Optimization Of Probabilistic Data Association Algorithm And Its Application In Interference - Limited Wireless Communication System

Detection of a vector of symbols simultaneously transmitted over the interference-limited multiple-input multiple-output (MIMO) channels is of fun-damental importance for future wireless communications. This is because the general mathematical model of MIMO detection underpins a wide range of rel-evant applications including (but not limited to) the symbol detection for spa-tial multiplexing multiple-antenna systems, the multiuser detection (MUD) for code-division multiple-access (CDMA) and space-division multiple-access (S-DMA) systems, as well as the equalization for band-limited channels with inter-symbol interference (ISI). More importantly, with the evolution of wireless net-works, it is inevitable to invoke the "network MIMO" concept, an combination of the general MIMO mathematical model and the cooperative communica-tions, to deal with the ubiquitous multiple mutually interfering transmissions taking place in overlapping signal domains. Associated with this evolution, it is identified that the achievable performance of the highly spectrum-efficient wireless networks is typically interference-limited, rather than noise-limited as for the traditional point-to-point systems. Against this backdrop, detection of a vector of symbols over the interference-limited MIMO channels under various novel constraints has imposed considerable new challenges.In this thesis, the probabilistic data association (PDA) method that is well-suited for large-scale MIMO systems is investigated. The PDA algorithm is essentially an interference-modelling approach based on Gaussian approxima-tion. The greatest advantage of the PDA algorithm is that it imposes a low computational complexity which increases as a cubic function of the problem size, while providing competitive performance. The major contributions of this thesis are summarized as follows.Firstly, a multi-vector PDA approach is proposed for symbol detection in spatial multiplexing MIMO systems. By designing a joint detection (JD) struc-ture for multiple consecutive symbol vectors of the same transmission burst, more a priori information is exploited when updating the estimated marginal probabilities for each symbol per iteration. Therefore, the proposed multi-vector PDA-JD detector outperforms the conventional single-vector PDA de-tector in the context of correlated input bit streams, which is achieved at the cost of a higher but retains polynomial-time computational complexity. Final-ly, the conventional single-vector PDA detector is shown to be a special case of the proposed multi-vector PDA-JD detector.Secondly, the rigorous mathematical relationship between the vector s-pace of transmitted bits and the vector space of transmitted M-ary rectangular quadrature amplitude modulation (M-QAM) symbols in spatial multiplexing MIMO systems is thoroughly investigated. It is revealed that the two vector spaces are linked by linear/quasi-linear transformations which are explicitly characterized by some transformation matrices. This result may potentially facilitate many signal processing problems of wireless communications. For instance, when used in the symbol detection problem, it enable us to trans-forms the conventional three-step "signal-to-symbol-to-bits" decision process to a simpler "signal-to-bits" decision process.Thirdly, with the aid of the above modulation transformation matrices, a unified bit-based PDA (BPDA) detection method of rectangular M-QAM sym-bols is proposed for spatial multiplexing MIMO systems. It is shown that the novel linear natural mapping based B-PDA approach attains an improved detec-tion performance, despite dramatically reducing the computational complexity in contrast to the conventional symbol-based PDA aided MIMO detector. In addition, It is revealed that the linear natural mapping based B-PDA approach-es the lower bound performance provided by the binary reflected Gray mapping based B-PDA, which either requires the idealized perfect modulation matrix es-timation or imposes higher computational complexity. Therefore, a conclusion is drawn that it is preferable to use the linear natural bit-to-symbol mapping, rather than the nonlinear binary reflected Gray mapping when the B-PDA de-tector is used in uncoded MIMO systems.Finally, the applications of the PDA method to some more sophisticated systems which are of great interest to the contemporary wireless community is explored. In particular, a base station (BS) cooperation aided distributed soft reception scheme using the PDA algorithm and soft combining (SC) is proposed for the uplink of multiuser multicell MIMO systems. The realistic19-cell hexagonal cellular model relying on unity frequency reuse (FR) is con-sidered, and local cooperation based message passing is used instead of a global message passing chain for the sake of reducing the backhaul traffic. It is shown that despite its low additional complexity and backhaul traffic, the proposed distributed PDA (DPDA) aided SC (DPDA-SC) reception scheme significant-ly outperforms the conventional non-cooperative benchmarkers. Furthermore, since only the index of the possible discrete value of the quantized converged soft information has to be exchangeed for SC in practice, the proposed DPDA-SC scheme is relatively robust to the quantization errors of the soft information exchanged. As an appealing benefit, the backhaul traffic is dramatically re-duced at a negligible performance degradation.