An outer loop link adaptation for BICM-OFDM that learns

Wireless BICM-OFDM systems usually perform some link adaptation procedure in order to adapt their transmission parameters to the changing channel. It is common practice to choose modulation and code rate based on thresholds on the signal-to-noise ratios (inner loop link adaptation), while these thresholds are shifted in an external control loop (outer loop link adaptation). This paper proposes a new approach for adjusting the threshold offset. Adaptive kernel regression is used in order to learn the relationship between offsets, the channel state, and packet error rates for each code rate in an online fashion. The proposed algorithm exploits this knowledge when selecting offsets. This is in contrast to current approaches, which do not anticipate the effect of changes to the offset but rely on probing only. Another advantage is that frequency-selective modulation can (but does not have to) be employed. Some less-known arguments in favor of frequency-selective modulation are pointed out.