Design and FPGA implementation of a new approximation for PAPR reduction

One of the main drawbacks of orthogonal frequency division multiplexing (OFDM) is its sensitivity to the peak-to-average power ratio (PAPR). To address this issue, a novel and efficient scheme based on multilayer perceptron (MLP) neural networks (NN) using radial basis function is proposed and implemented on FPGA. The proposed solution offers good performance as compared to previous algorithms in iterations, much lower complexity and low error rate. For a precise and less complex implementation of radial basis function, a novel approximation of the exponential function using Taylor series is proposed, implemented on FPGA and compared with Altera intellectual property core (IP core). It is shown that, the proposed approximation offers minimal resource consumption. In particular, we note a reduction of 9% and 6% in the use of lookup tables and DSP blocs.