter a TDMA network. In fixed networks, the precise time at which an earth station should transmit can be calculated. Provided the calculation is accurate, the earth station can transmit a reference burst (no traffic) timed to fall in the center of its time slot. When the frames containing the refer- ence burst arrive back at the earth station, the actual position of the burst can be checkediii, 63 TIME DIVISION MULTIPLE ACCESS (TDMA) 243 and corrections to the timing made if necessary. The station can then transmit traffic bursts with the correct timing, : In TDMA networks which lack sophisticated timing control, an earth station wish- ing to join the network can transmit a CDMA sequence at a low level, at an arbitrary time. ‘The CDMA sequence will inevitably collide with another earth station’s traffic burst, causing minor interference. The transmitting earth station can use a correlator to com- press the CDMA sequence into a single timing pulse using exactly the same process as is used to find a unique word. However, in this case the CDMA sequence is overwritten by interference from the traffic burst with which it collided. Given a suitably long sequence, coding gain can overcome the interference and a pulse will appear at the correlator output marking the end of the CDMA sequence. Alternatively, the transmitting station can use a shorter sequence and step the timing of the CDMA burst until it falls, in the empty slot allocated to that station, The position of the pulse within the TDMA frame gives the transmitting earth station the timing information needed to transmit its, bursts at the correct time. If the signal transmitted by the satellite cannot be monitored by the transmitting earth station, cooperative synchronization must be used instead. This situation arises when a satellite has multiple beams, or when satellite switched TDMA is used. A TDMA burst received in one beam can be retransmitted by the satellite in another beam that does not cover the transmitting station. A control station is required that can monitor the timing of each of the earth station's bursts as they arrive at the satellite and send out instructions to the earth stations when changes in timing are needed. In the Intelsat TDMA system, the control station determines a delay time, Dy, for each earth station that gives the time be- tween the start of a receive frame and the start of a transmit frame at that earth station. ‘The correct transmit time is then determined by the position of earth station’s burst within the transmit frame. If transmitting earth stations fall out of sync, the control station must send a do not transmit (DNTX) code to the station to tell it to stop transmitting because serious loss of data will occur to other users of the network when a station sends its bursts at the incorrect time. In satellite switched and multiple beam satellite systems, the cooperating control station must provide information to a new earth station that wishes to join the network. ‘The same techniques described above can be used, but an earth station within the receiv- ing beam must determine the timing of test transmissions and send that information to the transmitting station. ‘The availability of a global GPS time standard with better than Lis accuracy has made some of these tasks easier. (See Chapter 12 for details of the GPS system.) Transmitter Power in TDMA Networks TDMA works well in fixed networks carrying high-speed data streams. Transponders can be more heavily loaded because less backoff is needed with TDMA; only one RF signal is present in the transponder at any time and there is no third-order intermodulation, 80 backoff is needed only to keep PM—AM conversion at an acceptable level. Burst lengths can be made variable to accommodate stations that have different bit rates. High uplink transmitter power is required because every station must transmit bursts at a high bit rate and high bit rate signals occupy a wide bandwidth. Maintaining an adequate C/N ratio in the transponder forces the uplink earth station to use a high-power transmitter. For small earth stations such as VSATs and satellite phones, this is a major disadvantage compared to SCPC-FDMA.244 © cHapTene MULTIPLE Access EXAMPLE 6.3.2 TDMA in a VSAT Network ‘Asan example, consider a typical VSAT earth station in the United States which is part of a TD) network using a 54 MHz bandwidth transponder on a domestic Ku-band GEO satellite, The VSat ‘arth station has @ 1 m antenna that transmits a single 64 kbps signal at 14 GHz. Let's assume the TDMA network uses QPSK modulation and that al transmitters have a symbol rate of 30 Mbaug We wil Set (C/N) #20 4B, and then calculate the quired uplink transmit power. The Fllongg system parameters will be used: Earth station antenna gain = 41.5 4B, salelite antenra gain (on axis) = 32.0 dB, edge of beam loss = 3 4B, path oss at 14 GHz = 207.14B, receiver noise bandwidth = 30 MHz, transpenin noise temperature = 500 K, atmospheric and other losses = 1.0 dB, ‘The uplink power and noise budgets are Earth station transmit power = P.dBw Earth station antenna gain at 11 GHz = 41.5 dB Satellite antenna = 32.008 Edge of beam loss = 3.0 4B Other losses 1.0 6B Path loss at 1} GHz = -207.1 dB Power at transponder input ~ 137.6 dBW Boltzmann's constant = ~228.6 dBW/K/Hz ‘Transponder noise bandwidth 748 dBHz ‘Transponder noise temperature = 27.0 dBK ‘Transponder input noise power = ~126.8 dBW We requite (C/N)yp = P,/RT.B, = 20€B; hence P, ~ 137.6 + 126.8 = 20 dBW and P, = 308 BW or P, = 1200 W. Now consider the same earth station transmitting the same 64-kbps signal in a SCPC-FDMA VSAT network using QPSK with a symbol rate of 32 kbaud and a receiver noise bandwidth of 32 kHz. The uplink power budget is unchanged, but the noise power in the transponder, measured in a bandwidth of 32 kHz is - 156.5 dBW. To achieve (C/N)yp = 20 dB in the transponder now requires an uplink transmitter power of P= 20 + 137.6 ~156.5 = 1.1 dBW = 13 W. a ‘The above example illustrates a key problem with TDMA for any small earth sla- tion: uplink transmit power. No one is going to equip a I-m VSAT station with a 1200-W transmitter. Apart from the excessive cost, FCC regulation in the United States do not allow small VSAT stations to transmit more than 2 W to limit interference to adjacent satellites. If we change the multiple access technique for just two earth stations, so that each transmits a burst of QPSK signal at 64 kbaud for half the time, the uplink transmitter Power requirement is doubled to 4.1 dBW or 2.6 W. This makes wideband TDMA an un- likely choice in VSAT networks, and limits the number of stations that can share a TDMA frame in a low earth orbit satellite telephone system. The Iridium LEO system was de- signed to use a hybrid TDMA-FDMA multiple access scheme at L band to combine 4 small number of digital telephone transmissions into a 50-kbps QPSK signal. Similar tech- niques are used in some VSAT networks. EXAMPLE 6.3.3 TDMA in a Fixed Earth Station Network zn Example 6.2.1, three identical large earth stations shared a single 36-MHz bandwidth transpon- er using FDMA. The three earth stations transmitted signals with powers and bandwidths. given i . _h stax 00-W allow lites. teach mitter 1m un 6.3 TIME DIVISION MULTPLE ACCESS (TDMA) 245 by Station A: B= 1S MHz P, = 125 W = 21.04BW Station B: B= 10MHz 83 W = 19.2eBW StationC: B= SMHz 42.W = 16.2 dBW ‘The transponder total power output was 16 dBW with 3-dB output backoff and 105-dB transponder gain, ‘The three earth station accesses to the transponder are changed to TDMA, with a frame length of 1.0 ms, 2 preamble time of 10 y1s, and a guard time of 2 41s. There is no reference burst in the ‘TDMA frame. The signals are transmitted using QPSK, and within the earth stations the bit rates of the signals are Station A: Ry = 15.0 Mbps Station B: Ry = 10.0 Mbps StationC: Ry = 5.0 Mbps Calculate the burst duration and symbol rate for each carth station, and the earth station trans- mitter output power required if the transponder output backoff i8 set at 1.0 dB and the gain of the transponder with this output backoff is 104 dB, Compare the uplink (C/N) ratios in the transponder for FDMA and TDMA operation given that station A’s transmission has a (C/N)yp ratio of 34 dB ‘when the transponder is operated in FDMA. ‘The transponder must carry a total bit rate of 15 + 10 + 5 = 30 Mbps within the 1.0-ms frames. Thus each frame carries 30 Mbps X 0.001 s = 30 kb. The three preamble and guard times take up 3 X (10 + 2) = 36 ys in each frame, leaving 1000 ~ 36 = 964 ys for transmission of data, Hence the burst bit rate is Rerun = 3OKD/964 ws = 31.12Mbps. 7 ‘Since we are using QPSK for the transmissions, the burst symbol rate on the link is Reruns = 31.12 Mbps/2 = 15.56 Mops 7 Each of the stations must transmit at the same burst rate of 15.56 Msps,,The burst lengths can be calculated from the time available in each frame for data transmission and the number of, bits each station must send in a } ms TDMA frame. The time available for data transmission is 964 145, which must be shared in proportion to the number of bits each station sends in a frame, The number of bits in a frame is 30,000, so the sharing of bits and times within a frame is given below Station A: N= 15,000bits 7, = 482.0 us Station: —R, = 10,000bits Ty = 321.3 us Station: Ry = 5,000bits Te = 160.7 ws We can easily check to see if these results are correct. Each earth station must have the stated average bitrate, so if we multiply the burst duration for each earth station by the burst bit rate for the transponder, 31.12 Mbps, we must have the correct number of bits/frame for each station. Station A:T, = 482.0 us N, = 482.0 ps X 31.12 Mops = 15,000 Station B: Ty = 321.348 Ny = 321.3 as X 31.12 Mbps = 10,000 Station: Te = 160.7 zs y= 160.7 us X 31.12 Mbps = 5,000 Each station must transmit at the same symbol rate of 15.56 Msps, regardless of the number of bits sent per frame. In the previous FDMA example, a transponder output power of 20 W = 13 dBW ‘was achieved with a total earth station power of 250 W = 24 dBW and a transponder gain of 105 dB. With TDMA, we are using a 1 dB transponder output backoff, and a transponder gain of 104 dB, so the transponder output power is now 16 — 1 = 15 dBW, an increase of 2 dB, and we have lost | dB of gain in the transponder. This requires an earth station output power, from each earth station, of P= M4241 =27W = 500W i,CHAPTER 6 MULTIPLE ACCESS TDMA requires a substantial increase in earth station transmitter power, relative to FD, for a low capacity earth station which joins high-capacity stations in a TDMA system. Jn this ex, ‘ample, a station that was transmitting 42 W when the transponder was operated in FDMA must noy, transmit 500 W when the transponder is operated in FDMA, ‘The uplink (C/N)uy for station A’s 15 MHz signal was 34.0 dB when the transponder yas operated in FDMA. With QPSK and a burst rate of 15.56 Msps, the noise bandwidth of the ear station receiver, assuming ideal RRC filters, will be 15.56 MHz. The output of station A has been increased from 21 to 24 dBW, s0 the input power level at the transponder will also have increaseg by 3 dB, Hence the uplink (C/N),y ratio in the transponder for station A signals is (C/N) = 34.0 + 10 logio(15.0/15.56) + 3 = 36.8 dB Since all earth stations transmit at the same power level and with the same burst rate, and all the sig. nals have the same noise bandwidth, the (C/N).» ratio for each of the signals in the transponder is identical, at 368 dB. This is 28 4B higher than for the FDMA operation, but at the expense of a large increase in the total uplink power from the three earth station transmitters. . Satellite Switched TOMA ‘One advantage that TDMA has when used with a baseband processing transponder is satel. lite switched TDMA. Instead of using a single antenna beam to maintain continuous com- munication with its entire coverage zone, the satellite has a number of narrow antenna beams that can be used sequentially to cover the zone. A narrow antenna beam has a higher gain than a broad beam, which increases the satellite EIRP and therefore increases the ca- pacity of the downlink. Uplink signals received by the satellite are demodulated to recover the bit streams, which are structured as a sequence of packets addressed to different receiving earth stations, The satellite creates TDMA frames of data that contain packets addressed to specific earth stations, and switches its transmit beam to the direction of the receiving earth station as the packets are transmitted. Note that control of the TDMA net- work timing could now be on board the satellite, rather than at a master earth station, In the above example, the VSAT earth station could transmit data to a baseband pro- cessing satellite using SCPC-FDMA to permit the use of a small antenna and low power transmitter. The satellite could then use satellite switched TDMA to send that data to mul- tiple earth stations, creating a mesh VSAT network. It is difficult to create a mesh VSAT network using SCPC-FDMA. Baseband processors are considered in more detail in the next section. 6.4 ONBOARD PROCESSING ‘The discussion of multiple access so far has assumed the use of a bent pipe transponder, which simply amplifies a signal received from earth and retransmits it back to earth at a different frequency. The advantage of a bent pipe transponder is flexibility. The transpon- der can be used for any combination of signals that will fit within its bandwidth. The disadvantage of the bent pipe transponder is that it is not well suited to uplinks from small earth stations, especially uplinks operating in Ka band. Consider a link between a small transmitting earth station and a large hub station via a bent pipe GEO satellite transpon- der. There will usually be a small rain fade margin on the uplink from the transmitting station because of its low EIRP. When rain affects the uplink, the C/N ratio in the transpon- er will fall. The overall C/N ratio in the hub station receiver cannot be greater than the C/N ratio in the transponder, so the bit error rate at the hub station will increase quickly a rain affects the uplink. The only available solution is to use forward error correction64 ONBOARD PROCESSING 247 coding on the link, which lowers tlie data throughput but is actually needed for less than ‘5% of the time. The problem of uplink attenuation in rain is most severe for 30/20 GHz uplinks with small margins. Outages are likely 10 be frequent unless a large rain fade margin is included in the uplink power budget. Onboard processing or a baseband processing transponder can overcome this problem by separating the uplink and downlink signals and their C/N ratios. The baseband processing transponder can also have different mod- ulation schemes on the uplink and downlink to improve spectral efficiency, and can dynamically apply forward error control to only those links affected by rain attenuation. All LEO satellites providing mobile telephone service use onboard processing, and Ka- band satellites providing Internet access to individual users also use onboard processing. Baseband Processing Transponders ‘A baseband processing transponder has a receiver and transmitter similar to those found in an earth station. The received signal from the uplink is converted to an intermediate frequency and demodulated to recover the baseband signal, which is then processed and reassembled, The baseband signal is modulated onto a carrier at a downlink frequency and transmitted back to earth. The signals are invariably digital, although that is not a re- quirement of a baseband processing transponder. The advantage of this process is that up- link and downlink signal formats need not be the same, and that different forms of error correction can be applied to the uplink and downlink. The C/N ratios of the uplink and downlink are not tied together through the reciprocal formula (Eq, 6.10), If the C/N on the uplink is low, because of an uplink rain fade, for example, bit errors will be present in the recovered data in the transponder. The BER will depend only on the uplink C/N ratio. If the C/N on the downlink is high, as is usually the case for all star networks work- ing with a large hub earth station, no additional bit errors will occur on the downlink, Separation of the uplink and downlink signals allows different modulation methods to be used, as well as flexible error correction codes. In star networks, the C/N ratio of the uplink and downlink between the satellite and the hub station is usually high because of the large antenna gain and high transmit power of the hub station. The high C/N ratio ‘can be traded for a high level modulation such as 16-QAM, which reduces the bandwidth required for the uplink and increases the spectral efficiency of the communication system. 16-QAM sends four bits per symbol, and requires only half the bandwidth of QPSK. ‘As an example, consider a system in which half rate forward error correction and QPSK is used on the uplink from a small earth terminal. For a message data rate of Rj bits per second, the transmitted bit rate will be R, = 2 Ry bps. An RF bandwidth of RJ2 X (1 + ayy) = Ry(1 + cp) H2. will be required on the uplink, where cry, is the roll- off parameter of the (assumed) RRC filter in the transmitter. On the downlink to the hub station, where the C/N ratio is high, 16-QAM can be used without forward error correc- tion. The RF bandwidth required for the downlink will be Ry/4 X (1 + ay,). where ay is the roll-off parameter of the RRC filter in the transponder transmitter section. If we assume the same roll-off parameter, a, in both the uplink and the downlink transmitters ‘and receivers, the downlink requires only one-fourth of the uplink bandwidth to send the same number of bits. The fourfold reduction in downlink bandwidth represents a consid- erable improvement in the spectral efficiency of the satellite system, Star networks with small uplink earth stations are widely used for VSAT systems and satellite links that carry Internet traffic. One hub station linked to an Internet Service Provider (ISP) can connect to hundreds of user earth stations located at the users’ homes. This is a major growth area for Ka-band GEO satellites, a number of which are planned Oo COTEETE EE248 cHAPTER.® MULTIPLE ACCESS for launch in the 2002-2005 time frame. The user terminal is typically a 0.5- to 9 Bn antenna with a low power transmitter, 0.5-1 W output, that can send data to the Sateljzg at rates up to 400 kbps. The uplink operates in the 30 GHz band with a small rain fade margin, resulting in low C/N at the transponder input and requi ing the use OF error co, rection on the uplink during rain fades. It may be possible to remove all or most of error correction when the uplink is operating in clear air, allowing higher data speeds & cept during rain fades. Rain fading occurs for less than 5% of the time, and affects oly a small number of users at any instant. Under most conditions, only a small number of Users will suffer severe rain fading at the same time, so most of the links to the satelite will not need heavy error correction. This flexibility can increase the capacity of the satel lite links by a factor of two relative to a bent pipe transponder, at the cost of considerably increased complexity in both the satellite and the user terminal. Satellite Switched TOMA with Onboard Processing Baseband processing is essential in satellites using satellite switched TDMA, because dat Packets must be routed to different antenna beams based on the address of the destination earth station. The data in such systems is always sent in packets which contain a heade and @ traffic section. The header contains the address of the originating station and the address of the destination earth station When satellte switched TDMA is used, te \ransponder must extract the destination information and use it to select the correct down. link beam for that packet. The satellite is operating much like a router in a terrestrial daa transmission system. Switched beam operation of an uplink from a small earth station s ‘more difficult to achieve because it requires synchronization of the earth station transmit Fre With the satelite beam pointing sequence, in much the same way that a TDMA up- link operates. However, the uplink can operate in a small bandwidth which overcomes the chief disadvantage of classic TDMA—the requirement for high burst rate transmissions and high transmit power. Satellite switched TDMA can greatly increase the throughput of a transponder. Con- sider, for example, a satellite providing Internet access to individual users in the United States. The uplink and downlink beams atthe satellite must provide coverage over an area approximately 6° by 3°, as seen from the satellite. Antenna gain and beamwidth are re. lated by the approximate relationship G = 33,000/(produet of beamwidths in degrees) ‘This limits the maximum achievable satellite antenna gain to approximately 32.5 dB, A satellite with switched beam capability can have much narrower beams with higher gain than a satellite with a single fixed beam. The limitation on gain is the diameter of the antenna, which must fit inside the launch vehicle shroud. For launchers available in 2000, this limit is about 3.5 m. At 20 GHz, the uplink frequency for Ka band, an antenta with a circular aperture of diameter D = 3.5 m and aperture efficiency of ne = 65% has a gain G = qq (wD/A)* = 55.4 dB, and its beamwidth is approximately 75 A/D degrees = 0.32°. The corresponding downlink antenna for 30 GH» that has ¢ beamwidth of 0.32° and a gain of 55.4 dB has a diameter of 2.33 m. The switched beam satellite has an antenna gain almost 23 4B higher than the single beam satellite, which an be traded directly for reduction in uplink or downlink transmit power, and uplink or downlink data rate. However, the satellite must generate at least 170 beams to cover all of the United States with 0.32° beams, with a consequent increase in satellite entenna complexity, Satellite switched TDMA and multiple beam antennas-are a feature of most of the Proposed Ka-band Internet access satellites". The Astrolink satellites, for example, have6.5 DEMAND ACCESS MULTIPLE ACCESS (oAMA) 249 105 spot beams for links to small user terminals. The satellite uplink (30 GHz) antenna has a diameter of 2.5 m and the downlink antenna has a diameter of 3.25 m. There are five spot beams for links to hub stations; the large antennas used by the hub stations al- low a lower gain antenna with a broader beam to be used on the satellite. Coverage of the United States with multiple beams is not always provided uniformly. Differences in population densities and the frequency of heavy rainfall make it advanta- geous to provide more system capacity to metropolitan areas, and also to provide higher link margins to areas with more frequent heavy rainfall, such as Florida and the south- eastern states. In the most sophisticated of large GEO satellites, a steerable phased array antenna can be used, with control of beam pointing from the ground via the satellite's telemetry and command link. The antenna beams can then be moved to provide coverage of areas with highest demand for traffic. The growth of the terrestrial optical fiber net- ‘work will eventually fulfill the need for high-speed access to the Internet. Where direct, access to an ISP is available via optical fiber, the transmission rate is likely to be higher and the cost to the user is likely to be lower. As the fiber network spreads through met- ropolitan areas, an Internet access satellite can concentrate its service on less well popu- lated and rural areas, A steerable beam antenna allows the geographical capacity of the satellite to be reconfigured throughout its lifetime. a Demand can be used in any satellite communication link where traffic from an is intermittent. An example is an LEO satellite system providing links to mo- bile telephones. Telephone voice users communicate at random times, for periods rang- ing from less than a minute to several minutes. As a percentage of total time, the use of an individual telephone may be as little as 1%. If each user were allocated a fixed chan- nel, the utilization of the entire system might be as low as 1%, especially at night when demand for telephone channels is small. Demand access allows a satellite channel to be allocated to a user on demand, rather than continuously, which greatly increases the num- ber of simultaneous users who can be served by the system. The two-way telephone chan- nel may be a pair of frequency slots ina DA-SCPC system, a pair of time slots in a TDM or TDMA system, or any combination or FDMA, TDM, and TDMA. Most SCPC-FDMA. systems use demand access to ensure that the available bandwidth in a transponder is used as fully as possible. In the early days of satellite communication, the equipment required to allocate channels on demand, either in frequency or time, was large and expensive. The growth of cellular telephone systems has led to the development of low cost, highly integrated con- trollers and frequency synthesizers that make demand access feasible. Cellular telephone systems use demand access and techniques similar to those used by satellite systems in the allocation of channels to users. The major difference between a cellular system and a satellite system is that in a cellular system the controller is at the base station to which the user is connected by a single hop radio link. In a satellite communication system, there is always a two hop link via the satellite to a controller at the hub earth station. Controllers are not placed on the satellites largely because of the difficulties in determining which Tinks are in use, and who will be charged for the connection. As a result, all connections pass through a controlling earth station that can determine whether to permit the requested connection to be made, and who should be charged. In international satellite communi- cation systems issues such as landing rights require the owner of the system to ensure250 cuarrens mutmipLe access that communication can take place only between users in preauthorized countrey zones, The presence of the signals from all destinations at a central earth station alg st ows security agencies the option of monitoring any traffic deemed to be contrary te the national interest. Demand access systems require two different types of channek a common signg, ing channel (CSC) and a communication channel. A user wishing to enter the communi. cation network fist calls the controling earth station using the CSC, and the controls, then allocates a pair of channels to that user. The CSC is usually operated in Fandom, access mode (see Section 6.6) because the demand for use of the CSC is relatively low, messages are short, and the CSC is therefore lightly loaded, a requirement for any Ds link. Packet transmission techniques are widely used in demand access systems because Of the need for addresses to determine the source and destination of signals. Section 69 discusses the design of packets for use in satellite communication systems, Bent pipe transponders are often used in demand access mode, allowing any cop. figuration of FDMA channels to be adopted. There seem to be few standards for demang access systems in the satellite communication industry, with each network using a diffe. cat proprietary configuration. Figure 6.11 shows a typical 54 MHz bandwidth Ku tang transponder frequency plan for the inbound channels of a VSAT network using frequency division multiple access with single channel per carrierand demand access (FDMA-SCPC. DA) on the inbound link. The individual outbound RF channels are 45 kHz wide, to ae. commodate the occupied bandwidth of 64-kbps bit streams transmitted using QPSK and RRC filters with « = 04. A guard band of 15 KHz is allowed between each RF channel, 80 one RF channel requires a total bandwidth of 60 kHz. A 54 MHz bandwidth transpon- der can accommodate 900 of these 60 kHz channels, but it is unlikely that all arc used at the same time, Many VSAT systems are power limited, preventing the full use of the transponder bandwidth, and the statistics of demand access systems ensure that the likelihood of all the channels being used at one time is small. Considerable backoff is required in a bent pipe transponder with large numbers of FDMA channels, as discussed earlier in this chapter. ‘The outbound link of this particular VSAT network is a continuous TDM bit stream transmitted through a separate transponder. A second transponder is used to allow for the differences in transponder gain needed for the inbound and outbound channels of the VSAT system. In VSAT systems, the inbound and outbound channels are usually symmetric, offering the same data rate in opposite directions. Internet access systems are often <———— Transponder bandwith 54 MHz——~———+ 64 kbps QPSK channels lesc}} 2 || 3 |} 4 |] 5 Ile 4898 | | 899 | [csc == == 60 kHz channel spacing 15 kHz guard bands between channels. FIGURE 6.11 Frequency plan for a 54-MHz transponder carrying 900 demand access channels, Each channel has an occupiad RF bandwidth of 45 kHz and carries ané 64-kbps signal, Channel 1 and channel 900 are common signaling channels (CSC) used by the ‘demand assignment system to set up access to the other 898 channels.e used at se of the that the ackoff is discussed it stream Ww for the the VSAT pmmettic, are often 65 DEMAND ACCESS MULTIPLE ACCESS DAMA) 251 asymmetric, because requests for information can be short but the resulting replies may be lengthy. The packet length of the TDM signal in the outbound direction may be fixed, which suits a symmetrical network, or variable, which better suits an Internet channel capable of downloading large files or video from the Internet. ‘The common signaling channels (CSC) shown in the inbound transponder in Figure 6.11 are located at the ends of the transponder bandwidth. When a VSAT earth station wants to access the satellite, it transmits a control packer to the satellite on the CSC frequency and ‘waits for a reply. The control packet is received by the hub earth station and decoded. The control packet contains the address of a terrestrial or satellite destination for the call, DA, the address of the station requesting the connection, RA, any other relevant data (such as a character, CP, to indicate that this is a control packet with no traffic data), and a cyclic redundancy check (CRC) that is used in the receiver to check for errors in the packet. The control station records both origination and destination station addresses and measures the Guration of the connection in order to generate billing data. In a true demand access sys- tem, the control station allocates the VSAT an uplink frequency and a time slot of specified duration in the outbound TDM frame. If the hub station has a large volume of data to send to a particular VSAT station, it can allocate a longer time slot in the TDM frame to that station, This is important in Internet access systems where a large file of video or other ‘multimedia data may have to be sent. The timeslots usually come in multiples of a fixed ‘minimum duration so that clock rates and buffer sizes are compatible. Ifthe system becomes busy and many stations are requesting large files, throughput to any one station will slow down toward the standard minimum rate, exactly as in a terrestrial Internet server. The outbound link transmits a continuous bit stream so that receivers can maintai carrier phase and bit clock synchronization. The data is organized into a sequence of pack- cts, addressed to the receiving stations, and organized into a frame, One frame contains ‘one packet for each earth station, as illustrated in Figure 6.12. In many TDM systems there is always a time slot with addresses and other information for delivery of data to cach earth station in the network, but there may be no data to send. In this case, the packet is assigned a special character to indicate no traffic. Once an inbound frequency and an outbound time slot are allocated to the VSAT station, the connection can be completed and data transfer or voice communication can begin. The hub earth station designed to receive FDMA-SCPC signals has multiple re- ceivers operating at different center frequencies and allocated to the many transponders on the satelite. ‘The block diagram of a receiver for one transponder with a bandwidth of 54 MHz is illustrated in Figure 6.13. The receiver amplifies and down-converts the received sig- nal to an intermediate frequency of 700 MHz and then to a second IF at 70 MHz. Indi- vidual FDMA-SCPC channels within the band 43 to 97 MHz are down-converted to a standard IF frequency of 2 MHZz in this example using local oscillators with frequencies 41 MHZ through 95 MHz in steps of 60 kHz. There are a total of 900 such 2-MHz IF receivers to cover all the frequency slots in the transponder bandwidth. End of TOM tre lomo, 0 > FIGURE 6.12 End of a TDM frame outbound to VSAT stations. A, station address; CRC, cyclic redundancy check; EOF, end of frame marker.252 cuaprens mucmpce access Downlink at £, GH2 LNA tage rejzction 700MH2 700 Mie 70 Me BPF iFampltiee Mixer "hpe SHXHEHS- RHE First Lo Second LO £,~ 700M 630 Mele roms x a [HERS EKHERD ke tg Ora b—-ECHEk; Ch 4 data Ch data Channel selection ou OMe psx ‘ters iz Bit recovery ‘Thitd LOs Mires 'BPFS amplifiers. democuitors ‘ensure ‘spaced 60 kHz FIGURE 6.13 Receiver for FOMA-SCPC hub station. Channel selection is by bandpass fiters in the 70 Miz IF section of the receiver and the third local oscillator. Each chennel Selection filter and third local oscillator is at a different frequency, spaced in 60 ktir inare- ments across the IF band. D, demodulator; 8, bit recovery circuit, The hub earth station transmits a continuous TDM signal to all VSAT stations in the network. The symbol rate and the bandwidth of the TDM signal depends on the max. imum bandwidth that the VSAT receiver can use, based on noise considerations. If the Retwork is symmetrical and uses all 900 possible FDMA-SCPC channels at a bit rate of 64 kbps per channel, the TDM signal must have abit rate of 900 X 64 kbps = 57.6 Mbps, ‘ignoring packet overhead. This bitrate is likely to be much too high for a VSAT station, resulting in low C/N ratio in the receiver. The transponder can be partitioned to carry ‘multiple groups of TDM signals with lower bit rates better suited to VSAT receivers EXAMPLE 6.5.1 FDMA-SCPC-DA A VSAT network consists of 250 Ku-band VSAT earth stations sharing one inbound and one outbound ‘transponder on a GEO satelite. The transponder bandwidth is 54 MHz. The transmit data bit rae for {he VSAT stations is 64 Kbps. Statistics for the VSAT network show that each VSAT generates an avs crane it ate of 5 kbps with random time of aval of data. The average outbound data rate per SAT stations 20 Kbps. The inbound data link operates in FDMA-SCPC using demand access, QPSK ‘modulation with a = 0:5 RRC filters, and half rate forward error correction. The outbound deta link {es 2 single continuous TDM stream, QPSK modulation, anda 16-bit CRC wordin each pace. De- tGimine the bitrate and bandwidths ofthe VSAT and hub station receivers, sugges a frame and packet size for the TDM link, an SCPC-FDMA frequency plan, and a demand access method Inbound Link: VSAT to Hub {The data rate at the VSAT station averages 5 kbps, bul arrives in the form of variable length mes Sages with random arrival times, The data are transmitted from the VSAT station on a half rate FECstations in n the max- ons, If the Dit rate of | 57.6 Mbps, AT station, 65 DEMAND ACCESS MULTIPLE ACCESS DAMA) 253, bit stream at 128 Kbps carrying 64 kbps of data, as a QPSK symbol stream at 64 kbaud. The link has a = 0.5 RRC filters, so the occupied bandwidth of the QPSK signal is R,(1 + a) = 96 kHz. ‘The IF receivers in the hub earth station will have filters with a noise bandwidth of 64 kHz (equal to the symbol rate of the signal) and total bandwidth of 96 kHz. Allowing a 20% guard band between RF channels requires a carrier-to-carrier spacing of 115 kHz, ‘The maximum number of channels that can be carried by the inbound transponder is '54,000/115 = 469, Several of these channels would have to be designated as common signaling ‘channels, so the maximum usable communication channels would be about 460. Ifthe transponder is power limited, fewer channels can be carried. Each VSAT channel transmits data at 64 kbps, but data arrives at an average rate of 5 Kbps. ‘Assuming the data does not require transmission in real time, a buffer can store data until a total of 120 kb has been collected, The data are then packetized (see Section 6.7) and overhead bits are ‘added for addresses, CRC, etc. The VSAT station can then send a request for a channel to the hub station and the data can be downloaded in about 2 s, The average time to collect 120 kb atthe VSAT is 24 5. This scenario favors a SCPC-FDMA multiple access scheme using a CSC to obtain access to the satellite once every 24 s. If we assume that it takes 2 $ to establish the connection and 2 s to transmit the packets, a total of 4 s of satellite transmission is used every 24 s, giving a VSAT sta- tion loading of 16.6%. This means that (in theory) six VSAT stations can share the same RF chan- nel. In practic, itis impossible to load the channels to 100% of their capacity because data arrives, ‘at random time intervals causing temporary overload when a large volume of data arrives at the same time. If we assume a 66% load factor for the inbound link, we can share one inbound channel between four VSAT stations, which requires a total bandwidth of 54/4 = 13.5 MHz in the transpon- det, Demand access has clearly achieved considerable savings in bandwidth and power in this case. Outbound Link: Hub to VSAT ‘The hub station transmits @ continuous TDM frame consisting of many sequential packels addressed to exch VSAT station, There are 250 VSAT stations with an average outbound data rate of 20 kbps teach, If we apply half rate FEC to the outbound data stream and use QPSK modulation, we will transmit at 20 kbaud per station, With 250 stations, the average outbound symbol rate is 5 Mbaud, and the outbound data rate is 5 Mbps because we are using QPSK modulation with balf rate FEC. ‘The occupied bandwidth of the signal with a = 0.5 RRC filters is 75 MHz, and the VSAT receiver noise bandwidth will be 5.0 MHz. Let us assume a frame length of 5 s. On average each VSAT will receive 100 kbps during @ 5.5 period, but the statistics of the traffic suggest that there will be megabits of data for some sta- tions and none for others in any given time period. Within the 5-s frame, if the frame time is di- vided equally, each earth station could receive data fora period of 20 ms, and would receive 10,000 data bits at 3 Mbps. If we assume a 5% overhead allowance for the packet, there will actually be ‘9500 data bits delivered during the 20-ms period, plus 500 overhead bits. If there are no data bits to be delivered to @ given station, only the overhead portion of the packet, 500 bits, needs to be transmitted. That allows other stations to use the spare time in the frame to send additional data at ‘a much higher than average rate, ‘Demand access is most valuable when the traffic mix changes a great deal. The multiple ae~ ‘cess system described here was designed to meet the needs of the average data rates transmitted on the inbound and outbound links, If many of the stations arc inactive, the other stations can have i ‘creased data rates. For example, suppose only 50 of the VSAT stations are active. Each VSAT sta. tion can transmit at its maximum data rate of 64 kbps, and will deliver data as fast as the terminal can supply it. To increase the inbound data rate above 64 kbps requires a wider channel bandwidth, and a receiver in the hub station with a wider IF filter. Alternatively, the VSAT station could trans- mit two carriers. The limitation on inbound data rate is likely to be VSAT EIRP and the resulting uplink C/N ratio in the transponder. SCPC-FDMA does not offer as much flexibility © change data rates as. TDM. ~ vane ‘On the outbound link with only 50 VSAT stations active, the packet length for the active sta- tions can be increased by a factor of five. Short packets must still be sent to all stations to main- tain synchronization of the VSAT receiver, but need consist only of a control packet, which is 5007 i bits in length. The outbound link has a bit rate of 5 Mbps. Each VSAT station receives a cong) il packet of 500 bits every 5 s if there is no data to be sent to that station, Hence the 200 inactive sy. ‘ tions receive an average bitrate of 100 bps, and the 50 active stations can share the remaining bi, ‘The bit rate for the 50 active stations is 5 Mbps ~ 200 x 100 bps = 4.98 Mbps, an average rae ‘of 99,600 bps for each active station. The normal packet length is 20 ms and delivers 10,000 bit ‘The packet can now be extended to almost 100 ms for each of the active stations. The actual packe, length is 100 ms — 80 us = 9992 ms to allow for the time required in each frame to send a sin, gle control packet to each inactive VSAT. Thus the active stationsreceive 499,600 its in each packer, ‘of which 500 are control bits, giving a data rate of 99.82 kbps. TDM frames which offer variable | packet length can easily accommodate a widely changing mixture of data rates delivered to each SAT station. A field within the packet header can tell the VSAT station how many dat in the packet, allowing great variability. ‘254 CHAPTER 6 MULTIPLE ACCESS | 6.6 RANDOM ACCESS (RA) : Random Access is a widely used satellite multiple access technique where the trafic density from individual users is low. For example, VSAT terminals and satellite mobile telephones often require communication capacity infrequently. These users can share transponder space without any central control or allocation of time or frequency, provided the average activity level is sufficiently low. Ina true random access network, a user trans- mits packets whenever they are available. The packet has a destination address, and a source address. All stations receive the packet and the station with the correct address stores the data contained in the packet. All other stations ignore the packet, unless itis designated as a broadcast packet with information for all stations. In satelite communi- cation systems, the network is more usually a star configuration, with a single hub and ‘many small earth stations or portable terminals, Inbound packets are received by the hub earth station and forwarded to their destinations. Early work on random access techniques for radio channels was done at the University of Hawaii, where the system was called ‘Aloha and was known by the generic term packet radio. | Random access cannot be used when traffic density exceeds 18%, and therefore makes inefficient use of the bandwidth available in the transponder. Although there is a saving in transmission time because no call set up is required, the low throughput and poor spectral efficiency has restricted random access use in satelite systems to cases where traffic bursts are short and highly intermittent. In general, itis used on single SCPC-FDMA channels, rather than on whole transponders, The common signaling channel, described in the previous section, is an example of an SCPC-FDMA random access channel withia a transponder that can successfully use random access because itis lightly loaded. 6.7 PACKET RADIO SYSTEMS AND PROTOCOLS Data transmission between computers or terminals requires agreed methods by which con- nections are established and data is transferred. When we make a telephone call, there are conventions and etiquette which define how a telephone connection is established and when each person speaks. For example, you decide to call your friend John Doe. You lift the telephone handset and hear dial tone. The telephone system is telling you that it is ready for you to dial a number. You dial the telephone number of your friend and wait 10 hear a ringing tone. The telephone system is telling you that it is trying to attract the ssc a imthe traffic lite mobile } can share Provided \user trans: ress, and a ect address unless it is, ; communi sle hub and by the hub techniques was called therefore h there isa jughput and cases where "PC-FDMA 1, described nel within | jaded. which con- I, there are blished and Joe, You lift ou that itis and wait 10 » attract the 67 PACKET RADIO SYSTEMS AND PROTOCOLS 255 attention of your friend. If your friend answers, you expect to hear “Hello, this is John Doe.” You might reply, “Hello, this is Bill Smith. How are you?” If the person who an- swers just says “Hello,” you cannot be sure that you have reached John Doe, and you ‘might have to say “Hello, this is Bill Smith, is that John Doe?” If you dialed a wrong number and have connected to the wrong person, you might say “Sorry, I must have di- aled a wrong number.” Then you would put the phone down and try again, If you weren't sure whether the answerer said “John Doe” or “John Roe” because of noise on the line, you would ask for a repeat transmission. "The process of creating a telephone connection to a friend makes use of signals pro- vided by the telephone system and a set of procedures based in human etiquette and in- telligence. Humans can readily determine whether they have reached the correct person, ‘and how to proceed if the call does not go through correctly. A data transmission system lacks the intelligence of a human, and cannot readily adapt to changing responses in the way that humans can. Data transmissions make use of packets and protocols to ensure that automatic connection and transfer of data can be achieved reliably without human in- tervention, One dictionary definition of protocol is, “The code of ceremonial forms and courtesies of precedence, etc., accepted as proper and correct in official dealings, as be- tween heads of state or diplomatic officials”"*. It is this sense of the word protocol that is used to describe the rules by which two data terminals can connect to each other through a communication system and then transfer data. ‘The creation of protocols for data transmission is a very large subject, with many books and papers devoted to the design and performance of different schemes. In this text, only the briefest summary of the subject is included. The widespread ownership of per- sonal computers and growth of the Internet helped spur development of efficient and pow- erful protocols like TCP/IP. The International Standards Organization (ISO) has created a seven layer model for machine to machine communication known as the open systems interconnection (OSI) which separates the functions of different parts of the system, The 1SO-OSI model is shown in Figure 6.14. Although the model is widely quoted as de~ scribing the structure of data communication systems, it rarely seems possible to identify seven separate layers within any given system. The lowest layer, the physical layer, is the ‘one with which this text is concerned—the transport of bits from one point to another. Regardless of the method of transportation, the ISO model assumes that bits are carried across the physical layer, in both directions, possibly arriving with some errors. The ‘Appication Presentation Session Transport Network Data link control Physical FIGURE 6.14 |SO-OSI seven layer model for digital communications links.PEE EEE 256 cHAPTER.6 MULTIPLE ACCESS remaining six layers of the model are embedded in the hardware and software of the te. minals at each ends of the link. The second layer of the model provides error detection and correction, either in hardware or software, and the remaining layers are responsibie for organizing the data transfer, from making connections to billing for the service, Terrestrial data communication has evolved through a series of protocols, beginning with a standard by IBM, known as HDL, through X.25 to ATM (asynchronous transfer mode, ATM uses a 53-byte packet and was designed to be transmitted over fiber-optic line networks using the DS-3 44.736 Mbps transmission rate according to an IBE standard (P8026), Digital cellular radio systems, although using three different standards, are compatible, and any cellular telephone designed for TDMA can work with any provider's TDMA system, Satel. lite communication systems have not evolved a set of standards, and many systems use proprietary protocols. For example, the Iridium, Globalstar, and ICO Global LEO/MEO sate lite communication systems all use digital transmission with different protocols. Handse designed for use with the Iridium system cannot communicate with Globalstar or ICO Global satellites. There is now intense interest in designing GEO satellite systems that are compat. ible with the ATM protocol so that wide band satellite links can be connected directly to ter- restrial data networks", Generically, this is known as wideband by satellite. Typically, the long delay time inherent in transmission via a GEO satellite creates problems when terfacing to a terrestrial protocol designed for much shorter delays. Special interfaces are needed at the earth stations that allow the protocol to be adapted for satellite use. i Data cannot be transmitted as a continuous bit stream in most cases, because ad. ditional bits for addresses, error control, and other additional information that is not part of the message data must be inserted into the bit stream. Data are sent in packets, usw- ally with an agreed length and content, using a structure that is very similar to the TDMA frame described in Section 6.3. Each packet typically consists of a header, which con- tains address and control information, a block of message data, and a closing section | with error control bits and an end of packet flag. One protocol and packet design that has been widely accepted for use in amateur satellite systems is AX.25. The AX.25 pro- tocol is based on X.25, a protocol developed for terrestrial data communications, and is used by amateur radio operators in a terrestrial data communication network. The pro- tocol was adapted for use in amateur radio LEO satellites with VHF and UHF transpon- ders operating in a store and forward mode. Several of these satellites were built and orbited, providing a method for amateur radio operators to send messages by satellite”, Figure 6.15 shows the structure of the AX.25 packet. All packets begin and end with a unique word, called a flag, O1111110, which is not allowed to appear in any other part of the packet. The flag marks the end of one packet and the start of the next packet, so that the receiving data terminal can extract the packet contents correctly. The general format of the packet contents is a header, followed by message bits, followed by a cyclic redundancy check (CRC). The header contains addresses, in the form of amateur radio call signs, for the sender and intended recipient, and control information that helps the receiving station identify the contents of the packets. The control bits, for example, tell the receiving ter- minal how long the packet is, and define whether this is a broadcast packet, intended to be viewed by all receiving stations, or a packet for a specific recipient. Control bits also spec- ify the type of packet—some packets contain no message bits and are sent to convey syS- tem information. The CRC allows the receiver to check whether the. packet was received correctly, and to call for a retransmission if an erroris detected. The interested reader should refer to reference 17 for further details of the amateur radio satellite system. All data transmission system must have some form of protocol, and data is almost al- ways sent in packet form. Thus whenever multiple access techniques are discussed and dig- ital data are transmitted, it can be assumed that some form of packet transmission is used. Ht . —8 CODE DIVISION MULTIPLE ACCESS (CDMA) 257 ihe: tor Flag Address Control} FOS | Flag eSPonsible. abits 112-S60bits «Bits Tits Shits rvice, ~ ‘AX.25 Un-rumbered frame beginning sfer mode) networks 6). Digit ~ nse: - oa [ Fes assess [Cowal] in| Message | Fs | Flea oe 5 Bbits 112-560bits Bbits abits Nx 8 bits: 16 bits 8 bits TS Use MEO sate, |AX25 Information frame ; conn Flag =01111110 FCS = Frame check sequence - 4 FIGURE 6.15 AX.25 packets. The unnumbered frame (packet) is used only for control, oily oe ‘and contains no message data. All packets (frames) start and end with Flag, a unique word Y to tee vith six ones. Data messages are processed to ensure that six ones do not appear as @ Typically, String anywhere except Flag, The value of NV can be anywhere from 1 to 256. FCS; Frame 8 when in- control sequence. efaces are ecause ad- is not part kets, use he TDMA /hich cons ng section lesign that X.25 pro- ns, and is . The pro- transpon- built and satellite!” dend with other part cet, so that format of edundancy signs, for ng station ng ter- ided to be also spec- mnvey sys s received Jer should | @8 CODE DIVISION MULTIPLE [ACCESS (CDMA) ee —————— Code division multiple access is a scheme in which a number of users.can occupy all of the transponder bandwidth all of the time, CDMA signals are encoded such that in- Formation from an individual transmitter can be recovered by a receiving station that knows the code being used, in the presence of all the other CDMA signals in the same bandwidth. This provides a decentralized satellite network, as only the pairs of earth stations that are communicating need to coordinate their transmissions. Subject to transponder power limitations and the practical constraints of the codes in use, stations with traffic can access a transponder on demand without coordinating their frequency (as in FDMA) or their time of transmission (as in TDMA) with any central authority. Each receiving station is allocated a CDMA code; any transmitting station that wants to send data to that earth station must use the correct code. CDMA codes are typically 16 bits to many thousands of bits in length, and the bits of a CDMA code are called chips to distinguish them from the message bits of a data transmission. The CDMA chip sequence modulates the data bits of the original message, and the chip rate is always ‘much greater than the data rate, This greatly increases the speed of the digital trans~ mission, widening its spectrum in proportion to the length of the chip sequence. As a result, CDMA is also known as spread spectrum. Direct sequence spread spectrum (DS- §5) is the only type currently used in satellite communication; frequency hopping spread spectrum (FH-SS) is used in the Bluetooth system for multiple access in short range 1o- cal area wireless networks'®, = saroeaeaeeal CDMA was originally developed for military communication systems, where its purpose was to spread the energy of a data transmission across--wide bandwidth to make detection of the signal more difficult (called low probability of intercept). Spreading the energy in a signal across @ wide bandwidth can make the noise power spectral density (NPSD) in the receiver larger than the power spectral density (PSD) of the received sig- nal, The signal is then said to be buried in the noise, a common feature of DS-SS signals, almost al- d and dig- nis used.CHAPTER 6 MULTIPLE access and the signal is much harder to detect than a signal with a PSD greater than the receiver's NPSD. The correlation process that recovers the original data bits from a DS.SS Spread spectrum signal is also resistant to jamming, the deliberate transmission of a radio Signal at the same frequency to blot out someone else's transmission. Both of these attributes are valuable in tactical military communication systems, CDMA has become popular in cellular telephone systems where it is used 10 en. hance cell capacity. However, it has not been widely adopted by satelite communication systems because it usally proves to be les efficient, in terms of capacity, than FDMA and TDMA. The Globalstar LEO satelite system was designed to use CDMA tor mult. Ble access by satelite telephones; the advantage of CDMA in this application ie soft hand. off. More details of the Globalstar system can bé found with, reference 4.0 The GPS navigation system uses DS-SS CDMA for the transinision of signals that Lous codies have been developed specifically for this purpose, such as Gold and Kasam codes'. rhe DS-SS codes will all be treated as pseudonoise (PN) sequences in this discus- Sion, Pscudonoise refers tothe spectrum of code, which appears to bea randens sequence oF bits (or chips) with a Mat, noiselike spectrum, ‘The generation of a DS 66 signal is puustated in Figure 6.16, We will begin by assuming thatthe system uses baebang signals Modulator | Ej Incoming bit stream Fy bps Fe py Spreading PN sequence NAy bps: FIGURE 6.16 The basic principle of a direct sequence spread spectrum (CDMA) system. F2ch incoming message data bit is multiplied by the same PN secuenen: In this example the message sequence is -1 +1 and the PN sequence is 41 +1 rt 44 1-1.68 CODE DIVISION MULTIPLE ACCESS (coma) 259, Ry, converted to have levels of +1 and —1 V corresponding to the logical states | and 0, is multiplied by a PN sequence, also with levels +1 and ~1 V, ata rate MX Ry chips per second, Each data bit results in the transmission of a complete PN sequence of length M chips. In the example shown in Figure 6.16, the seven chip spreading code sequence is 1110100, which is converted to +1 +1 +1 —1 +1 ~1 ~1. The spreading sequence mul- tiplies the data sequence 0 1, represented as ~1 +1, leading to the transmitted sequence |] Pa |B Bs el CO rae Phase shifters Adder Jo Output vo=—by~ Bp + Bs ~ By + Be + By + by Bits clocked Bits clocked in Register contents Output in Register contents Output oft 444444 elt |i Vom t frat tt tues oft |r afralt 1 11 t]uss nol ta] afasialt tt t]ues ufaa aati afwet f2]4 21 4 141 t]v=8 wlar-at tt [t}yet wi4-t1 41 4 1] m=7 Yo=-7 FIGURE 6.17 Data bit recovery using an IF correlator (matched filter). In this example the PN sequence is seven bits long for illustration. The CDMA chips from the receiver are clocked into the shift register serially and the shift register contents passed through phase shifters and added. The phase shifters convert ~1 chips to +1 when the correct code is in the shift register such that all the voltages add to a maximum when the received sequence is, correct. This figure shows the shift register contents and adder output for the chip sequence in Figure 6.16, Note that a high spurious output of 5 occurs at the third clock step, indicating that the seven bit sequence used here for illustration has poor autocorrelation properties.260 chapters MULTIPLE Access , C oO CL __, Moser TE t= Ct UT LI Aa] Vcoming spread 8 roam Recovered it sreum GAR Despreading PN sequence FIGURE 6.18 A baseband correlator for dispreading CDMA signals. The original bit stream is recovered by multiplying the received signal by a synchronized copy of the PN sequence that was used in the transmitter. Ata CDMA receiver which knows the seven bit code, there will be a correlator that has the code stored as multiplier settings. Figure 6.17 illustrates the correlation process, Received chips are clocked into a shift register of length equal to the code sequence — seven stages in this case. The word in the shift register is identified as b, by .. . by. At each clock cycle the seven chip word with chip values b, in the correlator shift register are multiplied by +1 or ~1, corresponding to the chips in the code sequence, by the blocks marked phase shifters in Figure 6.17. Note that received chips are clocked into the correlator from the left, so the code sequence appears reversed (written from right to left) in the phase shifters in Figure 6.17. The outputs of the phase shifters are added to give the output word vp = pyb, + pads + Psbs + pads + pabs + pybg + pyby. The value of will be +7 or —7 when the correct code sequence exactly fills the seven stages of the shift register. Figure 6.17 shows one process by which the code sequence can be detected, The shift register is originally filled with +1 chips, giving an output from the adder of = +1. The sequence generated in Figure 6.16 is clocked in from the left, The adder output yields values of vy = +1 +5 +3 +3 +1 —1 as the chip sequence moves into the register. On the next clock cycle, when all seven bits of the sequence are in the seven stage register, the output of the adder is ~7. A threshold detector after the adder with 4 threshold level of 6 would detect the threshold crossing and output a logical 0 for the first data bit. As the next 7 bits are clocked through the shift register, the output of the adder fluctuates between —| and +5, reaching v) = +7 or ~7 when the code se- ‘quence fills the shift register, Note that the seven bit sequence used for illustration here would be a poor choice for a CDMA system because of the high spurious output from the adder at the third clock step. PN sequences used in CDMA systems are required to have good autocorrelation and good cross-correlation properties to minimize false threshold crossings. Multiple shift registers like the one shown in Figure 6.17 can be Operated in parallel with each input delayed by an increment of one bit. If there are NV shift registers, 1V possible code positions are tested at each chip clock cycle giving faster code acquisition, When the code sequence is long, the multistage shift register detector shown in Fig- ure 6.17 becomes unwieldy. Chip-by-chip multiplication is used instead. The multiplier has inputs 6, and p,, where b, is the received chip and p; is a stored PN sequence chip.ee N 6.8 CODE DIVISION MULTIPLE ACCESS (coma) 261 ‘The output of the multiplier is then integrated over the duration of the code sequence to yield a value +N or —N where N is the number of chips in the PN sequence. The process js illustrated in Figure 6.18 for the seven chip code of Figure 6.16. The incoming signal at the left side of the figure is multiplied by the despreading code sequence to give the ‘output at the right of the figure. In practice, a low pass filter is used instead of an integrator to avoid the need to synchronize discharge (dumping) of the integrator contents when a bit is detected. If the correct code is present in the input signal, the output of the multiplier is +1 or ~1 as each chip is present in the multiplier. In a practical DS-SS CDMA system, there will be several other CDMA signals present at the correlator input as well as the wanted code, The integrated value of the multiplier output for cross-correlation of any ‘other code with the stored code will yield a value +1 or —1 if the codes have ideal cross- correlation properties. Given 2 sufficiently large value for M (i., a long PN code sequence), it is possible to detect the bits of the wanted code in the presence of a large number of unwanted CDMA signals. ‘The seven chip code used to illustrate DS-SS code correlation is not a good code for a CDMA system. It exhibits poor autocorrelation because at one position in the shift register the adder output is +5, too close to the peak value of +7. Noise added to the incoming code could easily push the correlator output above the threshold caus- ing a detected bit error. Ideally, we would like to have CDMA codes of length M chips that have autocorrelation values of +1 or ~1 everywhere except when the code is aligned correctly, when the value should be M. When a different code is clocked into the shift register in Figure 6.17, the cross-correlation should be +1 or ~I on all clock cycles. Very few known codes have these ideal properties. There are Barker codes with sequence lengths up to 23 chips that meet these requirement, but practical DS-CDMA systems normally use longer codes with nonideal correlation properties. For example, GPS satellites use 1023 chip Gold codes for the C/A (course acquisition) code se- ‘quence that are built up from maximal length sequences which are easy to generate with a shift register. Practical CDMA systems use BPSK waveforms and correlate the received sig- nals at IF rather than baseband. The shift register shown in Figure 6.17 is typically a single stage multiplier, as shown in Figure 6.18, and the incoming signal and the PN sequence are BPSK waveforms with 0° or 180° phase shifts. Multiplication of two identical, cophased BPSK waveforms yields an output of +1. If the phase of the in- put waveform is reversed (indicating that the original data bit was a 0 rather than a 1) the output is —1. Coherent phase detection is required so that the IF waveforms can be added in phase, but the correlation principle is the same. The main difficulty in DS- SS CDMA receivers is that the received signal is buried in the noise, so the usual techniques for carrier recovery cannot be used. Baseband correlation is rarely used in DS-SS CDMA systems because the signals entering the correlator have C/N ratios less than 0 dB (negative C/N), so the signals always look like noise. A complete DS-SS receiver and correlator for the GPS C/A DS-SS signal is described in Chapter 12. The GPS C/A code sequence is 1023 chips in length, so chip-by-chip multiplication is used in the receiver. ‘The bandwidth occupied by the original data signal with bit rate R,, if transmitted using BPSK and a = 04 square root raised cosine (RRC) filters, would be 1.4 Ry Hz. ‘The spread spectrum signal occupies a bandwidth M x 1.4 R, Hz, and must be received through an IF RRC filter with a noise bandwidth of M x R, Hz, Suppose that a BPSK receiver with the appropriate RRC filter with noise bandwidth R, Hz receives the BPSK signal with C/N = 11 dB. If we do not change the power level of the original BPSK sig- nal by the process of spreading it into a bandwidth of 1.4 M R, Hz, the C/N in the spread——— 262 CHAPTER 6 MULTIPLE ACCESS spectrum receiver will be 11 ~ 10 log,oMf dB. If M is large (c.g., 1023 as in the GPS C/ code) the CDMA signal will have a C/N ratio in the receiver much less than 0 dp; 19.1 dB for the GPS C/A signal example. The despreading process using a correlator to recover the original signal adds a processing gain equal to the numerical value M to the (C/N)ss ratio of the received spread spectrum signal. Hence the signal-to-noise (S/N)... ratio in the spread spectrum receiver after the correlator is given by (S/Nou = (C/N)ss + 10 logioM (6.14) ‘This $/N ratio must be sufficiently high for the receiver to recover the bits of the trans. mitted signal with a reasonable bit error rate. For example, if a BER no larger than 10°$ is required, (S/N).a must be greater than 11.0 dB, allowing a 0.4-d8 implementation mar. gin with no forward error correction. DS-SS CDMA Capacity In a DS-SS CDMA system where there are a number of CDMA signals present at the in. put to each receiver, it is usual to treat the unwanted (interfering) CDMA signals as noise, If a receiver has an input containing Q signals, each at a power level C watts, and the receiver thermal noise power is N, watts, the (C/N)iq ratio for the wanted signal is approximately (C/N)ig = 10 logio( CAM, + (Q ~ 1) x C)] dB (6.15) where [N, + (Q — 1) x C) watts is the total noise at the receiver input. The term (Q ~ 1) x C = I watts is the power of the Q — | interfering CDMA signals, (Note that N, and C must be added in watts, not decibel units.) The correlator in the receiver adds a processing gain of 10 logy Mf dB to the input C/N, as seen in Eq, (6.14), and outputs a correlated signal with a signal-to-noise ratio (S/N)oa. Hence the output S/N ratio for the bit st?eam inthe receiver is given by = . lhe | (9/Noa = 10 logiol CAM + (O~ 1) X O)] + 1 logioM dB; (616) If Q is a large number, itis probable that [N, + (Q — 1) ¥ C] = @ ~ 1) x C watts, and then Eq. (6.16) reduces to (S/N)ou = 10 logigl 1/(Q ~ 1)] + 10 logo Mf = 10 loglM/(Q — 1)} 4B 6.17) If Qis also large such that M>> 1 then (S/N)ou = 10 logiM/Q) 4B (6.18) Examination of Eq, (6.18) shows that Mf, the number of chips in the spreading code must be 10 times larger than Q if the output S/N ratio is to be greater than 10 dB, and that the system capacity is independent of the thermal noise power in the receiver. The bit rate of each signal is given by R, = RN = B/{N X (1 + a)] (6.19) where R, is the chip rate and B is the transponder bandwidth. The total bit rate for the transponder is given by M x Ry = BX M/{Q X (I + a) If M must be 10 times larger than Q to allow demodulation of the spread signal without many bit errors, the total bit rate through the transponder in bits per Hertz using CDMA will be numerically less than one-tenth of the bandwidth in hertz. This results in poor utilization of the RF band- width when CDMA is used, compared to FDMA or TDMA, as the following example demonstrates, a6.8 CODE DIVISION MULTIPLE ACCESS (CDMA) 263 EXAMPLE 6.8.1 CDMA in a Fixed Earth Station Network ADS-S$ CDMA system has a number of earth stations sharing a single 54 MHz bandwidth Ka-band transponder. Each station has a different 1823-bit PN sequence which is used to spread the trafic bits into a bandwidth of 45 MHz. The transmitters and receivers use RRC filters with a = 0. and the chip rate is 30 Mops. Determine the number of earth stations that can be supported by the CDMA system if the correlated output S/N = 12 dB. Equation (6.17) gives (S/N) = 1248 10 log(M/(Q ~ 1)] = 30.9 - 10 1og(Q- 1) Hence 10 log(Q ~ 1) = 18.94B Q=7-1=76 Bach of the carriers has a bit rate of 30 Mbps/1023 ~ 29.33 kHz, so the transponder carries a to- tal bit rate of 77 X 29.33 kbps ~ 2.258 Mbps. A 54-MHz bandwidth transponder operated in DMA. ‘or TDMA would have a much higher capacity ‘The capacity of the system can be improved by adding half rate forward error (FEC) control to the baseband signal to reduce the S/N required for detection of the bits in the receiver. If the FEC system has a coding gain of 6 dB, we can use S/N = 12 ~ 6 = 6 dB. Using Eq, (6.18), because we now know M=>Q 64B = 10 log(M/Q) gives Q~ M/4 = 255 channels. The data bit rate of each channel (before application of half rate FEC) is now 14.66 kbps and the total throughput of the transponder is 255 * 14.66 Kbps = 3.74 Mbps. This is still well below the capacity of a FDMA or TDMA system. We can conclude that CDMA is useful in commercial systems only where efficient use of satellite capacity is not important, or where the ease with which stations can leave and join the net- ‘work outweighs the loss of efficiency, or where power limitations in the transponder ensure that it cannot be heavily loaded. a EXAMPLE 6.8.2 CDMA in an LEO Satellite Network ‘An LEO satellite communication systems uses direct sequence CDMA as the multiple access method for groups of terminals within each of its multiple antenna beams. The terminals generate and re- ceive compressed digital voice signals with a bit rate of 9.6 kbps. The signals are transmitted and received at a chip rate of 5.0 Mbps as BPSK modulated DS-CDMA. In the absence of any other CDMA signals, the input power level at the receiver input is ~ 146.0. dBW for one CDMA signal, and the noise temperature of the receiving system is 300 K. The satelite transmits 31 simultaneous CDMA signals. Find the S/N ratio for the 9.6-kbps BPSK signal after despreading, and estimate the BER of the data signal, given a system implementation margin of | dB. If two of the multiple ‘beams from the satellite overlap, s0 that a second group of 3} DS-CDMA signals is present at the receiver, find the BER of the wanted signal ‘The thermal noise power in the receiver is N, = k7;B,,. For the chip rate of 5.0 Mbps with BPSK and ideal RRC filters, B, = 5.0 MHz. Hence 228.6 + 248 + 67.0 = —136.8 dBW = 209 x 10" W ‘There are 30 interfering CDMA signals overlaid in the 5-MHz bandwidth of the receiver fil- ter. The total interfering power is 1 = 30 P, = ~146.0 + 148 dB = —131.2 dBW = 7.59 x 10 W ‘The carrier-to-noise plus interference ratio must be calculated in watts, not dBW, because we cannot add noise and interference in decibel units, only in watts. The carrier-to-noise ratio in the264 cHaPTen 6 MULTPLE ACCESS receiver for the wanted CDMA signal is CAN, +1) = 251 x 10-"/(2.09 x 10-4 + 759 x 10-14) 251/968 = 0.0259 = -15.9 eB The carrier power is well below the noise plus interference power, so the wanted carrier ig hidden below the noise and interference. This is called a low probability of intercep signal. CDMA ‘was first used by military radio communication systems because detection ofa signal which is below, the noise floor is difficult. ‘The coding gain, G., for the CDMA receiver is given by the chip rate divided by the bit rte Gi. = R/Ry = 5.0 Mops/9.6 kbps = 5208 = 27.2 4B Hence, after correlation of the wanted code (despreading), the S/N ratio of the 9600-bp, BPSK signal is S/N = ~15.9 + 272 = 11.348 ‘With an implementation margin of 1 dB, the effective S/N is 10.3 dB = tio. For BPSK, the BER is Pe 10.7 as a power ra $ erfel ViCTNVarnal = $erfel3.27] = 2x 10°* Ifa second group of 31 signals is present at the receiver from an overlapping satelite beam, there will be additional interference which lowers the C/(N + 1) ati. The interfering power from 31 signals is 1=31XxP, —146.0 + 14.9 dB = ~131.1 dBW = 776 x 10° Ww Hence the new C/(N, + 2) ratio is CAN, + D = 2.51 X10 251/174, = 18.408 "12.09 x 10" + 7.59 x 10 + 7.76 x 10-4) ors ‘After correlation of the wanted code the $/N ratio of the 9600 BPSK signal is S/N = ~18.4 + 272 = 8.84B With an implementation margin of 1 dB, the effective S/N is 78 JB = 6.02 as a power ratio, For BPSK the BER is P, Serfel VICTNVarmin] = $erfo(245] = 3 x 10+ We would need to add forward error correction to the baseband signal to improve the bit er yor rate. To achieve a BER of 10°* in this case, a coding gain of about 3 dB would be adequate With half rate convolutional coding, 2 coding gain of 5.5 to 6 dB is typical, which would provide 4 margin of 3 dB over a BER of 10° and a baseband data rate of 48 kbps. This bit rate will sup- Port a single digital speech channel with LPC linear predictive coding compression. The advantage of overlapping beams in a mobile satellite system is that the wanted signal can be transmitted by both satellites (using different CDMA codes) and blockage of one beam by an obstruction on the ground does not cause loss of the signal if the other beam can still be re. ceived. The wanted signal from both satellites cam be combined at baseband using a rake receiver, which improves the BER, With optimum combining of the same baseband signal, the BER will be the same as for a single beam with 31 users. a MPLE 6.8.3 GPS ‘The Global Positioning System (GPS) uses direct sequence CDMA for both the C/A and P code {cansmissions. The design and operation of GPS is discussed in detail in Chapter 12, from which this example of a direct sequence CDMA system is drawn,88 CODE Division, MULTIPLE access (coma) 265, tease ay ie PS satlites are 1023-bit PN sequences, formed = iA Gola ote a slime 0 me i 10.GPS satcites a Visible, so interference wi ; wot aume fr Simplicity that thee than ning Overltid COMA signals. In the ample we ei atsume * Signals, ate all r ith equal power, ‘There are variat ns wit po ‘" the transmitted ove beige salen Cava one ees ieoath, $0 there will be atiatons in tere lite close to the horizon actice pe The received Be Pover level of individual satelite signals in OWEr lev table below, assuming Q 8 typical C/A signal is een 1.023 Mbps using BPSK san ys The Cy, ‘given by the downlink budget in oe ‘ code is transmitted at a bit rate of NE" Moise bandwidth ig assumed to be 2 MHz, tlie agp (eBay 268 met ~ 1868 p, canynem fin (ap) ol The interference frm sing Sieting " is given by ‘A code Spectrum signals of equal power for the C, T= 1600 4 95 = = 891 x 10 W ‘he AHEM Noise Povey in a noi ; 2B Kis kB, where The De nrg sai = lately smal decease FS interference fron Sher Steins ay LA co The C/N for one Ca as AN + 1) = ~1407 aBw vw ». dB, the C/N ratio for the wanied signal oe Nor © Feceiver, thermal noise is the dominant xample 6.8.2, * Signal wig nis The theoretical M6 interfering signals is Coding gain f the S/N ratio of the 8 ~ (1492 0 81025 4, correlated cy mae i code sequence j 19.348 "™i08 ideal correlation ig Be~34 391 2 lo8ap npitce v on man" from the sorrelator is a 1-kbps polar binary cae mine te eH The Primary use OF this signal in 2 commercial GPS receiver is to de tom exch oe ore cya ode seauencee We time that code sequences arrive from each of the GPS sates Prong guenees, ai int code sein ace are moan apter 1 Navigation data are Modulated onto ter details) samples in succession for egg 0 da ROD ine tation margin of | dB in recone 1 at 50 He, so there are 20 svor20-v Y of th te ¥tton message alloning an implemen. Zyfie 2 BER around 6319/4 Signa “lective S/N afterthe coe 988 ih Vat: gving «BER slose to 19 i Sop 29 samples i inerove ie ror by fir of cut Theoretical, there wij WBF nh Te he nag cate message repeats every 1% 0 and ving ™ overwritten by new datay ys. The vigation message once every 2 eo Tepeats every 30 s The error sed SnD ee th CPS sete MA as a way to obtain ® high-speed bitstream from wie timing inte a oxcellent use of CD for ‘an be obtained, an essential ingredient foCHAPTER 6 MULTIPLE ACCESS any time of arrival position location system, and also a low speed data steam that provides the ny. igation information for the solution of the location problem, ‘The example here uses nine interfering satelites with the same power level. Most Gps receivers select the four strongest satellite signals to use in the position location soltion. A moe realistic scenario would have four satelite atthe maximimn receive power level and the remain at a lower level, since GPS satellites orbit in constellations of four, with one constellation always TRIE semProve the accuracy of postion location measurements. Thus we should expect less they 07 B degradation in C/N due to interference by other satellites’ CDMA signals, and the prof bility of a bit error in the navigation message then becomes very small. Jerestingly, the eastern European satelite navigation system known to the Westera work 35 GLONASS uses FDMA for multiple access. There are a maximum of 64 individual C/A signa, in the GPS system, from a possible 64 GPS satelite, each of which could be allocated its own Ri frequency within a | MHz band. If all the parameters of the GPS system were held the same, by Mnultiple aecesses were changed from DS-CDMA to FDMA, a multichannel receiver could eceine {oar (or more) 1 kbps BPSK data streams with C/N ratio of 108 dB. This is essentially how the GLONASS system works, with accuracy comparable to that achieved in the GPS system and a mura simpler receiver. 6.9 SUMMARY aT Multiple access is the process by which a large number of earth stations interconnect their links through a satellite. In frequency division multiple access (FDMA), stations are separated by fre- quency, while in time division multiple access (TDMA), they are separated in time. In code divi- sion multiple access (CDMA), stations use spread- spectrum transmissions with orthogonal codes to share a transponder without interference. Multiple ‘access may be preassigned or demand (DAMA), de- Pending on whether or not it responds to changing Itaffic loads. Frequency division multiple access is the most widely used multiple access scheme. In it each earth Station is assigned frequency bands for its uplink transmissions. Because of the TWT backoff re- quired to reduce intermodulation distortion with bent pipe transponders, the spectral efficiency (ie., the number of channels that can be carried per ‘megahertz of bandwidth) degrades with the number of stations that access a transponder. FDMA is widely used with VSAT earth stations and SCPC systems where the uplink from the earth station is at a low power level In time division multiple access (TDMA), earth stations transmit in turn, Since only one carrier is Present at a time, no TWT backoff is required and thus full transponder EIRP is available, TDMA per- formance does not degrade with the number of ac- cesses, TDMA transmissions are organized into frames; a frame contains one or two reference bursts that synchronize the network and identify the frame and a series of traffic bursts. Each participating sta- tion transmits one traffic burst per frame, Frames and individual traffic bursts are identified by stan- dardized bit sequences called unique words. One of the major technical problems in implementing ‘TDMA is synchronization. Once synchronization is acquired, it must be maintained dynamically 10 compensate for orbital motion of the spacecraft, TDMA is often combined with FDMA, $0 that a small number of carth stations share a TDMA. frame forming one FDMA access to a transponder. This is called MF-TDMA, In code division multiple access (CDMA) sta- tions transmit at the same time and in the same fre- ‘quency bands using spread-spectrum (SS) techniques. CDMA avoids the centralized network control re- uired for synchronization in TDMA, but tends to achieve rather poor spectral efficiency. The Global- star LEO satellite system was designed to use CDMA, with the advantage that an earth station can receive the same signal from more than one satelite at the same time, allowing soft handoff between satellites, Random access is used in systems that have low ‘taffic requirements and can tolerate less than 18% utilization of the RF channels. The advantage of ran- dom access is that no central network control is needed. Digital links between computers require protocols to ensure efficient transfer of data, and invariably use some form of packet communication, Satellite sys- ‘tems have tended to use proprietary protocols, with the result that different satelite systems are not ‘compatible.PRoptems © 267 | tems, Prentice-Hall, Englewood Clifs, NJ, 6th Ed., 1998. "4, Fennet.G. Staemuek, Introduction 10 Communication ‘Systems, Addison-Wesley, Reading, MA. 3rd Ed, 1984, 1992 “4, wovwglobalsta-com |S. K, Miva, ed, Saree Communications Technology, KDD Engineering and Consulting, Tokyo, Japan, 1981 6, Tw T. Ha, Digital Satellite Communication, McGraw- Hill, New York, 1990. 1. Inelsat TOMAIDSI System Specification (TDMAIDST Traffie Terminals), (BG-42-65E Rev. 2), Intelsat, ‘Washington, DC, June 23, 1983. 8, Technical Requirements for Inmarsat Standard-A Ship Earth Stavions (Issue 2) International Maritime Satel- Ite Organization, London, UK, February 1983 9, Rover M, Gacuarot, Satelite Communications, Lifetime Learning Publications, Belmont, CA, 1984, | 10. K. Feuer, Digital Communications: Satellite Earth Sta- tion Engineering, Prentice-Hall, Englewood Cliffs, NJ, 1983. I. VK. Baancava, D. Haccoun, R. Matyas, and P. NuspL, Digital Communications by Sallie, John Wiley & Sons, New York, 1981 PROBLEMS 12, J, 5. Suen, Ir, Digital Communications by Satellite, Prentice-Hall, Englewood Cliffs, NJ, 1977, 13, Astrolink URL. |. Spaceway URL. 15. Webster's New World Dictionary, 2nd Ed, William Collins ++ World Publishing Company, 1980, 16. Ian F. Axviuoiz and Scone-He Jens, “Satellite ATM Networks: A Survey.” IEEE Communicarions Maga- zine, Vol. 35, No. 7. pp 30-39, July 1997. 17. Dawporr, Amateur Satellite Handbook, American Radio Relay League. AARL, Newton, CT. 18, www Bluetooth.com 19. wwwrglobalstar.com 20, “Broadband via Satellite.” IEEE Communications Mag- azine, Special Issue, July 1997. 21. “Broadband Satellite Network Performance.” IEEE Communications Magazine, Special Issue, March 1999, 22. C. K, Tom and V. 0. K. Li, “Satelite ATM Network Architectures,” [EEE Nenvork, SeptemberfOctober, 1999, 23, B. G. Evans and R. Tarazouns, “Future Multimedia ‘Communications via Satellite,” Imemational Jounal of Satellite Communications, 14, 467474, 1996, 1, You are designing an FDM-FM-FDMA analog link that will occupy 36 MHz of an INTELSAT VI transponder. The uplink and downlink center fre- quencies of the occupied band are 5985.5 MHz and 3760.5 MHz. The distance from the satellite to your ‘arth station is 40,000 km. The saturation uplink flux density for your uplink is ~75 dBW/m? and the satel- lte’s G/T is = -11.6 dBK~'. At saturation the transponder EIRP for your downlink is 29 dBW and the earth station's G/T is 41 dBK”'. The transponder is linear in that the EIRP in dBW is BO 4B below the saturation value when the uplink flux density is backed off BO dB below saturation. The intermodu- lation carrier-to-noise ratio, (C/N), in dB, is related 1o the backoff BO in dB by (C/N) = 786 + 0.714 x BO In other words, at saturation the value of (C/N), is 7 86 dB. Find the maximum overall carrier-to-noise ratio (C/N), in dB that this link can achieve. What backoff must be used to achieve it? (When you need a frequency in your calculations, use the uplink or dovlink center frequency as appropriate.) Make your calculations for beam center. Problems 2 through 5 all involve a satellite and earth stations with the same specifications. Five earth stations share one transponder of a 6/4 GHz satellite. The satellite and earth station chara teristics are given below: Satellite ‘Transponder BW ‘Transponder gain Input noise temp, Saturated output power ~ 20 W (max) 4-GHz antenna gain 6-GHz antenna gain Earth station 4-GHz antenna gain 6-GHz antenna gain Receive system temp. = 36 MHz Path loss At 4 GHz, Ly At 6 GHz, L, 2, The stations all operate in a TDMA mode. Speech signals are sampled at 8 kHz, using 8 bits/sample. The sampled signals (PCM) are then multiplexed into 40-Mbps streams at each station, using QPSK,268 CHAPTERS MULTIPLE ACCESS 2. Find the bitrate for each PCM signal ». The number of speech signals (as PCM) that could be sent by each earth station, asa single access, with no overhead (i.e. no header or CRC, ete.). Ths is a TDM data stream. <. The shortest frame time for any TDMA scheme. 3 Assume that the TDMA system uses a 125-us frame time. Find the number of channels that each earth tation can send within the TDMA frame when: 1, No time is lost in overheads, preambles, and the like. ». AS-us preamble isadded to the beginning of each ‘earth station's transmission. ¢. AS-us preamble is added to each station's trans- mission and 2-us guard band is allowed between every transmission. 4, A 750-4s frame time is used instead of a 125-us frame, Find the new channel capacities of the earth stations for the cases in Problem 3 above. 5. Find the carth station transmitter power and re- ‘ceived (C/N) when the system is operated: In TDMA with the transponder saturated by each earth station in turn b. In FDMA with 3-dB input and output backoft. 6. A digital communication system uses a satellite transponder with a bandwidth of 54 MHz. Several carth stations share the transponder using QPSK ‘modulation using either FDMA or TDMA. Standard ‘message data rates used inthe system are 80 kbps and 20 Mbps. The transmitters and receivers in the sys- tem all use ideal RRC filters with a = 0.25, and FDMA channels in the satellite are separated by 100- kHz guard bands. When TDMA is used, the TDMA frame is 125 xs in length, and a 2-us guard time is required between eachaccess. A preamble of 148 bits ‘must be sent by each earth station at the start of each transmitted data burst. a. What is the symbol rate forthe 80-kbps and 2.0- Mbps QPSK signals sent using FDMA? b, What is the symbol rate of each earth station's transmitted data burst when TDMA is used? ¢. Calculate the number of earth stations that can be served by the transponder when 80-Kbps channels are sent using (i) FDMA and (ii) TDMA. 4, Calculate the number of earth stations that can be served by the transponder when 2.0-Mbps channels are sent using (i) FDMA and (i) TDMA. 7. The capacity of the TDMA system described in Problem 6 can be increased substantially by using satellite switched TDMA. In ageoup of earth stations, ‘each station sends a 2.0-Mbps signal to every other earth station in every frame, It takes 1 118 10 repos. tion the satellite antenna beam from one earth station, to another. Only the downlink antenna beam ig switched; the uplink uses a common zone beam, The frame length to be used is 1000 ys, with a 148.bi, preamble and 2-us guard times between transmis. sions arriving at the satellite The extra antenna gain atthe satellite is traded for an increase in the data rate by using 16-QAM on the downlink, Other parame. ters of the system are unchanged. ‘a, Find the number of earth stations that can share the transponder. b. Find the total data throughput of the transponder after all preamble bits have been removed. 8. An LEO satellite system transmits compressed digital voice signals to handheld terminals (sat. phones). The satphones work in groups of 10. The inbound bit stream from the satphone to the satellite is at 10 kbps. The data are sent as a BPSK signal, ‘The outbound bit stream from the satelite is ata bit rate of 100 kbps, and consists of packets addressed to each of 10 satphones. This signal is sent using QPSK, and all 10 satphones receive the 100 kbps bit stream, ‘The system operates in L band where rain fading can be ignored, but blockage from buildings and trees is a significant factor. The satellite uses onboard pro- cessing and multibeam antennas. The links use square root raised cosine (RRC) filters with a = 0.5. In this question we will be concemed only with the links between the satelite and the satphones, and ideal RRC filters will be assumed. a, Whatis the noise bandwidth of the narrowest band- iter in: (i) the satphone receiver and (ji) the receiver for the inbound link? b. What is the occupied RF bastdwidth of the radio, signals of: (i) the inbound link (phone to satellite) and Gi) the outbound link (satellite to phone)? . The inbound link has clear air (C/N)p = 18.0 4B and the BPSK demodulator on the satellite has an im- plementation margin of 0.5 dB. What is the clear air BER in the baseband of the satellite receiver? 4, What is the available fade margin {for (C/No on the uplink to the satellite ifthe inbound link operat- ing threshold is set at BER = 10“? €. The outbound fink has clear air (C/N}p = 18.0dB and the QPSK demodulator in the satellite phone has animplementation margin of 08 4B. What isthe cleat air BER? £, What is the available fade margin [for the over- all (C/N)p on the downlink to the satphone] if the outbound link operating threshold is set at BER = 10°57PROBLEMS 269 avband satelite broadcasts digital television The stations share the transponder using FDMA, sriver the United States. The nominal bit ate with SI-KHz guard bands between the edges of the Pe signi 28 Mops The gta signa can con- RF sina ‘The RRC filters used in the VSAT trans- mo prrecorded NTSC video signals, QPSK _ miter and the hub sation resis hve roll Pes cused, and error mitigation technigues factor « = 0.4. To minimize intermodulation between modal _metmployed that provide an effective coding gxin of signals, the ransponders operated with 3-dB output Sip [Coding gin of 6 aB means that when the back off ete etthe received signal is X 4B, the BER 4, Calculate the RF bandwidth occupied by each gorresponds to C/N = (X + 6) 4B.) VSAT transmission. The OPSK demodulator in the receiver hasan iM 4, Clealate th tn vs serene 6d Tere ee re pm ea ‘Feivers use ideal RRC filters with a = 025. doris bandwidth limited oa the occupied bandwidth of the RF TV a Er ialece tal ¢, Calculate the clear air C/N ratio or a received sig- | seu? nal atthe hub station, and the link margin, if the nam- mpressed tb, What is the symbol rate of the transmitted QPSK her of VSAT stations in the network is increased (0 me Ml ad the noise bandwith ofthe eats terminal jhe number you ealeulated in) above. Remenber als (at i z 110. The reriver? ‘that the power available from the transponder is fixed. sal Tre minimum permited BER after errr miee- Aging mre stations to he nworkowersthe power onthe recive 10 Whats the misimem pt per chanel at the enponder oat mitted (C/N)p for the digital TV receiver? Pat 3 rt 4. The Ka-band link suffers rain attenuation that re- ee (C/No in the receiver by 7 dB for 0.1% of the The VSAT network described in Part) i modified to be operate with TDMA on the VSAT uplinks instead year. Ifthe BER is 10° under the 0.1% year condi- tions, what is the clear air (C/N) vale? of FDMA. There are 100 VSAT stations in the net- work. | _e. Anew coding algorithms developed that provides «Fung gain of 7€B witha bitrate hat increases to The TDMA ‘ame has 8 duration of 2 ms and is So'Mps Assuming thatthe RRC filters inthe sys- Made up of 100 buts from the 100 VSAT stations. se square tem can be changed to match the new symbol rate, ‘There is a preamble of 100 symbols at the start of 5. In this tds implementation ofthe new coding algorithm im- cach VSAT station bersh and each burst is separated the links prove the system performance? If so, what is the new from the next burst by a guard time of 1.0 ys (C/N)q margin? a. There are 100 VSAT station RF bursts in each 40. This problem is about multiple access techniques frame of 2.0 ms, and 100 guard times of 1.0 j1s. What is the duration of each station's burst? in the inbound link of @ VSAT network. This set of questions compares the operation ofa Ku-band satel-__, Each VSAT station must deliver 128 Kops of 0 etaneponderin FDMA. in TDMA, and in FDMA- in the form of 128 k symbols, every second, Hove RA. There are three parts to the problem. ‘many data symbols are there in each RF burst, and Part 1 what is the total number of symbols per burst after tn a sur vetwor "accounting forthe 100 symbol preamble at the be- pone earees es yee ginning of each burst? Hence find the burst rate for eee ena jansmitter with an output power of ‘N® VSAT wansmistions symbols per second. LW and an EIRP of 41 dBW froma 1.1-m diame- ¢ Hfall the VSAT stations, and the hub receiver, have roan The aramid data signals have a bit RRC filters wit roll-off factor a = O4, wat is the rate of 128 kbps and are transmitted using QPSK RF bandwidth occupied in the transponder’ nodulation and half rate FEC, giving a symbol rte {the symbol rate ‘of transmissions were increased ar 128 keps, At the bub station the overall C/N ra unit all $4-Mliz bandwidth of the transponder were Pa algae TE EPO EY elas tania of VSAT stations in the network? 16 dB in clear air. ‘The (C/N) ratio for one channel in satelite d. The transponder can be operated with 4B out- transponder is 19.0 dB, and the (C/N)uq aio for one put backoff when TDMA is used. and implemen- tation margin of the hub receiver is 1.5 dB. The EIRP ‘channel in the hub receiver is 19.0 dB. The threshold C/N ratio in any hub station receiver for BER = 10°% of the VSAT stations must be increased because the noise bandwidth of the hub receiver has increased. By is 9.0 dB. This includes the receiver implementation margin of 05 dB. comparing the symbol rate with 100 FDMA VSAT yA in Fading and trees oard pro-270 CHAPTER 6 MULTIPLE ACCESS stations in Problem | with the TDMA symbol rate for 100 VSAT stations in part (b) above, estimate the decibel increase in EIRP required from each VSAT transmitter, ‘Comment on the feasibility of transmitting this power level from a VSAT station. Part 3 ‘The FDMA system described in Part 1 is used with random access to serve a very large number of VSAT stations. All the parameters of Part | are the same, ‘except that each station has a small amount of data to send at varying intervals of time. The average mes- sage data rate for each VSAT station is 5.0 kbps and the maximum permitted channel loading is 12%. a. How many VSAT stations can share each RF frequency? '. What is the maximum number of VSAT stations in the network when the number of RF channels is the value you calculated in Part 1(c)? 11. This problem examines the use of a Ka-band satel- lite to provide connection to the Internet from a small two-way terminal, The problem is in three parts. The first part establishes the design of the communications links and terminals, The second part examines the ca- pacity of the satellite, The third part looks at changes that must be made to support portable terminals, Part 1 Communication links Description of the satellite communication system ‘A Ka-band GEO satellite is located at longitude 100° W. Star networks can be built with a single hub station, two transponders on the GEO satellite, and a number of earth stations, identified here as VSATS "The major parameters of the system components are given below. You may not need all of these parame- ters to answer the questions, and additional parame- {ers are given in the individual questions. ‘The Kacband satellite serves the United States Coverage of the 48 contiguous states is achieved by ‘regional beam, but the satellite also carries an ad- vanced antenna system with satellite switched spot beams that allow data packets to be transmitted t0 small earth stations with a high EIRP. This allows high-speed data transmission in the outbound link. ‘The system is designed primarily to support In- ternet access via satelite, with highly asymmetrical links. Requests for access to the Internet are made by users at a low data rate through the satellite's region bbeam, Replies from the Internet can be received at a high data rate using the satellite's spot beam, System values Uplink frequency for transponder | 28.2 GHz Downlink frequency for transponder 1 21.7 GHz Uplink frequency for transponder 2 28.1 GHz Downlink frequency for transponder 2.21.6 GH, Range to satelite (all stations) 38,000 km Satellite transponders Saturated output power sow ‘Transponder bandwidth S4Mie ‘Transponder input noise temperature S00 K ‘Antenna gain, on axis, regional beam 33 dB ‘Antenna gain, on axis, switched spot beam 48 4B VSAT station parameters ‘Transmitter output power Low ‘Transmit frequency 282GHe Receive frequency 217 ‘Antenna diameter 05m Aperture efficiency 65% Receiver system noise temperature (clear air) 250K Receiver system noise bandwidth TBD Hub station parameters Maximum transmit power 100 w ‘Transmit frequency 28.1 GHz Receive frequency 21.6 GHz Receiver system noise temperature (clear air) 250K Antenna diameter 40 m Aperture efficiency 65% Receiver system noise bandwidth TBD Atmospheric losses and miscellaneous losses In clear air at 28 GHz. 2.0 6B In clear air at 21 GHz 2.0 6B Constants: Boltzmann's constant, k, = 1.38 X 10°” JK = ~228.6 dBWIKIH Part 1 Problems C/N ratios in clear air conditions ‘Make all calculations for the worst case of a VSAT station that is located on the ~3 dB contour of the satellite antenna beam (regional or spot) and fora hub station on the ~2 dB contour of the regional beam. ‘The spot beam is used only for transmissions at 21 GHz from the satellite tothe customers’ earth stations. All other links use the satellite's regional beam. 4, Calculate the free space path loss for a 38,000-km path at 28.2 and 21.7 GHz, b. Calculate the gains of the hub and VSAT antennas at frequencies of 28.2 and 21.7 GHz. tion sat sat the Ps ce in_¢,Caleuate the inbound overall C/N in the hub sta fion receiver in a noise bandwidth of 128 kHz when Ie VSAT has 2 transmitter output power of 1 W and the regional beam on the satellite. Make the " jperall C/N calculation for a single QPSK signal which is transmitted by transponder | at an output power of | W. 4 Calculate the outbound overall C/N in transponder with hub station transmit power of 1 W. Make your ‘calculation in a receiver noise bandwidth of | MHz, for a single QPSK signal, with the output power of iransponder 2 set at |W and the spot beam of the satellite transmitting to the customers’ terminals. Estimate the beamwidth of the spot beam from the ~ satellite. Using a map of the United States, estimate the minimum number of spot beam positions required to serve the entire United States. Part 2 System performance ‘Connection to the Internet is achieved by the follow: ing procedure. ‘The customer sends a connection request, in the form of a data packet, to the hub station via the satel- lite and its regional beam. The hub station decodes the request and notes the location of the station. The connection between the Internet and the hub is ¢s- tablished, through an Internet Service Provider (ISP) ‘and the public switched telephone network (PSTN). ‘Arresponse from the ISP is sent back to the customer using the satellite's spot beam. Since the packet from the customer contains the VSAT station location, the hub station can send instructions to the satellite to point the spot beam in the correct direction when transmitting packets to the customer. Note that with ‘linear transponder (bent pipe) on the satellite, the ‘beam pointing instructions must be sent to the satel lite atthe same time as the data packet. ‘The links between the ISP and the customer inthis system are highly asymmetric. The customer can send only short requests at a low data rate. The ISP can dump data to the customer at a high datarate, mainly because of the high EIRP of the satellite's spot beam transmissions. This mode of operation suits applica tions where the customer is browsing the Internet for information, or is requesting large files or video frames from the Internet. It works less well for send ing files from the customer to the Internet—asis done with e-mail, for example. In this problem you are ‘asked to design a VSAT network based on your re- sults from Part 1. Ka-band links are subject to high attenuation in rain, The outbound link is required to achieve «99.9% availability for a typical VSAT station for which slant path attenuation exceeds 7 dB at 21.7 GHz and 12 4B 3128.2 GHz, for 0.1% of anaverage year. The inbound ESCO TTC RE prosiems 271 link is required to achieve « 99.7% availability for a \ypical VSAT station for which slant path attenuation exceeds 4 dB at 21.7 GHz and 7 dB at 28.2 GHz, for (0.3% of an average year. ‘The link is declared unavailable if the BER ex- ceeds 10~* in the data stream supplied to the cus- tomer, or output by the hub station. Begin your analysis by assuming that 20 active SAT stations share the output power of transponder 1 equally at all times using QPSK-SCPC-FDMA. Half rate FEC coding is used in the inbound and the ‘outbound link and provides a coding gain of 5 dB at a BER of 10° in the recovered data stream. The im- plementation margin of the QPSK demodulators in the hub receiver is 0.5 dB, and in the VSAT receiver the implementation margin is 08 dB, Assume that there are always 20 active VSAT stations receiving, data from the outbound link in packet form, using TDM and a single QPSK carrier, Assume linear operation of the transponders, but include the effect of increased sky noise when rain is present on the uplink. ‘Transponder 1 (inbound, SCPC-FDMA) is oper- ated with 2-4B output backoff. ‘Transponder 2 (outbound, TDM) is operated with 1-4B back off. Part 2 Problems fa. Determine the clear air overall C/N required on the inbound uplink and downlink required for one ‘VSAP transmission to meet the 99,7% availabilty cri- terion, and the corresponding clear air C/N in the hub station receiver with (rain in the inbound uplink (i) rain in the inbound downlink, Remember to in- clude the effect of increased sky noise. b. Using the results you obtained in Part 1, and Part 2 problem (2), determine the maximum data rate for the VSAT request packets to meet the 99.7% avail- ability criterion with access to the transponder through the satellite's regional beam, with 20 active VSATs al any time, c. Determine the clear air overall C/N in the VSAT station receiver for an outbound data rate of | Mbps using QPSK-TDM to meet the 999% availablity criterion, for (iii) rain in the outbound uplink (iv) rain in the outbound downlink. Remember to in clude the effect of increased sky noise 4. Using the results you obtained in Part}, determine the maximum data rate that can be supplied to each ‘VSAT station with 20 active stations in the network at the same time, for the 99.9% availability criterion.272 cHaPTER.S MULTIPLE ACCESS Note that for the small percentages of time used here, you may assume that rain never occurs simultane- ‘ously in both the uplink and downlink, €. Ifyourresults from parts (b} and (dl) above show that either transponder 1 or 2 is not bandwidth limited, itis possible to optimize the system to transmit at higher bit rates. Redesign the VSAT and hub stations to increase the bit rates in either the inbound link, the outbound link, or both links, within the limits that the VSAT antenna diameter cannot exceed | m, and the transmit power cannot exceed 2 W. The hub station antenna diameter cannot exceed 5 m and the transmit power cannot exceed 200 W. You might also consider whether the number of simultaneous users can be increased, The satelite is leased and cannot be changed, except that the gain of the transponders can be adjusted to suit the earth stations used in the network. Part 3 Portable terminals The one advantage of radio systems over wired com- munications links is portability. This question asks you to design a portable Ka-band terminal which can bbe used to connect (o the Intemet (provided the cus- tomer has a clear view of the southern sky). The crit- ical clement in a portable communications link is the antenna. A large antenna provides a high data rate, but is cumbersome and must be pointed accurately at Ka-band frequencies. A small antenna is easier to set up, but cannot provide a high data rate. Let’s assume that the dimensions of the antenna are limited to the dimensions of a typical laptop computer—0.25 m X 0.2 m—with an aperture efficiency of 25%, and that some method is provided that helps the customer point the antenna beam toward the satelite so that there is no more than 1-dB loss of gain due to an- {enna mispointing. Because the portable terminals cannot achieve the same C/N ratios as the fixed terminals, separate ‘wansponders will be needed to service the portables. For convenience, we will call these transponders 3 Gnbound) and 4 (outbound) and use the same fre- quencies as transponders 1 and 2. The ability of the system to operate during rain fades on the outbound link is relaxed with an availability of 99.7% required in each direction, a. Calculate the gain and the beamwidth of the portable antenna at frequencies of 28.2 and 21.7 GHz. . Using your results from Part 1, find the inbound and outbound overall C/N ratios in the hub station and portable receivers using the conditions in Par in clear air conditions. Don’t forget to allow an ext 1 4B of loss to account for antenna mispointing ©. Assume that 10 active stationsshare each transpon, der, Determine the maximum data rates that customer, ‘can achieve on the inbound and the outbound links with 99.7% availablity of the inbound and outbound links, 4, Transponders 3 and 4 can be switched into base. band processing mode. In this mode, the incoming QPSK signal is demodulated to baseband, the data bits are recovered and then remodulated onto a car. rier for transmission as a new QPSK signal. This allows the transponder to transmit at its rated out. PUL power at all times despite uplink attenuation ‘The bit error rate for the link is then the sum of the BERs on the uplink and the downlink. Rework your solution to part (c) above using baseband proces. sors for both inbound and outbound links and de- termine the new data rates for the inbound and out. bound links. ¢. Draw ablock diagram of transponder 3 when used in its baseband processing mode. Your block diagram should include all the filters, amplifiers, mixers, oscillators, modulators and demodulators, and all ‘ther important blocks. Label each filler and ampli- fier with a center frequency and bandwidth, and in- dicate the gain of cach amplifier. Label all oscillators with their frequencies. f. Comment on the performance of the fixed and portable Ka-band Internet link system. If the transponders on the GEO satellite cost $1.5 M per year each to lease, and the service provider's costs to Support the customer base that shares these transpon- ders are $ 05 M per year, what would you expect to hhave to charge the customer for access to the Inter~ ret when using the fixed terminal and the portable terminal? You can establish a charging structure made up of a monthly fee plus a per minute access charge. ‘Assume that you can achieve a continuous level of activity of 20 fixed or 10 portable terminals for 12 h per da Each user can be assumed to connect to the In- ternet for 15 min once each day, bat is active (in the sense of data transfer over the satellite) for I min per day. How do the data rates and the charges you pro- pose for the portable Internet access service compare to typical charges for cable modem service?