A Novel Constrained Estimator for Selective Feedback in OFDM and its Implications

Reduced feedback schemes facilitate scheduling and rate adaptation in orthogonal frequency division multiplexing systems while reducing the number of subchannels (SCs) for which channel state information (CSI) is fed back. In the popular selective feedback scheme, each user feeds back CSI about only its strongest SC to the base station (BS). We focus on the less studied, but practically important problem of transmitting reliably even on SCs that are fed back by no user or few users. We derive a constrained, non-linear minimum mean square error estimator that enables the BS to estimate the power gains of all the unreported SCs of a user. The novelty of the estimator lies in its exploitation of the structure of feedback generated by the selective feedback scheme. It improves the average cell throughput compared to several conventional approaches without any additional feedback overhead. The improvements occur for both uncorrelated and correlated SCs, and for the channel-aware greedy and fair round-robin schedulers