CDMA Receivers for High Spectral Utilization MPRG

Transcription

1 CDMA Receivers for High Spectral Utilization 19

2 Types of CDMA Receivers Conventional Single User Receivers Multiuser Receivers 20

3 Why Use Advanced Receivers? CDMA is interference limited CDMA subject to near/far problem on reverse channel Interference and near/far problem limit system capacity These problems are not inherent to CDMA, but rather the result of using conventional correlator Multiuser and Single User receivers mitigate these problems 21

4 Receiver Architectures Multiuser Multistage RAKE Receiver Estimate the interference and subtract it out. Number of stages : 2-3 Number of RAKE fingers : 2 Conventional RAKE correlators employed at each stage Single User Mobile Receiver Linear fractionally spaced adaptive filter of length 45 Adaptive algorithms: Normalized LMS, Linearly Constrained CMA, Griffiths algorithm, Soft Decision Directed Normalized LMS, etc. 22

5 Single User Receiver Objectives Design and implement a mobile receiver capable of operating in heavily loaded and multipath environments Compare existing receiver structures and algorithms on a common simulation testbed Develop novel blind algorithms for CDMA interference rejection Implement promising algorithms on a DSP Map the algorithms from the DSP to the COLT reconfigurable FPGA processor 23

6 Single User Receiver Structures 24

7 Receiver Structures T s T s T s T s r(t) r(n) Fractionally-spaced y(n) transversal filter T s T d ) d(n) The The Complex Complex Linear Linear Fractionally-Spaced Fractionally-Spaced Receiver Receiver (CLFSR) (CLFSR) Adaptation algorithm e(n) d(n) Decision Directed Training 25

8 What Single User Receivers Do Correlation MFR CLFSR Cross correlation of filter weight vector with the spreading code of each user 26

9 Algorithms: Trained Algorithms based on MSE cost function Adaptive receivers offer significant capacity gains Capacity curves: Synchronous AWGN channel, E b /N 0 = 7.5 db 27

10 Algorithms: Trained The adaptive receivers are near-far resistant Capacity curves: Asynchronous AWGN channel, E b /N 0 = 7.5 db 28

11 Results Can adaptive receivers exploit multipath? Capacity curves: Urban channel (COST 207 model), E b /N 0 = 23.5 db 29

12 Hardware Architecture Control Bus I arm SRAM Signal at IF NCO BPF AGC A/D I DDC Q Digital Dual Port SRAM fixed point DSP System Executive floating point DSP Q arm Data Bus Adaptive 30

13 Multiuser Receiver Objectives Design and implement a multiuser multistage RAKE receiver for interference cancellation and multipath mitigation at the base station. Provide significant capacity gains and near/far resistance compared to current CDMA receivers with a complexity that makes implementation feasible. Development of analytical and simulation tools to study multiuser interference cancellation for CDMA. Address implementation issues not considered in the literature. 31

14 CDMA Receivers CDMA Receivers Single User Multiuser Conventional Adaptive s Optimal (MLSE) Sub-optimal Linear Non-linear Decorrelator MMSE Multistage Reduced Tree Search Neural Networks Decision Feedback Conventional First Stage Decorrelating First Stage Succ. Canc. First Stage Decorrelating Decision Feedback Successive Cancellation 32

15 Multiuser Receivers Offer significant capacity gains over conventional receiver Provide robustness in near/far situations Parallel cancellation provides excellent tradeoff between complexity and performance 33

16 Approach R(t) Initial Multiuser Estimation MAI Regeneration - Multiuser Estimation Improved Estimates User 1 Synch 1 Z 1,i x 2a 1 (t-t 1 ) T User 1 Synch 1 Z 1,i User K Synch K Z K,i x 2a k (t-t k ) T User K Synch K Z K,i 34

17 Performance AWGN Significant improvements are observed with few stages of interference cancellation for different channel conditions. Multistage cancellation tends to the single user bound as the number of stages increases. Rayleigh Fading 35

18 Interference Cancellation Scheme User 1 User 2 C 1 C 2 c Colt-based s FPGA-based Interference Cancellation and Multiuser Receiver Differential Decoder User Bit Streams User 3 User 4 C 3 C 4 Σ - Σ Σ User 1 User 2 User 3 C 1 C 2 C 3 Σ c Colt-based s FPGA-based Interference Cancellation c C = 0.75 User 4 C 4 Σ - Acquisition & Tracking Receiver Stage 1 Receiver Stage 2