Symbol Error Probability Minimization for Symbol-Level Precoding in Massive MIMO Communications

Symbol-level precoding exploits the symbol constellation structure and transforms multi-user interference into a useful signal by taking into account both channel state information and data symbols, which has recently emerged as a novel paradigm for future wireless communications. The symbol-level precoding highly depends on the modulation and less works consider PAM and QAM constellations. In this paper, we focus on PAM and QAM constellations and propose a symbol-level precoding design to minimize the symbol error probability (SEP). We first analyze the SEP for PAM and QAM symbols and derive an upper bound of SEP. To minimize the SEP for all users, we directly design the transmitted signal, which is a linear combination of channel vectors. In addition, to reduce the complexity brought by the SEP function, we further relax the SEP expression and propose a near-optimal solution of the transmitted signals. Simulation results demonstrate that the proposed scheme achieves superior performance than other existing precoding schemes in terms of SEP.