Skip to Main Content
Linear transmitter designs that minimize the MSE or the uncoded BER do not guarantee good bit error performance when any channel correction code is introduced and the QoS is measured in terms of packet-error rate (PER). This paper presents a robust power allocation strategy that maximizes the cut-off rate using partial Channel State Information (CSI) at the transmitter. The cut-off rate has been considered for a long time as the practical bound on the performance that could be achieved by feasible codes, providing an exponential upper bound on the error probability for the average over the ensemble of codes with block length N. A Bayesian formulation is used to design an optimal transmitter based on noisy CSI estimates. The paper analyzes the cost function, comparing the results with the algorithms that minimize the uncoded BER, minimize the MSE and maximize the mutual information.