PAPR Reduction in OFDM Signal Using Machine Learning Base Tone Reservation

The orthogonal frequency division multiplexing (OFDM) is a wireless communication modulation and multiplexing technology that has been widely used in practice due to its salient advantages, i.e., high spectral efficiency and multi-path fading resistance. However, OFDM signals is known to suffer from High Peak-to-Average Power Ratio. PAPR reduction techniques are subject of extensive research. Among them, Tone Reservation technique is considered a viable solution, as it is capable of reducing the PAPR of OFDM signals without scarifying the BER performance; this is achieved by reserving some subcarriers to control the peaks. Assigning appropriate signals to these subcarriers is required to maximize the PAPR reduction performance. In this paper, we apply two machine leaning algorithms, namely Support Vector Machine (SVM) and Artificial Neural Network (ANN) to help determine appropriate peak canceling signals for the Tone Reservation PAPR reduction technique. Numerical results on a relatively small subcarrier OFDM system have shown that both SVM and ANN offer good performance in assigning peak canceling signals as compared to the solution obtained by the exhaustive search, with ANN being a slightly better than SVM.