Link adaptation in MIMO-OFDM with practical impairments

Performing link adaptation in multiple-input-multiple-output orthogonal frequency division multiplexing (MIMO-OFDM) systems is challenging due to the problem of mapping the channel state information to a frame error rate (FER) value. This difficulty comes from the spatial and frequential selectivity of the channel, which makes the different symbols in a codeword to observe different signal to noise ratio values. Moreover, practical impairments like non-Gaussian noise, different frame lengths, or channel nonlinearities can severely affect the adaptation procedure in real scenarios. In this paper we study different FER prediction techniques, which can be classified in parametric, non-parametric and semi-parametric. We evaluate the performance of the FER predictors under practical impairments, and compare the achieved throughput when used in conjunction with link adaptation algorithms.