Detection of Spatially Modulated Signals via RLS: Theoretical Bounds and\n Applications

This paper characterizes the performance of massive multiuser spatial\nmodulation MIMO systems, when a regularized form of the least-squares method is\nused for detection. For a generic distortion function and right unitarily\ninvariant channel matrices, the per-antenna transmit rate and the asymptotic\ndistortion achieved by this class of detectors is derived. Invoking an\nasymptotic characterization, we address two particular applications. Namely, we\nderive the error rate achieved by the computationally-intractable optimal\nBayesian detector, and we propose an efficient approach to tune a LASSO-type\ndetector. We further validate our derivations through various numerical\nexperiments.\n