Finite-alphabet BER-oriented optimization for non-uniform DGSM

Develop a rigorous BER-oriented evaluation of the non-uniform activation aided distributed generalized spatial modulation scheme using finite-alphabet mutual-information analysis and jointly optimize the activation probabilities, power allocation, and precoding matrix.

Background

The paper optimizes activation probabilities and per-stream power allocation by maximizing an upper bound on mutual information under Gaussian signaling. The authors note that assessing symbol-detection reliability and bit-error-rate performance requires moving beyond this Gaussian-input upper-bound analysis to finite-alphabet mutual information.

The unresolved extension requires a joint treatment of activation probabilities, power allocation, and the precoding matrix, rather than optimizing only the activation probabilities and power allocation. The authors explicitly defer this concrete evaluation and optimization to future work.