Multi-antenna systems, also known as multi-input multi-output (MIMO) systems, introduced by Foschini and Telatar in the late 90s, can increase spectral efficiency by utilizing the spatial domain. While spatial multiplexing MIMO fully exploits this increased spectral efficiency, it makes detection very complex at the receiver. There have been, therefore, a lot of works on reducing the complexity of detection algorithm. Among them, the K-best detection algorithm shows a BER performance very close to optimum ML detection, and also is VLSI friendly.
Sorting plays an important role in K-best detection. Distributed sorting, previously proposed in the literature, reduces this sorting complexity and thus leads to, in some cases together with increased throughput, area reduction in VLSI implementations. However, it is observed that it also leads to severe BER performance degradations. In this thesis distributed sorting in conjunction with interleaving is proposed to recover that degraded BER performance.
The concept of ‘skew’ is introduced such that performance improvements can be expected by removing or reducing these ‘skews’, which is done by interleaving. Finding optimal interleaving patterns is almost impossible, and therefore, a general systematic heuristic interleaving rule is proposed. It is shown that, in the context of 802.11n, using the proposed scheme, 2.5dB improvement in SNR at BER $10^{-4}$ can be achieved with a hardware overhead of 34%, compared to the conventional distributed sorting scheme.