On combining SDMA and B-IFDMA: Multi-user detection and channel estimation

Abstract

In this paper, we combine two different multiple access, namely, Space Division Multiple Access (SDMA) and Block-Interleaved Division Multiple Access (B-IFDMA), for uplink transmission, and we name the combined multiple access SD/B-IFDMA. Since SDMA suffers from multiple access interference (MAI) and B-IFDMA from intersymbol interference (ISI), a multi-user detector that can counteract both MAI and ISI is needed for SD/B-IFDMA. We propose three low complexity linear multi-user detectors for SD/B-IFDMA, namely, Zero Forcing (ZF), Minimum Mean Square Error (MMSE) and Non-Iterative Constrained Least Squares (NICLS). The NICLS is a heuristic multi-user detector that tries to improve the performance of ZF without the necessity of having to estimate the noise variance. Additionally, we also address the channel estimation procedure for SD/B-IFDMA by applying orthogonal time multiplexing training and using Chu sequence as the training sequence. Two estimators are considered for channel estimation for SD/B-IFDMA, namely, frequency domain Least Squares (LS) and time domain low complexity Maximum Likelihood (lcML). From bit error performance, it is shown that the MMSE multiuser detector performs best followed by NICLS and ZF. For performance assessment of the channel estimators, ZF multiuser detection is used. It is shown that the lcML outperforms LS with the penalty of having higher computational complexity.