Solution: The signal power is 0 dBm at 1m and -90 dBm at
km. The AMPS receiver has 30 kHz bandwidth, which corresponds to
a thermal noise power of -129.2 dBm, which becomes -123.2 dBm when
the 6 dB noise figure is added. Therefore, we have for
(a) Solving for SNR(d)=15 dB yields
km.
(b) With probability 99%, the shadowing attenuation does not exceed
dB. Solving for SNR(d)=15+13.8 dB yields
km.
Solution: Wavelength at 1800 MHz is 1/6 m, and 6 km/h = 10/6 m/s,
so that
Hz. Duration of 1000-bit packets is
T=0.1 ms at the specified data rate, so that
.
This value of
and the fading margin F=20 dB correspond to
and
. The value of
can be found from
to be 1/3960, and p=1-q.
(a) The probability that the file is successfully sent is the probability that K consecutive slots are successful, where K=5000 is the total number of 1000-bit packets in the 5-Mbit file. Therefore,
since the channel state in the slot in which the first packet is sent has steady-state probability distribution. This value turns out to be about 0.28.
If level crossing rate analysis is to be used, note that the average
crossing rate of level
is
,
i.e., segments made up of one bad channel period and one good channel
period have average length
. Since
is the probability
that the channel is in good state, the average duration of a good channel
period is given by
, which in
the present case (
Hz) amounts to about 0.4 s.
Assuming that the duration of a good channel period is exponentially
distributed, and given that the time needed to transmit the file is
0.5 s, the probability that the channel remains good for at least 0.5
s given that it is good at the beginning of the transmission is
. The probability of successfully transmit
the file is then given by this number multiplied by the steady-state
probability that the channel is good,
, and is equal to
0.283, which is consistent with the result we found previously, as expected.
(b) If after a failure we wait for a long time (with respect to the channel's coherence time), successive attempts are approximately independent of each other and the number of attempts needed to succeed is a geometric random variable whose mean is the inverse of the probability of success of a single attempt, i.e., about 3.57.
(c) By repeating the discrete Markov computation doubling T and dividing K by two gives a numerical result which is very close to the one found in (a). In the level crossing approach, nothing would change. Therefore, doubling the packet size would not help.
Solution:
(a)
The number of full-duplex channels available is as follows:
AMPS: 10000/30 = 333; IS-54: 10000/30
3 = 1000;
GSM: 10000/200
8 = 400. For CDMA,
can be found as
where W=10 MHz, R=10 kbps, and K is the number of users.
In order to guarantee that
dB = 4, a maximum of
about 250 users can be simultaneously supported.
(b) In the first three cases, nothing changes, since sectorization does not increase the number of channels per cell. In CDMA, a threefold gain in capacity is achieved, and 750 users can now be simultaneously active.
(c) In the first three cases, some frequency reuse plan must be used, so that the number of channels per cell is correspondingly reduced (e.g., by a factor of 7). In CDMA, no frequency reuse plan is needed, and the effect of the multicell structure is to increase the interference by about 50%. Therefore, the number of simultaneously active users which can be supported in the CDMA system is now 500.
(d) Each sector has now 333/21 channels, i.e., some sectors will have 15 and some others will have 16. For 16 channels, GOS of 2% is achieved at 10 Erl (9 Erl for 15 channels). If we have 900 users per cell, there will be 300 users per sector sharing the 16 channels. In order for the GOS to be met, each user can generate up to 1/30 Erl which, with 2 minutes as the average call duration, corresponds to a calling rate of one call per hour.