High performance MIMO communications with imperfect channel knowledge

The tremendous popularity of wireless communications during the last three decades is mainly attributed to the convenience of allowing user mobility. Driven by a huge market demand, future wireless communication standards are targeting high data rates, high performance, and low cost at high user mobility. This brings forth significant challenges due to the limited available radio spectrum and the random fading nature of the wireless channel. Today, Multiple Input Multiple Output (MIMO) technology is incorporated in almost every evolving standard because of the potential gains that can be achieved using MIMO in terms of diversity, spatial multiplexing, and interference reduction. These gains come at no loss in power or spectral efficiency, however they require accurate channel estimation in addition to increased receiver complexity. Channel estimation is usually acquired though pilot transmission. Due to the large number of channel coefficients that need to be estimated with MIMO communications, pilot transmission creates an overhead and reduces the total system throughput. This problem is especially acute under high mobility conditions which necessitate estimating the channel more frequently.;To alleviate this problem, blind and differential MIMO schemes have been proposed. Although these schemes do not require channel estimation at the receiver, they are less appealing because of their degraded performance and restricted achievable data rates. In this dissertation, we develop high rate blind and differential MIMO communication schemes, namely modified block code for blind MIMO and block differential MIMO. Based on a generalized likelihood ratio detection approach, we derive a high performance unified receiver structure which accommodates variations in the availability and the reliability of the channel information at the receiver. Hence, this unified receiver structure, which we call the Unified Generalized Likelihood Ratio Detector (UGLRD), can be applied to a blind MIMO system, to a differential MIMO system, and to situations where an imperfect channel estimate is available at the receiver. We account for errors in channel estimation both due to noise and due to time variations induced by Doppler effects. We show that the UGLRD outperforms the Mismatched Maximum Likelihood Detector (MMLD). Furthermore, the performance of the UGLRD approaches that of an optimal Maximum Likelihood (ML) detector which assumes perfect channel knowledge independent of the amount of error encountered in channel estimation. The enhanced performance of the UGLRD comes at increased computational complexity at the receiver. The Ordered Branch-Estimate-Bound algorithm for the UGLRD (OBEB-UGLRD) is developed to implement the UGLRD efficiently. The OBEB-UGLRD algorithm offers significant computational reduction at no loss in performance compared to brute force search. We analyze the performance of the UGLRD and give a closed form solution for the Pairwise Error Probability (PEP) from which the Bit Error Rate (BER) of the system can be tightly upper bounded at high SNRs. To sum up, this dissertation presents high data rate high performance low complexity MIMO communication schemes suited for high mobility future wireless communication applications.