A layer-adaptive M algorithm for multiple-input multiple-output channel detection

We present a new detection algorithm for multiple-input multiple-output channels called the CLAM algorithm. The CLAM algorithm is similar to the classical M algorithm for searching the detection tree, except that it varies from one stage to the next the number of children extended from each retained node. These numbers are optimized based on knowledge of the channel at the receiver. This simple enhancement not only reduces complexity, it also enables the CLAM to significantly outperform the M algorithm. For example, on a 4-input 4- output Rayleigh-fading channel with 64-QAM inputs, the CLAM algorithm outperforms the M algorithm by 2 dB at a BER of 10"3, falling only 0.6 dB short of the joint maximum-likelihood detector, while simultaneously reducing the average search complexity by 12%.