Impact of Interference Model on Capacity in CDMA Cellular Networks. Robert Akl, D.Sc. Asad Parvez University of North Texas

Transcription

1 Impact of Interference Model on Capacty n CDMA Cellular Networks Robert Akl, D.Sc. Asad Parvez Unversty of North Texas

2 Outlne Introducton to CDMA networks Average nterference model Actual nterference model Optmzed capacty 2D Gaussan user model Conclusons

3 Frequency Frequency Code Dvson Multple Access (CDMA) Overvew Multple access schemes FDMA TDMA CDMA Call 4 Call 3 Call 10 Call 11 Call 12 Call 7 Call 8 Call 9 Frequency Call 1 Call 2 Call 3 Call 4 Call 2 Call 4 Call 5 Call 6 Tme Call 1 Tme Call 1 Call 2 Call 3 Tme Code

4 Spread Spectrum: Drect Spreadng 1 bt perod 1 chp perod Data Sgnal Data Sgnal Interference f PN-code f Coded Data Sgnal PN-code Decoded data Sgnal Recovered Data Sgnal f f

5 Factors Affectng Capacty Power Control c2 c1 Pt1 Pt2 Pr1 Pr2 Base Staton Pt1: Power transmtted from c1 Pt2: Power transmtted from c2 Pr1: Power receved at base staton from c1 Pr2: Power receved at base staton from c2 Pr1 = Pr2 d2 d1 Dstance

6 Factors Affectng Capacty (cont.) Soft handover of calls Tme

7 Factors Affectng Capacty (cont.) Unversal frequency use A E F D A A G C B A A A A A A A A A TDMA or FDMA Reverse lnk vs forward lnk CDMA Forward lnk Reverse lnk Voce actvty factor

8 Relatve Average Inter-cell Interference Model Ij Relatve average nterference at cell caused by n j users n cell j Cell r j m rj x, y10 10 nj I j E dax, Cj m 2 r x, y/ Aj y A rj Cell j r m 2 s nj j Ij e Aj Cj m r ( x, ( x, y) y) da( x, y) B da Back

9 Interference Matrx F j, M M1 M2 MM where F[ j, ] and n j I j / n j for, j 1,..., M, s the number of users n cell j Hence, the total relatve average ntercell nterference experenced by cell s M j1 I njf j, C I 2 1 I 2 1 F[1,2] M M1 M2 MM

10 Relatve Actual Inter-cell Interference Model Interference matrx F cannot be calculated n advance Instead, a new nterference matrx U s computed as follows For a user k n cell j, the relatve actual nterference offered by ths user to cell s s Uj e k 2 r r j m D Hence, the total relatve actual nter-cell nterference at cell caused by every user n the network s I M j1 n j k1 U, for j j k E k users n cell j

11 Actual Interference Matrx U Example: for a new call n cell 2, compute row matrx U[2,] for = 1,,M usng equaton D U [ M] Update 2 nd row of nterference matrx U by addng the above row matrx to t. U[ j, ] U M M1 M2 MM

12 Capacty The capacty of a CDMA network s determned by mantanng a lower bound on the bt energy to nterference densty rato, gven by E I b 0 RE n I b Eb 1 / W N 0 for 1,..., M F W = Spread sgnal bandwdth R = bts/sec (nformaton rate) α = voce actvty factor n = users n cell N 0 = background nose spectral densty Let τ be that threshold above whch the bt error rate must be mantaned, then by rewrtng Eq. F n I W / R 1 E b 1 / N 0 for 1 c 1,..., M eff G Back

13 Capacty Cases Equal capacty: all cells have an equal number of users n n for all Optmzed Capacty: A set of users n each cell obtaned by solvng followng optmzaton problem max n subject to M 1 n n I, c eff for 1,..., M., H

14 Smulatons Network confguraton COST-231 propagaton model Carrer frequency = 1800 MHz Average base staton heght = 30 meters Average moble heght = 1.5 meters Path loss coeffcent, m = 4 Shadow fadng standard devaton, σ s = 6 db Processng gan, W/R = 21.1 db Bt energy to nterference rato threshold, τ = 9.2 db Interference to background nose rato, I 0 /N 0 = 10 db Voce actvty factor, α = These values n Eq. G gve upper bound on the relatve nterference n every cell, c_eff =

15 Smulatons Equal Capacty Average nterference Users n each cell: 18 Actual nterference Users n each cell: 17

16 Smulatons Optmzed Capacty Vs Actual Interference Capacty Optmzed Capacty usng average nterference = 559 Smulated Capacty usng actual nterference = 554

17 More Smulatons Actual Interference Smulated Capacty = 564 Smulated Capacty = 568

18 Indvdual Cell Capacty Comparson Comparson of cell capacty for 3 smulaton trals. Comparson of average cell capacty for 50 smulaton trals.

19 Extreme Cases Usng Actual Interference Non-Unform Dstrbuton Maxmum network capacty of 1026 wth best case non-unform user dstrbuton Maxmum network capacty of 108 wth worst case non-unform user dstrbuton

20 Model User Dstrbuton by 2D Gaussan Mean = 0 and standard devaton = 200 Mean = 0 and standard devaton = 500

21 Model User Dstrbuton by 2D Gaussan Mean = 0 and standard devaton = 900 Non-zero mean, standard devaton between

22 Conclusons Actual nterference model s computatonally ntensve. Capacty obtaned usng average nterference s close to the capacty obtaned usng actual nterference for unform user dstrbuton. Average nterference model cannot predct extreme varatons n network capacty under nonunform user dstrbuton. Can use 2D Gaussan dstrbuton to model unform and non-unform user dstrbuton.