Codebook Design Based on Self-Organizing Map Clustering for Limited Feedback MIMO Systems

Codebook design is a core technology in limited feedback multiple-input multiple-output (MIMO) communication system. However, conventional codebook designs usually assume that the channel vectors obey a uniform distribution. Motivated by the excellent classification and analysis abilities of clustering algorithms, we propose a self-organizing map (SOM) clustering based codebook design method that can adaptively learn an arbitrary propagation environment. By utilizing the statistical distribution of cumulative channel samples to construct quantized codewords, the proposed method is able to update the codebook adaptively according to the instantaneous channel state information. Simulation results are consistent with theoretical analysis, and demonstrate that the proposed codebook design method can achieve a significant performance improvement in the achievable rate region compared with the conventional one, especially in non-uniform channel distribution scenarios.