Вы находитесь на странице: 1из 34

Problem 1 a) The frequency band allocated to a certain cellular standard enables 60 traffic channels/cell.

When the whole band is licensed to one operator, what is the amount of traffic the operator can serve with 2 % blocking probability? The traffic is assumed to be Erlang-B distributed. b) Competition is introduced by sharing the frequency band between three operators, each getting 20 traffic channels/cell. What is the amount of traffic the three operators can together serve with 2% blocking? Each operator will have about the same network configuration, and subscribers can only be served by their own operator. Solution a) The Erlang B-table gives T1 = 49.64 Erlang b) The Erlang B-table gives for one operator (20 TCHs) T1 = 13.18 Erlang T3 = 313.18 = 39.54 Erlang By giving all frequency resources to one operator instead of one third to each of three operators would give 25.5 % higher total capacity.

Problem 2 a) A GSM900 operator has 36 carriers which are used with the reuse factor 9. How large traffic can the operator serve with 2 % blocking probability. The network is ideal with spatially homogeneously distributed traffic and equisized cells. b) To meet the increasing capacity demand the operator opens a GSM1800 network with 108 carriers given the same coverage using same base stations as the GSM900 network and having the same reuse factor. How large traffic the operator can now serve with 2 % blocking probability? c) Calculate the traffic capacity with 2 % blocking probability if all user terminals are dual-band able to use both frequency band. How many % larger the capacity is compared to the situation in b) where only single-band terminals are used? Solution a) Under the given conditions there are 36/9 = 4 carriers/cell. Assuming that in each cell two timeslots are used for signalling there are 48 2 = 30 traffic channels/cell. From Erlangs B-table the traffic capacity for 2% blocking probability is 21.93 Erlang/cell.

b) If it is assumed that the number of timeslots is proportional to the number of carriers, there will be 90 traffic channels/cell in the GSM1800 system. Applying again Erlangs B-table the traffic capacity in the new system is 78.31 Erlang/cell Now the total traffic capacity is 21.93 + 78.31 = 100.24 Erlang/cell. c) When all terminals are dual-band each terminal will have 120 channels available. Erlangs B-table gives then a traffic capacity of 107.40 Erlang/cell. Compared to case b) where only single-band terminals are used 7.1% higher traffic capacity is obtained

Problem 3. a) A mobile phone subscriber generates on average 20 mErlang traffic during the busy hour. (Occupies a channel for 1.2 minutes.) How many subscribers in a cell having 30 traffic channels will cause a blocking probability of i) 0.5%, ii) 5%, iii) 50%? b) How many subscribers can be served at the different blocking levels? c) What is the average channel load at the different blocking levels?

Solution a) The traffic in a cell is obtained from the Erlang B-table. The number of subscribers that can be served is obtained by dividing the cell traffic with the average subscriber traffic.

19.03 = 9515 951 . 0.02 24.80 B = 5% T = 24.80 Erlang N = = 1240.0 1240 0.02 5811 . B = 50% T = 5811Erlang N = . = 29055 2905 . 0.02 B = 0.5% T = 19.03Erlang N =

b) The portion of subscribers being blocked cannot be served. Therefore

B = 0.5% N served = (1 B ) N = (1 0.005)951.5 = 946.7 946 B = 5% N served = (1 B ) N = (1 0.05)1240 = 1178 1178 B = 50% N served = (1 B ) N = (1 0.5)2905.5 = 1452.75 1452
c) The average channel load at a given blocking level is the served traffic divided by the number of channels

B = 0.5% =

a1 BfT = a1 0.005f19.03 = 0.631 N 30 a1 BfT = a1 0.05f24.80 = 0.785 B = 5% = N 30 . a1 BfT = a1 0.5f5811 = 0.969 B = 50% =
N 30

One can see that allowing a higher blocking level more subscribers can be served and the spectral efficiency is increased. However, a high blocking level might not be a good marketing argument!

Problem 4 1 000 000 people are living uniformly distributed in the service area of a cellular operator. The traffic can be modeled with the Erlang-B distribution and there are 50 traffic channels/cell. Without considering coverage matters, how many cells must the operator use to obtain 2% blocking in the following cases. a) The cellular phone penetration is 10% and each subscriber generates 25 mErlang traffic during the busy hour. b) The cellular phone penetration has increased to 80% and each subscriber still generates 25 mErlang traffic during the busy hour. c) The cellular phone penetration has increased to 80% but the new subscribers coming after the first 10% will produce a traffic inversely proportional to the penetration during the busy hour (the first new subscriber 25 mErlang). d) If the population density is 100 inhabitants/km2, what would be the cell size in the three previous cases based on pure capacity considerations?

Solution a) T = PopulationPenetrationUser traffic = 106 0.1 0.025 = 2500 Erlang With 50 traffic channels and 2 % blocking target the Erlang B table gives the maximum cell traffic T50,0.02 = 40.24 Erlang . Thus the needed number of cells is

N cell =

T T50,0.02

2500 = 62.18 63 40.24

b) T = PopulationPenetrationUser traffic = 106 0.8 0.025 = 20000 Erlang

N cell =

T T50,0.02

20000 = 497.01 498 40.24

c) Now the user traffic vs. penetration model can be written in the form

To =

0.1 0.025 p

and the total traffic is

T = T p = 0.1 +

0.8

T p = 0.1 p

= 7698.60 Erlang
0.1

0.8 dp = 2500 1 + ln ( p ) = 2500 (1 + ln 8 ) 0.1

N cell =

T T50,0.02

7698.60 = 191.32 192 40.24

d) With a population density of 100 inhabitants/km2, one million population is spread over 1000000/100 = 10000 km2. Then a simple estimate of cell size is the whole area divided by the number of cells:

Aa =

10000 = 158.7 km2 63 10000 Ac = = 52.1 km2 192

Ab =

10000 = 20.1 km 2 498

Problem 5 A cellular network operator has been granted a 210 MHz bandwidth. He intends to use an FDMA/TDMA/FDD-system with 200 kHz carrier spacing and 8 timeslots/carrier in both down-link and up-link. The co-channel reuse factor is 7. a) How many traffic channels are available in each base station? It is assumed that all signaling is multiplexed on the traffic channels. b) How much traffic can be served in a base station, if the blocking probability target is 2 %, and the traffic obeys Erlang-B? c) The frequency regulator requires an average channel load of 90%. To which approximate value will the blocking probability increase? Solution a) With FDD (Frequency Division Duplex) e.g. the down-link can use half of the available frequency resource, so the total number of carriers is

N carrier =

total DL bandwidth 10 = = 50 carrier spacing 0.2

The number of carriers in a base station is obtained by dividing the total number of carriers with the co-channel reuse factor, and taking the nearest smaller integer

50 N BS _ carrier = int = int ( 7.143) = 7 7


As each carrier will contain 8 timeslots, the number of traffic channels in a base station is

N BS _ TCH = 8 7 = 56
b) The amount of served traffic is given by

Tserved = (1 B ) Toffered = 0.98 45.88 = 44.96 Erlang


The numerical of the offered traffic is obtained from the Erlang B table. c) The average fractional channel load is TCH =

(1 B ) Toffered
N BS _ TCH

Insertion of different offered traffic values in the Erlang-B expression

TN B = N! N Tk k =0 k !

Toffered 50 51 52 53 54 55 56 57 58 59 60

B 0.0458 0.0537 0.0621 0.0709 0.0801 0.0896 0.0994 0.1094 0.1195 0.1297 0.1399

0.8520 0.8618 0.8710 0.8793 0.8870 0.8941 0.9006 0.9065 0.9120 0.9169 0.9215

Toffered 55 55.1 55.2 55.3 55.4 55.5 55.6 55.7 55.8 55.9 56

B 0.0896 0.0906 0.0916 0.0925 0.0935 0.0945 0.0955 0.0965 0.0974 0.0984 0.0994

0.8941 0.8948 0.8955 0.8961 0.8968 0.8974 0.8981 0.8987 0.8993 0.9000 0.9006

It appears that an offered traffic of 55.9 Erlang gives a fractional channel load of 0.9, and the blocking probability is 9.84 %.

Problem 6 In a cellular service area the offered traffic density is 4 Erlang/km2. The number of traffic channels in a base station is 36, and the blocking probability target is i) 1%, ii) 2%, iii) 5%, iv) 10 %. Omnidirectional base station antennas are used a) If the cell size can be dimensioned for maximum capacity, what is the cell area for the given blocking probability targets? b) Calculate the maximum path length for the given blocking probability targets, 1) with circular cell shape, 2) with hexagonal cell shape, and 3) with quadratic cell shape. Solution a) The cell size (area) is obtained by dividing the traffic values for the different TB,36 . Numerical blocking probabilities with the traffic density: Acell = results are given in Table 6.1 Table 6.1 TB,36 Acell/km2 B = 1% B = 2% 25.51 27.34 6.378 6.835 B = 5% B = 10% 30.66 34.50 7.665 8.625

b) The area of a cell with circular shape is A = R 2 R = The area of a cell with hexagonal shape is A = 6

The area of a cell with quadratic shape is A = R 2

R 3 RR= 2 2

A 6.75

R=

A2

R R R R

Table 6.2 B = 1% Acell/km2 6.378 Rcicrle/km 1.425 Rhexagon/km 1.567 Rquadrate/km 1.786 B = 2% 6.835 1.475 1.622 1.849 B = 5% 7.665 1.562 1.718 1.958 B = 10% 8.625 1.657 1.822 2.077

P9. In a cellular system a gain term (e.g. BS power or antenna gain) is increased with 3 dB. This increase is fully utilised to increase coverage. How many percent will i) the cell radius, ii) the cell area increase under the assumption of a single slope average path loss model with the path loss exponent n = 2, 3, 4, or 5?

L1 = L0 + 10n log ( r1 ) r L2 L1 = L = 10n ( log ( r2 ) log ( r1 ) ) = 10n log 2 L2 = L0 + 10n log ( r2 ) r

r2 A2 L 10 n L 10 n 2 = 10 = 10 = 10L 5n r1 A1
n 2 3 4 5 r2 r1 1.413 41.3% 1.259 25.9% 1.188 18.8% 1.148 14.8% A2 A1 1.995 99.5% 1.585 58.5% 1.413 41.3% 1.318 31.8%

down-link
BS, Pbs= 40 dBm comb. filt. L = 3 dB ant+feed. L = 4 dB BS ant. G=10 dB radio ch. L = Lp ant+feed. G = 0 dB MS, SMS= -100 dBm

up-link
MS, Pms 29 dBm
ant+feed. G = 0 dB radio ch. L = Lp BS ant. G=10 dB feeder L = 4 dB div. comb. G = 4 dB BS, SMS= 102 dBm

The figure shows the system parameters of a cellular phone radio link that should be considered in the radio link budget. The average path loss as function of the distance d is Lc = 133, 8 + 33, 8 lg(d ) . a) Determine the allowed radio channel loss in both down-link and up-link. b) How large is the cell radius based on the average path loss (giving 50 % coverage probability at the cell border)? Consider both down-link and uplink.

SOLUTION a) In the down-link

Pbs Lcomb. filt . L feed + Gbs L p,dl + Gms Sms L p,dl Pbs Lcomb. filt . L feed + Gbs + Gms Sms 40 3 4 + 10 + 0 + 100 = 143dB

In the up-link

Pms + Gms L p,ul + Gbs L feed + Gdiv Sbs L p,ul Pms + Gms + Gbs L feed + Gdiv Sbs 29 + 0 + 10 4 + 4 + 102 = 141dB

b) Based on the allowed down-link and up-link path loss

133.8 + 33.8 log Rdl = 143 Rdl = 10(

143133.8 ) 33.8 141133.8 ) 33.8

= 1.87km = 1.63km

133.8 + 33.8 log Rul = 141 Rdl = 10(

P10. In this task the indoor coverage probability at the cell border should be estimated when the corresponding outdoor coverage probability is known. The received power level is obtained from the radio link budget, and it si given by the expression Prx = S + SFM = Ptx Ltx + Gtx L p + Grx Lrx , where S is the receiver sensitivity level, Ptx , Ltx , and Gtx are the transmit power level, antenna feeder sysstem loss, and antenna gain respectively, Grx and Lrx are the receiver antenna gain and feeder loss. The slow fade margin is SFM = INVQ(1 p ) , and the average path loss with the actual radio link parameters L p = 127.1 + Lw Ahms + 35.2log r and

Ahms = 2.6hms 3.9 . The function INVQ( ) gives the argument of the Qfunction when the value of the Q-function is known. Lw is the outer wall average penetration loss. The outdoor MS antenna height is 1.5 m, shadow fading standard deviation = 6 dB, and the independent log-normal average wall penetration loss is 10 dB and standard deviation 8 dB. All other parameters than SFM, Ahms and average path loss remain unchanged in the indoor case. Calculate the indoor coverage probability on the first floor (MS antenna height 5 m) at the cell border, where the outdoor coverage probability is p = 0.90.

SOLUTION In the outdoor situation

SFM = INVQ(0.1) 6 = 1.28 6 = 7.68dB Ahms = 2.6 1.5 3.9 = 0dB


In the indoor situation

L = 10 9.1 = 0.9dB

Ahms = 2.6 5.0 3.9 = 9.1dB

SFM = 7.7 0.9 = 6.8dB = INVQ(1 p ) tot = INVQ(1 p ) 6 2 + 82 = 10 INVQ(1 p ) 6.8 INVQ(1 p ) = = 0.68 1 p = 0.248 p = 0.752 10

The Q- and INVQ-function values can be obtained from the graph in the next slide

1 0.5 0.2 0.1 0.05 Q(x) 0.02 0.01 0.005 0.002 0.001 0 1 x 2 3

QFUNCT3.dsf

P11. Based on log-normal shadow fading with 8 dB standard deviation, determine the shadow fading margin needed in radio link budget calculations in a single microcell with the cell coverage probability target a) 90%, b) 95%. Obtain the path loss exponent using the COST231 WalfischIkegami average path loss model when hr = 24 m and hbs = 8 m. Solution From the COST 231 Walfisch-Ikegami average path loss model the terms containing log(d) must be identified, and the sum of the coefficient divided by 10 gives the path loss exponent. There should not be any other terms containing d, so this approach is not valid for d < 0.5 km

hroof hms PLE = n = 0.1(20 + kd ) = 0.1 20 + 18 + 15 hroof 24 8 = 0.1 38 + 15 = 4.8 = 1.67 24 n

Based on the normalised values SFM/ in the figure on next page

SFM (90%) = 8 0.544 = 4.35 dB SFM (95%) = 8 0.963 = 7.70 dB

Normalised shadow fade margin as function of propagation parameters for different cell coverage probabilities 1.2 1.0 0.8 0.6 SFM/SF 0.4 0.2 0 0.2 0.4 1 Fu = 0.80 Fu = 0.95

Fu = 0.90 Fu = 0.85

SFM_outage.dsf

1.5

SF/n

2.5

P12. In the lecture material the effective gain of a mast-top amplifier was found to be 9.08 dB when the antenna noise temperature equals the standard temperature To = 290 K. The equipment parameters are Gmta = 12 dB, Fmta = 2 dB, Frx = 10 dB, L fs = 4 dB To which value is the effective gain reduced when the antenna noise temperature due to man-made noise is 10 To? SOLUTION The general expression for the BS receiver total input noise is

L fs 1 Gmta Pn _ rx = k (Ta + Tmta )B + kT fs B + kTrx B L fs L fs


and the gain reduction equals the noise increase due to the mta effective gain is

Pn _ rx Gmta = 10 lg Pn _ rxo

G L fs 1 mta 10T + F ( o ( mta 1) To ) + L To + ( Frx 1) To L fs fs = 10 lg L fs 1 10To + To + ( Frx 1) To L fs L fs

G L fs 1 mta T + T T fs + Trx ( a mta ) + L fs L fs = 10 lg L fs 1 Ta + T fs + Trx L fs L fs

G L fs 1 mta k (Ta + Tmta ) B + kT fs B + kTrx B L fs L fs = 10 lg L fs 1 kTa B + kT fs B + kTrx B L fs L fs

10

Gmta ( Fmta + 9 ) + L fs 1 + ( Frx 1) L fs = 10 lg 10 + L fs 1 + ( Frx 1) L fs

Gmta ( Fmta + 9 ) + Frx L fs 1 = 10 lg 9 + Frx L fs

11

101.2 100.2 + 9 + 10 100.4 1 = 10 lg = 10.88 dB 0.4 9 + 10 10


The resulting effective gain is

Gmta _ eff = Gmta Gmta = 12 10.88 = 1.12 dB

Conclusion: In high background noise environments a large part of the mta advantages is lost

12

P13 Based on the single cell coverage probability target, derive an expression of the coverage probability as a function of the distance normalised to the cell radius and draw the graph when = 6 dB, n = 4, and the cell coverage probability is i) 0.90, ii) 0.95. SOLUTION With log-normal shadow fading the coverage probability on distance r is

S Prx ( R ) + 10n log ( r R ) 10 log ( r R ) SFM = Q Pcov ( r ) = Q n SFM SFM SFM Pcov ( R ) = Q = 1 Q = INVQ ( 1 Pcov ( R ) ) 10log ( r R ) Pcov ( r ) = Q INVQ ( 1 Pcov ( R ) ) n

From the graph of SFM = f ( n , Pcov ( R) ) we get

SFM = f ( 1.5, 0.9 ) 0.494

SFM = f ( 1.5, 0.95 ) 0.923

13

Normalised shadow fade margin as function of propagation parameters for different cell coverage probabilities 1.2 1.0 0.923 0.8 SFM/SF 0.6 0.494 0.4 0.2 0 0.2 0.4 1 Fu = 0.80 Fu = 0.90 Fu = 0.85 Fu = 0.95

SFM_outage.dsf

1.5

SF/n

2.5

14

10-0

10-1 Fu=0.95 10-2 Pcov(r) 10-3 Fu=0.90

10-4

10-5
SFM_outage.dsf

r/R

15

P14 Cell coverage probability planning target is i) 90 %, ii) 95 %. The planning approach is based on j) connection to a dedicated base station (single cell), jj) connection to the best base station in a 7 cell cluster. The path loss exponent is 4, and the following four log-normal shadow fading cases are considered: k) = 4 dB, kk) = 6 dB, kkk) = 8 dB, and kkkk) = 10 dB. a) What is the obtained coverage in case jj), if the coverage planning has been based on case j)? b) How many times can the number of base stations be reduced, if the coverage planning can be based on case jj) instead of case j)? SOLUTION a) The coverage probability will depend on the shadow fade margin (SFM) which depends on the single cell border coverage probability and on the parameter /n. Using the graph Fu, single _ cell = f ( Pcov ( R), n ) we get the value for Pcoverage ( R ) , and the using this value the asked cell coverage probability can be obtained from the graph Fu,multi _ cell = f ( Pcov ( R ), n )

16

Single cell coverage probability


1.0 /n = 1.0 1.1 1.2 1.3 0.8 1.4 1.5 1.6 1.7 0.6
Fu 0.4 1.8 1.9 2.0 2.1 2.2 2.3 2.4 2.5 = /n 0.2
Cell_coverage_prob.dsf

0.2

0.772 0.614 0.765 0.868 0.8 0.4 Pcov(R) 0.6 1.0 0.694 0.737 0.822 0.850

17

0.999 0.999 0.990 0.979

Middle cell coverage probability in a 7 cell cluster


1.0 /n = 1.0 1.1 1.2 1.3 0.8 1.4 1.5

0.6 Fu 0.4 1.6 1.7 1.8 1.9 2.0 2.1 2.2 2.3 2.4 2.5 = /n 0.2
Cell_coverage_prob.dsf

0.2

0.772 0.614 0.765 0.868 0.8 0.4 1.0 Pcov(R) 0.6 0.737 0.822 0.850 0.694

18

The results obtained from the graphs are collected into the following table /n = 4/4 =1 0.614 0.979 0.772 0.999 /n = 6/4 = 1.5 0.694 0.990 0.822 0.999 /n = 8/4 =2 0.737 0.999 0.850 0.999 /n = 10/4 = 2.5 0.765 0.999 0.868 0.999

Fu1 = 0.90 Pcov(R) Fu7 Fu1 = 0.95 Pcov(R) Fu7

b) Using the graph Fu,multi _ cell = f ( Pcov ( R ), n ) we get the value for Pcov ( R ) , and the using these values the SFM/-values are obtained by

SFM = INVQ ( 1 Pcov ( R ) )

The reduction of the necessary SFM is obtained as

SFM = ( SFM1 SFM 7 )


N cell = Aserv _ area Acell N cell 1 N cell 7 = Acell 7 Acell 1 = ( R7 R1 ) From the average path loss we obtain
and as the number of cells is
2

19

( R7

R1 )

= 10SFM /(10n)

= 10SFM / 20
/n = 6/4 = 1.5 0.694 0.507 0.405 0.240 4.482 1.675 0.822 0.923 0.502 0.005 5.508 1.185 /n = 8/4 =2 0.737 0.634 0.417 0.210 6.752 2.176 0.850 1.036 0.519 0.048 7.904 2.484 /n = 10/4 = 2.5 0.765 0722 0.422 0.197 9.190 2.881 0.868 1.117 0.523 0.058 10.590 3.385

Fu1 = 0.90 Pcov1(R) SFM1/ Pcov7(R) SFM7/ SFM N1/N7 Fu1 = 0.95 Pcov1(R) SFM1/ Pcov7(R) SFM7/ SFM N1/N7

/n = 4/4 =1 0.614 0.290 0.345 0.399 2.756 1.373 0.772 0.745 0.467 0.083 3.312 1.464

20

Middle cell coverage probability in a 7 cell cluster


1.0 /n = 1.0 1.1 1.2 1.3 0.8 1.4 1.5

0.6 Fu 0.4 1.6 1.7 1.8 1.9 2.0 2.1 2.2 2.3 2.4 2.5 = /n 0.2
Cell_coverage_prob.dsf

0.2

0.345 0.405 0.417 0.422

0.467 0.502 0.519 0.523

0.4

Pcov(R) 0.6

1.0

Вам также может понравиться