Quasi-maximum-likelihood detector based on geometrical diversification greedy intensification

This letter proposes a quasi optimum maximum likelihood detection technique based on Geometrical Diversification and Greedy Intensification (GDGI). The presented detector scheme is shown to achieve almost optimal performance for all signal-to-noise ratio (SNR) values and a cubic computation complexity in the problem dimension. It possesses a regular structure well suited for hardware implementation. Simulation results show that for a system with a high dimension of n = 60, the loss is approximately 0.35 dB at BER=10^-5 compared to an optimal decoding.