Supervised Learning Method for Link Adaptation Algorithm in Coded MIMO-OFDM Systems

Abstract

The traditional link adaptation scheme requires packet error rate based on channel state information (CSI). The adaptive selection of the modulation and coding schemes (MCS) is usually decided according to estimation of packet error rate. Unfortunately, the integration of channel coding, OFDM technology and multi-antenna technology in Multiple-Input, Multiple-Output-Orthogonal Frequency Division Multiplexing(MIMOOFDM) wireless communication systems makes traditional prediction of packet error rate very complicated. This paper proposes a new framework based on supervised learning approach k-nearest neighbor (k-NN) algorithm for adaptive modulation and coding (AMC) in MIMO-OFDM wireless systems. With the singular value decomposition (SVD) of the channel matrix, the signal-to-noise ratio (SNR) on each spatial stream is extracted as a feature set. A classification scheme is then proposed to match channel implementations to different MCSs. The simulation results show that the proposed framework can successfully classify each MCS and perform perfect selection of MCS for frequency flat fading channels.