Efficient PAPR reduction through random constellation mapping

In wireless communications research and commercialization, Orthogonal Frequency Division Multiplexing (OFDM) has been established as a key PHY layer enabling technology (e.g., in both 4G and 5G cellular network standards). However, the Peak-to-Average-Ratio (PAPR) problem can lower system performance and has been under intense research in recent years. The trade-off between high PAPR reduction factor and high signal redundancy and/or computational complexity has not been carefully addressed by many works on this problem, and some approaches with state-of-the-art PAPR reduction performance suffer from high bit redundancy. In this paper we propose a novel method for PAPR reduction through random constellation mapping. The proposed method first turns the input symbol into a number of candidate symbols by randomly one-to-one mapping the signal constellation at each position within the symbol. Then the candidate symbol with the lowest PAPR is sent out along with the its candidacy number. This results in very low signal redundancy, and simulation results also show that significant PAPR reduction gains can also be obtained.