On Reducing the Complexity of Tone-Reservation Based PAPR Reduction Schemes by Compressive Sensing

In this paper, we describe a novel design of a Peak-to-Average-Power-Ratio (PAPR) reducing system, which exploits the relative temporal sparsity of Orthogonal Frequency Division Multiplexed (OFDM) signals to detect the positions and amplitudes of clipped peaks, by partial observation of their frequency content at the receiver. This approach uses recent advances in reconstruction of sparse signals from rank-deficient projections using convex programming collectively known as compressive sensing. Since previous work in the literature has focused on using the reserved tones as spectral support for optimum peak-reducing signals in the time-domain, the complexity at the transmitter was always a problem. In this work, we alternatively use extremely simple peak-reducing signals at the transmitter, then use the reserved tones to detect the peak-reducing signal at the receiver by a convex relaxation of an other-wise combinatorially prohibitive optimization problem. This in effect completely shifts the complexity to the receiver and drastically reduces it from a function of N (the number of subcarriers in the OFDM signal), to a function of m (the number of reserved tones) which is a small subset of N.

I. Introduction

In the last decade, the problem of high peak-to-average-power ratio (PAPR) in OFDM systems has been tackled by a variety of approaches, including coding techniques, constellation reshaping, tone-reservation, and selective mapping, to name a few. Although many of these reduction techniques are brilliant and extremely effective, the main obstacle limiting their actual implementation is commonly related to high complexity.