Broadband Communication

Systems

521316S
 Exercise 1
• Determine parameters for OFDM system operating
under the following conditions:
• bit rate Rb 20 Mbps
• tolerable rms delay spread  200 ns
• System bandwidth B  15 MHz
• Loss due to cyclic prefix max. 1 dB

2
 Example Design for OFDM (1)
• Let’s choose Tcp=800 ns*), to allow for timing errors<~0.6
s
• In order to have loss due to cyclic prefix less than 1 dB:
 T  Tcp   T  Tcp 
 10 log10    Ploss  log10    Ploss
 T   T 
   


 T 1  10 Ploss / 10
T
cp  T 
Tcp
1  10Ploss / 10
0.8s
Now, T  1/ 10
 3.9s
1  10

• Let’s choose T = 4.8 s => SNRloss=0.8 dB

*)Cp length should be 2-4 times max. delay spread,
3
depending on the data mod. and ch. coding robustness against ICI.
 Example Design for OFDM (2)
• Then subcarrier spacing becomes Df=1/T = 250 kHz
S

• Number of subcarriers becomes:

– N=B/Df = 15 MHz/250 kHz = 60
• IFFT/FFT of size 64 should be chosen
• To achieve 20 Mbps, each transmitted OFDM
symbol (including cp) must carry 96 bits of
information (96/4.8s = 20Mbps)

4
 Example Design for OFDM (3)
• 16-QAM with rate 1/2 coding gives
4*1/2*N/4.8s= 20 Mbps => N=48 subcarriers for
data
• QPSK with rate 3/4 coding gives
2*3/4*N/4.8s= 20 Mbps => N=64 subcarriers for
data
• However, 64 subcarriers would mean bandwidth of
64*250kHz=16MHz
• The first option is selected to maintain under
15MHz bandwidth (48*250kHz=12MHz).
5
 Example Design for OFDM (4)
• The receiver operates by using samples. Hence an integer
number of samples must be collected both from FFT interval
and OFDM symbol interval:
64-FFT => sampling freq. 64*250kHz=16MHz, BUT
16MHz*0.8s=12.8 samples for cp

• The parameters need to be readjusted to meet this

requirement.

6
 Example Design for OFDM (5)
Sampling rate is Rs  N FFT / TFFT s , then it must also be Rs  N cp / Tcp
N FFT N cp N FFT
   N cp   Tcp .
TFFT Tcp TFFT

To make sure that Ploss  1dB,

N 64
N cp  FFT  Tcp   0,8s  16.5.
TFFT 3.1s
TFFT=TS
Let's select N cp  13. Now the modified FFT integration
interval is TFFT  3.938s (resulting in Ploss  0.99dB)
and the new sampling frequency is Rs  13 / 0.8s  16.25MHz 7
 Example Design for OFDM (6)
• The bandwidth constraint needs to be rechecked
since the carrier spacing is slightly modified:
Df=253.90625kHz => (16-QAM needs 48 carriers
for data) B=12.1875MHz.

• Data rate requirement is achieved since OFDM

symbol interval is now a bit shorter.

8
 Example Design for OFDM (7)
• How the system parameters would be changed if the channel delay
spread is 1s and the number of subcarriers is kept the same?
• Let’s assume that 0.6 s is enough to cover timing errors etc. =>
Tcp=1.6s

1.6s
T 1/ 10
 7.8s
1  10

• Let’s choose T = 8.4 s => SNRloss= 0.9dB

• Then subcarrier spacing becomes Df=1/TS = 147 kHz

9
 Example Design for OFDM (8)
• 48 subcarriers results in 48* 147 kHz = 7 MHz system
bandwidth.

• 16-QAM with rate 1/2 coding gives

4*1/2*48/8.4s= 11.4 Mbps

• QPSK with rate 5/6 coding gives

2*5/6*48/8.4s= 9.5 Mbps

• The larger delay spread causes lowering the data rate and
lowering the system bandwidth if the number of subcarriers
is kept constant.

10
 Example Design for OFDM (9)
• How the data rate of 20Mbps can be maintained with system
bandwidth of 15 Mbps for delayspread of max. 1 s?

• Tcp=1.6s and TS = 6.8 s

• The subcarrier spacing is Df=1/TS ~ 147 kHz
• Number of subcarriers is N=B/Df = 15 MHz/147 kHz = 102
• IFFT/FFT of size 128 should be chosen.
• 16-QAM with rate 1/2 coding gives
4*1/2*102/8.4s=24.3Mbps
• QPSK with rate 5/6 coding gives 2*5/6*102/8.4s= 20.2
Mbps
When higher delayspreads must be tolerated, OFDM symbol length
must be increased to avoid large performance loss due to cyclic prefix.
This in turn results in narrower subcarrier spacing => synchronisation problems
11
and larger number of subcarriers => increased complexity due to larger FFT
 Discussion on the Choice of
Parameters
• The guard interval often isn't negligible compared to the
OFDM data symbol length (often, it's 1/4th of the useful
symbol size). Why not use a very long OFDM data symbol
after a guard interval in order to decrease the redundancy
(i.e. to minimise the loss due to cyclic prefic) ?

– Subcarrier spacing is inverse of the OFDM symbol length

– Subcarriers would be more closely spaced to keep
bandwith constant
=> tighter frequency and phase synchronisation
requirements

12
 Discussion on the Choice of
Parameters (2)
• If we define an OFDM system for a quasi-AWGN-channel
context (i.e. channel impulse response is short) - so, the
data throughput can be increased by choosing a short guard
interval.

– Long enough cyclic prefix relaxes timing requirements

– TX and RX filters also cause extra delay, i.e., lengthening
of channel impulse response

 Longer cyclic prefix makes system implementation easier

• CP >> channel delay spread

13
 Exercise 2
• Consider a multicarrier (MC) system, which has a total
passband bandwidth of 1 MHz. Assume that the channel
delay spread Tm = 20 µs. How many subchannels are
needed, so that each subchannel can be expected to be
approximately flat fading?

14
 Exercise 2 solution
• The coherence bandwidth of the channel Bc can be
calculated from the channel delay spread Tm,

• To ensure that each subchannel has flat fading, the

bandwidth is limited to BN = 0,1 Bc << Bc. Thus the number
of subchannels is

• Keeping in mind that N must be a power of 2 for the sake of

FFT and IFFT, it must be rounded upwards to the closest
suitable number. Thus the minimum value for N is 256. 15
 Exercise 3
• Consider a multicarrier FDM system (without and with
overlapping) operating in a channel with coherence
bandwidth Bc = 10 kHz.
a) Find a subchannel symbol time TN = 1/BN = 10Tm,
assuming channel delay spread Tm = 1/Bc. This should
insure flat fading on the subchannels.
b) Assume the system has N = 128 subchannels. If raised
cosine pulses with β = 1.5 are used, and the required
additional bandwidth due to time limiting to insure
minimal power outside the signal bandwidth is ε = 0.1,
what is the total bandwidth of the system?
c) Find the total required bandwidth of the system using
overlapping carriers separated by 1/TN, and compare
with your answer in part b).

16
 Exercise 3 solution a)
• The subchannel symbol duration is

17
 Exercise 3 solution b)
• If we assume raised cosine pulse, we get a symbol time

for each subchannel, where β is the rolloff factor of the pulse

shape.
• In a realistic implementation, the subchannels occupy a
larger bandwidth than under ideal raised cosine pulse
shaping due the time limiting of these pulse shapes. Let ε/TN
denote the additional bandwidth required due to time-
limiting of these pulse shapes. Hence the subchannels must
be separated by

18
 Exercise 3 solution b)
• The entire occupied bandwidth for non-overlapping
subchannels is then given by

19
 Exercise 3 solution c)
• This time β and ε do not affect the total system bandwidth
due to the subchannels overlap except in the first and last
subchannels. The total system bandwidth with overlapping
subchannels is given by

• where the approximation holds for N large.

• Thus we get (without approximation)

20
 Exercise 3 solution c)
• The total bandwidth using overlapping carries is less than
half of the non-overlapping bandwidth.

21
 Exercise 4
• Consider a high-speed data signal with bandwidth 0.5 MHz
and a data rate of 0.5 Mbps. The signal is transmitted over a
wireless channel with a delay spread of 10 µs.
a) If multicarrier modulation with non-overlapping
subchannels is used to mitigate the effects of ISI,
approximately how many subcarriers are needed? What is
the data rate and symbol time on each subcarrier? (We do
not need to eliminate the ISI completely, so BN = Bc is
enough for ISI mitigation)
b) Assume that the average received SNR (γs) on the nth
subcarrier is 1000/n (linear units) and that each subcarrier
experiences flat Rayleigh fading (so ISI is completely
eliminated). Suppose BPSK modulation is used for each
subcarrier. If a repetition code is used across all
subcarriers (i.e. if a copy of each bit is sent over each
subcarrier), then what is the BER after majority decoding?
What is the data rate of the system? 22
 Exercise 4 solution a)
• Aim: diversity gain obtained from subcarriers. The
condition imposed for ISI mitigation in this time
[compare to BN = 0,1xBc << Bc ->TN = 10Tm] is TN
= Tm = 10 μs

• The bandwidth of a subchannel BN = 1/TN = 100

kHz. Thereby the number of subchannels is

• and each subcarrier must carry

23
 Exercise 4 solution b)
• The bit error probability of BPSK in a Rayleigh
fading channel [1],(6.58) is

• where

24
 Exercise 4 solution b)
• Thus, for the 5 subchannels, it is possible to
calculate the following BERs

• Majority decoding means that the data is detected on each

subcarrier and that the decision is done by selecting the
symbol, which is the most common amongst the detected
symbols. Since N = 5, three symbols is enough to form 25 the
majority.
 Exercise 4 solution b)
• BER after decoding can be calculated from equation

• where the first term is probability that there is an error

in every 5 subcarriers, the second term is probability
that there is an error in 4 subcarriers (one is correct),
and the third term is probability that there is an error in
3 subcarriers (two are correct). 26
 Exercise 4 solution b)
• There is noticeable gain compared to a case where
n = 1. This is known as diversity gain.
• The total data rate of the system is the same as
the data rate of any of the subcarriers (since they
all have the same bits transmitted over them), i.e.,
R = 0,1 Mbps.

27
 Exercise 5
• Consider a channel with impulse response
h(t) = α0δ(t) + α1δ(t − T1) + α2δ(t − T2).
• Assume that T1 = 10 µs and T2 = 20 µs. You want to design a
multicarrier system for the channel, with subchannel bandwidth
BN = Bc/2. If raised cosine pulses with β = 1 are used, and the
subcarriers are separated by the minimum bandwidth necessary
to remain orthogonal, then what is the total bandwidth occupied
by a multicarrier system with 8 subcarriers? Assuming a constant
SNR on each subchannel of 20 dB, find the maximum
constellation size for MQAM modulation that can be sent over
each subchannel with a target BER of 10−3, assuming M is
restricted to be a power of 2. Find also the corresponding total
data rate of the system.
• Use as the upper bound for BER of an M-ary QAM, where
γ denotes the SNR.
28
 Exercise 5 solution
• Channel impulse response is h(t) = α0δ(t) + α1δ(t − T1) +
α2δ(t − T2).
• T1 = 10 µs and T2 = 20 µs.
• Subchannel bandwidth BN = Bc/2.
• Raised cosine pulses with β = 1 are used.
• Subcarriers are separated by the minimum bandwidth
necessary to remain orthogonal.
• Constant SNR on each subchannel of 20 dB = 100.
• Target BER of 10−3.
• MQAM modulation, where M is restricted to be a power of 2.

29
 Exercise 5 solution
• What is the total occupied bandwidth, if the
multicarrier system has 8 subcarriers?
• The delay spread of the channel Tm = T2 = 20 µs. Thus the
coherence bandwidth of the channel

• and the subcarrier bandwidth (this time)

• Note that this condition is not sufficient to avoid ISI (the

proper condition would be BN << Bc
30
 Exercise 5 solution
• The minimum frequency separation which allows the
subchannels to remain orthogonal over symbol interval
[0,TN] is 1/TN. For a raised cosine pulse with β = 1, the total
occupied bandwidth with 8 subcarriers is ([1] (12.4), where
ε = 0)

31
 Exercise 5 solution
• What is the maximum constellation size for MQAM
modulation that can be sent over each subchannel
with a target BER?
• The BER of an M-ary QAM constellation is upper bounded by

• Inserting the SNR (γ = 100) and the target BER (10–3), the
appropriate modulation order can be calculated as

• The closest power of 2 is 16, and therefore the maximum

constellation size for MQAM modulation is M = 16. 32
 Exercise 5 solution
• What is corresponding total data rate of the system?

• The associated data rate is

• Note that in this instance, no CP was defined, which means

that 1/T = 1/TN = BN.

33