DocumentCode
1133514
Title
Improved Viterbi decoder metrics for two-stage detectors in DS-CDMA
Author
Elezabi, Ayman ; Duel-Hallen, Alexandra
Author_Institution
American Univ., Cairo, Egypt
Volume
3
Issue
5
fYear
2004
Firstpage
1399
Lastpage
1404
Abstract
Modified branch metrics are proposed for single-user Viterbi decoders in two-stage detectors for convolutionally-encoded code-division multiple-access (CDMA) systems with random spreading sequences. The modifications are based on modeling the residual multiple-access interference (RMAI) after subtractive interference cancellation as conditionally Gaussian with time-dependent variance, where the conditioning is on the time-varying user crosscorrelations. A novel estimate of the variance of the total RMAI is presented, and used in the proposed branch metrics. Significant performance gains are demonstrated over the Euclidean branch metric of the standard Viterbi decoder.
Keywords
Gaussian processes; Viterbi decoding; code division multiple access; convolutional codes; interference suppression; radiofrequency interference; random codes; signal detection; spread spectrum communication; DS-CDMA; Euclidean branch metric; Viterbi decoder metrics; convolutional code; direct sequence code division multiple access; random spreading sequences; residual multiple-access interference; subtractive interference cancellation; time-dependent variance; two-stage detectors; Convolutional codes; Detectors; Interference cancellation; Iterative decoding; Maximum likelihood decoding; Multiaccess communication; Multiple access interference; Multiuser detection; Performance gain; Viterbi algorithm; CDMA; Code-division multiple-access; RMAI; modified branch metrics; residual multiple-access interference; single-user Viterbi decoders; subtractive interference cancellation; time-dependent variance; two-stage detectors;
fLanguage
English
Journal_Title
Wireless Communications, IEEE Transactions on
Publisher
ieee
ISSN
1536-1276
Type
jour
DOI
10.1109/TWC.2004.833489
Filename
1343868
Link To Document