Mutual information based rate adaptation for MIMO-OFDM systems with finite signal alphabets

We consider a single user MIMO-OFDM system where spatial multiplexing is applied and formulate rate allocation algorithms to adapt the used modulation alphabet on a per subcarrier basis. As a means to consider the system behavior including the channel code regarding bit or frame error rate (BER, FER) versus normalized average mutual information (NAMI) of an OFDM symbol for several M-QAM alphabets is investigated. To this end an approximation to the mutual information of MIMO systems with M-QAM signal alphabets is derived and compared to channel decomposition approaches to find a complexity/accuracy trade-off. Based on the mutual information results two algorithms using this channel knowledge are introduced. On the one hand, the error rates can be optimized assuming a fixed rate requirement and on the other hand the achieved transmission rate can be optimized under a BER/FER constraint. In both cases a NAMI threshold derived from the simulated system behavior for a specific channel code is used to adapt the allocated modulation alphabets via a greedy algorithm to achieve the target NAMI within an error margin.