A signal power adaptive regularized sphere decoding algorithm based multiuser detection technique for underdetermined OFDM/SDMA uplink system

• Published in: International journal of communication systems Vol. 32; no. 1
• Main Authors: Shahnaz, K V, Ali, C K
• Format: Journal Article
• Language: English
• Published: Chichester Wiley Subscription Services, Inc 10.01.2019
• Subjects: Adaptive algorithms; Decoding; Error detection; multiuser detection; OFDM/SDMA; Orthogonal Frequency Division Multiplexing; Parameter modification; rank‐deficiency; Regularization; Regularization methods; sphere decoder; Tikhonov regularization; underdetermined channel
• ISSN: 1074-5351; 1099-1131
• DOI: 10.1002/dac.3838

Summary This work considers a regularized Fincke‐Pohst Sphere Decoder (R‐FSD) to detect the multiuser data of orthogonal frequency division multiplexing/space division multiple access (OFDM/SDMA) uplink system with underdetermined and rank‐deficient channel. Rank‐deficiency of the detected channel coefficients makes error‐free multiuser detection (MUD) a difficult task. In literature, most of the papers deal with either a determined or over‐determined full‐rank system. The method proposed in this work transforms an original ill‐posed least squares (LS) problem to a well‐posed one at the receiver, by using the standard Tikhonov regularization method. This is an efficient, direct, and less complex approach where the channel is modified using a regularization parameter that adapts to the signal power at the receiver. The result obtained is compared with maximum likelihood (ML), zero forcing (ZF), minimum mean squared error (MMSE) and ordered successive interference cancellation (OSIC) based detection techniques. Underdetermined and rank‐deficient channel coefficients makes error‐free multiuser detection (MUD) a difficult task in orthogonal frequency division multiplexing/space division multiple access (OFDM/SDMA) uplink system. A regularized Fincke‐Pohst sphere decoder (R‐FSD) proposed in this work transforms an original ill‐posed least squares (LSs) problem to a well‐posed one. This is an efficient, direct, and less complex approach where the estimated channel matrix is modified using a regularization parameter that adapts to the signal power at the receiver.