MMSE-GDFE Lattice Decoding for Under-determined Linear Channels

摘要

Recently, the authors established the fundamentalrnrole of minimum mean square error generalizedrndecision-feedback equalizer (MMSE-GDFE) lattice decodingrnin achieving the optimal diversity-vs-multiplexing tradeoffrnof delay limited multiple-input multiple-output (MIMO)rnchannels. This optimality motivates the current work wherernwe exploit this technique in constructing an efficient decodingrnalgorithm for under-determined linear channels. Thernproposed algorithm consists of an MMSE-GDFE front-endrnfollowed by a lattice reduction algorithm with a greedy orderingrntechnique and, finally, a lattice search stage. By introducingrnflexibility in the termination strategy of the latticernsearch stage, we allow for trading performance for arnreduction in the complexity. The proposed algorithm isrnshown, through experimental results in MIMO quasi-staticrnchannels, to offer significant gains over the state of the artrndecoding algorithms in terms of performance enhancementrnand complexity reduction. From one side, when the searchrnis pursued until the best lattice point is found, the performancernof the proposed algorithm is shown to be within arnsmall fraction of a dB from the maximum likelihood (ML)rndecoder while offering a large reduction in complexity comparedrnto the most efficient implementation of ML decodingrnproposed by Dayal and Varanasi (e.g., an order of magnitudernin certain representative scenarios). On the other side,rnwhen the search is terminated after the first point is found,rnthe algorithm only requires linear complexity while offeringrnsignificant performance gains (in the order of several dBs)rnover the linear complexity algorithm proposed recently byrnYao and Wornell, and independently by Windpassinger andrnFisher.