Optimal Peak-to-Average Power Ratio Reduction in MIMO-OFDM Systems

Recent work has used convex optimization to minimize the peak-to-average power ratio (PAR) of OFDM signals subject to a constraint on the constellation error vector magnitude (EVM). This paper extends the PAR optimization technique to multiple-input multiple-output (MIMO) OFDM systems with channel precoding. In MIMO systems with a large OFDM symbol size, it is infeasible to solve the optimization problem by direct methods such as Cholesky factorization. Instead, we propose an iterative conjugate-gradient (CG) method to find an approximate solution with far lower memory and latency requirements. Simulation results are presented for a MIMO-OFDM system with 4 antennas and 1024 carriers. The PAR can be reduced from 11.5 dB to 4.3 dB for QPSK with -20 dB EVM, and from 11.5 dB to 5.5 dB for 16-QAM with -30 dB EVM. The tradeoff between PAR reduction and computational complexity is also examined to determine the number of CG iterations needed to reach within 1 dB of the globally optimal solution.