A FFT-Like MIMO Detection Algorithm

Abstract

This paper studies the multiple-input-multiple-output (MIMO) detection problem. Existing works model the MIMO detection as a tree or factor graph. This work adds a new member called the clique graph model to the MIMO detection graph model family so that MIMO detection can be transformed into finding the maximum clique in the constructed graph. Then, a clique based MIMO detection algorithm is developed, which uses an architecture similar to the Fast Fourier Transform (FFT) and can be executed in parallel naturally. The execution time of the algorithm can be proportional to log 2 N t where N t is the transmit antenna number. Simulation results are provided to evaluate the bit error rate (BER) performance of the proposed algorithm. INDEX TERMS Detection, graph model, MIMO, parallel processing.