ECE 5325/6325 Fall 2009: Exam 1 Solutions

1. (25 pts total) Question 1 was different between the 6325 and 5325 exams (but all other questions
were the same). For the other half who haven’t seen the questions, they are also included on this
problem. First, the 5325-only question 1.

5325 Question: You are called in to design a cellular system for Planarton, Utah. Unfortunately,
your path loss measurements indicate that path loss follows the log-distance model with path loss
exponent n = 2.95. You find a cellular standard that needs an S/I ratio of only 15 dB.

(a) (9 pts) What is the minimum N required, using omni-directional antennas? Solution: In dB,
we have S/I(dB) = 5(2.95) log 10 (3N ) − 10 log 10 6. Testing with various N , I found: for N = 4,
S/I(dB) = 8.14; for N = 7, S/I(dB) = 11.72; for N = 9, S/I(dB) = 13.33, and for N = 12,
S/I(dB) = 15.17. Thus N = 12 is the minimum N required.
(b) (7 pts) For your answer for (a), what are i and j? Solution: For a choice of i = 2, j = 2
provides N = i2 + ij + j 2 = 4 + 4 + 4 = 12.
(c) (9 pts) What is the minimum N required, using 120 degree sectoring? Solution: Note that
i0 = 2 for N = 4, 7, 9, 12 so I used S/I(dB) = 5(2.95) log 10 (3N ) − 10 log10 2 and tested N ≥ 4.
For N = 4, S/I(dB) = 12.91; for N = 7, S/I(dB) = 16.49. So N = 7 was the minimum.

6325 Question: S/I for a one-dimensional cellular system: Consider a cellular system deployed
along a infinitely long linear highway with all traffic coming from users in vehicles driving on the
highway, as shown in the figure. Assume BS towers serve a circular cell with radius R meters, and
so BSes are located every 2R meters along the highway. Assume that every N th BS reuses a channel
group, where N is the frequency reuse ratio. Assuming received power is proportional to 1/dn for
distance d, derive an expression for the signal to co-chanel interference (S/I) ratio for a mobile on
the edge of a base station’s coverage area, as a function of N and n.

Solution: The S/I ratio is the ratio of power from the serving BS to the total power from the
co-channel BSes. We typically use only the first tier of co-channel BSes because the further tiers
contribute much less interference power. Let received power be equal to cd−n for some constant c.
The serving BS is R away from the mobile, and the first tier BSes are R(2N + 1) and R(2N − 1)
away from the mobile as shown in the figure.
S cR−n 1
= =
I c[R(2N + 1)] + c[R(2N − 1)]
−n −n (2N + 1) + (2N − 1)−n
−n

where the cR−n multiplied all terms, so it cancelled out.

ECE 5325/6325 Fall 2009 2

2. (30 pts total) Since there are 5 forward channels per 120 kHz of spectrum, the total number of
ch = 416.7. Note that A = λH = 5 calls/hr 2 hr/call = 1/6.
forward channels is 10 MHz 0.125 MHz u 60

(a) (10 pts) For N = 12, and omni antennas, there are C = 416.7/12 = 34.7 channels per cell.
20
Using Erlang C, with P [delay] = 0.02, I read A = 20. Thus U = A/AU = 1/6 = 120 users per
cell. With 100 cells city wide, there are 12,000 total users.
(b) (10 pts) With 120 degree sectoring, N = 7 and three sectors per cell. So C = 416.7/(7 × 3) =
19.8 channels per sector. Using Erlang C, with P [delay] = 0.02, I read A = 13. Thus
13
U = 1/6 = 78 users per sector. With 100 cells city wide, there are 78(3)(100) = 23, 400 total
users.
(c) (10 pts) Using P [delay] = 0.02, C = 34.7, A = 20, H = 120 seconds, and t = 10 seconds,
   
C−A 34.7 − 20
P [delay > t] = P [delay > 0] exp − t = 0.02 exp − 10 = 0.0059 (1)
H 120
3. (30 pts total) We’re asked about the uplink, so the police radio is the TX and the BS is the RX.
(1) So Pt = 3 dBW. (2) The BS receiver has PN = F kT0 B. Since F = 103/10 = 2 for the BS
receiver, and B is 30 kHz, PN = 2(1.38 × 10−23 J/K)(294 K)(30 × 103 Hz) = 2.43 × 10−16 W, or
-156.1 dBW. (3) The gains are from the half-wave diplole (2.15 dB) plus 7.0 dBi at the BS, that
3×108
is, 9.15 dB. (4) The wavelength is λ = 705×10 6 = 0.4255 meters, and thus the path loss (free space)
4π10 m
at reference distance 10 m is Π0 = 20 log10 0.4255 m = 49.4 dB. The path loss beyond 10 meters is
Lp = 49.4 + 10(3.1) log 10 10dm . We are told to take into account the worst case 4 + 20 = 24 dB losses
caused by weather and building penetration. So the total losses are 73.4 + 31 log10 10dm . Plugging
in and solving for d,
X X
S/N (dB) = Pt (dBW) + dB Gains − dB Losses − PN (dBW)
 
d
11.8 = 3(dBW) + 9.15 − 73.4 + 31 log10 − (−156.1)(dBW)
10 m
d
31 log10 = 3 + 9.15 − 73.4 + 156.1 − 11.8
10 m
d = 10 m 1083.05/31 = 4.78 km

4. (15 pts total, or 3 pts each part)

(a) Buy more bandwidth / add more channels: True! Adding more channels, for the same traffic,
will reduce the probability of delay. Try on an Erlang C chart, keeping A constant, and
increasing C. The P [delay] will decrease.
(b) Replace more macrocells with microcells: True! There are higher number of microcells than
the number of macrocells they replace. For the same N , there is about the same C per cell,
but fewer users, and thus lower A.
(c) Increase the frequency reuse factor N : False. This is the wrong direction – increasing N
reduces C, the channels per cell, without changing A.
(d) Use umbrella cells: False. Umbrella cells are used to reduce dropped calls due to handover
errors. Umbrella cells segment the number of channels into more sub-groups, so the trunking
efficiency reduces, thus actually increasing the P [delay].
(e) Use repeaters: False. Using repeaters increases coverage in shadowed areas, but this actually
allows more users to request channels, thus P [delay] would slightly increase.