Professional Documents
Culture Documents
Measurements MHZ: Outdoor Microcell AT 1700
Measurements MHZ: Outdoor Microcell AT 1700
927
0-7803-0673-2192$3.00 1992 IEEE
Authorized licensed use limited to: Purdue University. Downloaded on June 07,2010 at 04:45:00 UTC from IEEE Xplore. Restrictions apply.
would be insufficient in this case, since the surround- Parameters for the different models have also been
ing buildings contribute significantly to the loss. A dual calculated using measured data as input. The param-
slope path loss model for 1700 MBr has b a n proposed eters were derived by using the Nelder-Mead Simplex
in [l]. In this model the LOS loss is described as in method [2] with the square error between measured
equation 1 which is the same as free space loss at 1700 and d e a l a t e d d s t r aa the function to minimize. Some
MHz if M = 37 dB and n = 2, z is the distance between restrictions were made on the parameters to assure con-
transmitter and receiver. These parameters have also vergence, for example no negative values were allowed.
been used in [l] for the first part of the LOS loss.
Loss(2) = M - 2" (1) 5 Line-of-sight results
However, measurements show that in some cases the
loss does not follow a single slope but has an increased The parameters for the LOS breakpoint model, table
loss factor after a certain distance. To be able to model 1, are based on the Stockholm measurements since this
this, an additional losa term is added after a break- is where data from distances larger than 100 m exist.
point. Ignoring the constant factor M,the loss is now The results show that the loss is slightly larger than
as in equation 2 before and as equation 3 after the free space (n=2) and that the breakpoint is unlikely to
breakpoint X L occur at distancea shorter than 150 m. There is also a
small positive correlation between 712 and c~ i. e. the
LOSSLl(2) = znl (2) further away the breakpoint is, the larger the value of
log(&) - lOg(Y1)
0
-5
where U(.) is the step function and -10
-15
f(Y) = 10 * log (E + 1)" [dB] -20
-25
y I z,the travelled direction in LOS. -30
The model is based on the observation that there is
a short distance ( 9 1 ) after the corner where the signal
-35
level still is of the same magnitude as in LOS, further -40 1
down the street there is a rapid increase in the loss
until the signal finally (&)converge to an n:th order
power law, equation 6, where m is the point where the
asymptote of f(z) = 0 dB. Figure 1 Measured data together with loss equations
928
Authorized licensed use limited to: Purdue University. Downloaded on June 07,2010 at 04:45:00 UTC from IEEE Xplore. Restrictions apply.
As shown in figure 1this will yield an additional error 7.1 Distance from wall
close to y1, on the other hand fewer parameters are
Fitting data to equation 7 gives the result in table 3.
needed. Using only equation 6 will give a small increase
to the squared error between data and calculated levels
as most of the data points are from distances larger
Level(z) = C + 10 . n . log(z) [dB] (7)
than yz meters; the same is valid for the 4-parameter
NLOS models in [l]. The fact that the levels from all Tx Distance from n C
the three tracks in figure 2 converge implies that there wall /m
must be some kind of common contributor t o the signal
level, regardless of track. This contributor may be the
wave diffracted a t corner Q3 [4], the diffraction term
from corner Q1 is, as proposed in [3], too small to be
of any influence.
3 1 0.5 -1.65
-1.78
-19.9
-17.54
1
Table 3 Variation of parameters with distance between
N, Tx Tz-QI/m n zo/m zllm xzlm receiver and wall W1.
1 1 62 1.69 0.80 1.02 1.06
1 2 62 2.02 0.67 0.70 1.16 In LOS the distance to the closest reflecting wall
2 I 1 I 62 1.8 0.96 1.20 I 2.71 seems to show low correlation with the regression pa-
I I I
I 21 21 62 I 2.15 I 0.97 I 0.97 I 1.03 I rameters according to these measurements.
3 1 62 2.11 1.81 1.83 10.6 In NLOS it becomes more important with the dis-
3 2 62 1.94 0.75 0.93 7.0 tance to the wall closest to the transmitter, wall W1
6 1 120 1.56 0.50 0.50 0.51 in figure 2, as this will determine how far down in the
6 2 120 2.3 1.73 1.74 1.78 crossing street reflexes from the LOS street walls will
8 1 150 2.02 3.42 3.44 74 reach. When turning around the corner close to the
8 2 150 1.8 3.33 3.57 58.8
Tx 2 0
~4 I
1
929
Authorized licensed use limited to: Purdue University. Downloaded on June 07,2010 at 04:45:00 UTC from IEEE Xplore. Restrictions apply.
very small and the result will be a diffuse wave that are very small and irregular, the only parameter that
may contribute to the fading but not to the mean path changes proportionally with the height is the break-
loss. The received power in three different NLOS paths, point z ~ The. parameters are calculated from data
1, 2,and 3 in 2 were measured with one transmitter on with distance between antenna and receiver larger than
each side of the LOS street. The results from transmit- 30 m. At shorter distances between transmitter and
ter 1 are shown in figure 3. There is an increase both of receiver the elevation of the main lobe of the antenna
y1, the point where the power starts to decrease, and becomes significant and here the antennas height to-
y2, where the curve converge towards the level in track gether with its vertical radiation pattern will determine
I. the path loss. This can however be used to reduce the
relative level/dB required dynamics of the radio equipment as omnidi-
rectional antennas, as the ones used in these measure
I I I
ments, often have a lower gain below the ground plane;
0 in this particular case about -20 dBi.
-5
7.3 Distance between corner &I and
-10
transmitter in NLOS
-15 An increasing distance between corner and transmit-
-20 ter should have a similar effect on the NLOS loss as
decreasing the distance to W1 in figure 2. Two mea-
-25 surements, number 2 and 6 in table 2 can be directly
-30 compared as they are measured in the same track but
with antennas a t different distance from the corner.
I I Here it can be noted that ~ 0 , y land do not only con-
0.1 1 10 verge when the antenna distance is increased but may
distancelm also decrease, as when transmitter 1 was used. This
Figure 3 Measured power in track 1,2 and 3 relative to
supports the theory that the reflections are determin-
the power between corners &I to 9 2 ing the level close to the corner and some other term
is determining the level further down the street.
Table 4 Parameters for the NLOS model eztracted from and NLOS. The measurements have shown that LOS
measurements in three trucks. path loss can be described as approximately free-space
loss and is fairly insensitive to placing in the street as
well as antenna height of the transmitter. For NLOS
7.2 Antenna height loss the placing in the street, the distances between
transmitter and corner as well as between corner and
The parameters in table 5 are derived for the LOS receiver are of great importance. A promising model
model, equations 2 to 4. for NLOS path loss has been proposed that takes these
I Antenna I nl I nz I 21, I parameters into consideration. It is based on measured
height/m m results that show that the additional loas does not occur
3.0 2.5 5.4 223 immediately as LOS is lost but some distance further
4.0 2.3 5.1 194 down the street. This distance is determined by sur-
I 1
I
I
5.0 I 2.4 1 5.4
I
I 191 I
ro'unding obstacles and transmitter placing and should
therefore be possible to calculate. After this there is
Table 5 Variation of parameter values for the dual slope an immediate loss that gradually converges towards an
LOS model with antenna height. exponential loss with an exponent close to two and a
reference point close to the point were the direct path
The LOS measurements were done in the same street was lost.
with the antennas at different height. The differences
930
,
Authorized licensed use limited to: Purdue University. Downloaded on June 07,2010 at 04:45:00 UTC from IEEE Xplore. Restrictions apply.
Acknowledgements References
The authors would like t o thank Professor Sven-Olof [l] Jan-Erik Berg.
Ohrvik a t our department for initiating this work, Mr Path loss and fading in microcells.
Per-Owe Klingvall and Mrs Katarina Borjeson for as- In TD(90) 065. COST, October 1990.
sistance during the measurement campaigns, Ericsson [2] J . A. Nelder and R. Mead.
Radio Systems for giving us radio equipment and Mr A simplex method for function minimization.
Jan-Erik Berg whose ideas and suggestions have been Computer Journal, (7):308-313, 1965.
most valuable. [3] Y. Nagata, Y. Furuya, E. Moriyama, M Mizuno,
This work was supported by the National Swedish I. Kamiya, and S. Hattori.
Board for Technical Development under contract no Microcellular propagation measurements in an ur-
88-01-357P. ban environment.
In International symposium Personal, Indoor and
Mobile Radio Communication, pages 341-346,
Kings College London, UK, September 1991.
IEEE.
[4] C. Bergljung and L. G. Olsson.
Rigorous diffraction theory applied to street micro-
cell propagation.
In Proc. IEEE Global Telecommunication Confer-
ence, 1991.
[5] J . H . Whitteker.
Measurement of pathloss at 910 mhz for proposed
microcell urban mobile systems.
IEEE Transactions on Vehicular Technology, pages
125-129, August 1988.
931
Authorized licensed use limited to: Purdue University. Downloaded on June 07,2010 at 04:45:00 UTC from IEEE Xplore. Restrictions apply.