Analysis of Clipping Noise and Tone-Reservation Algorithms for Peak Reduction in OFDM Systems

Orthogonal frequency division multiplexing (OFDM) suffers from a high peak-to-average power ratio (PAR). Tone reservation is a popular PAR reduction technique that uses a set of reserved tones to design a peak-canceling signal. In a previous paper by Krongold and Jones, an active-set approach was developed to efficiently compute the peak-canceling signal. In this paper, we consider the use of clipping noise, which is generated when the OFDM signal is clipped at a predefined threshold, to design the peak-canceling signal. To this end, the clipping noise is analyzed as a series of parabolic pulses under tone-reservation constraints. The single-pulse case and the multiple-pulse case are treated. The analysis explains peak regrowth and the constancy of the clipping noise power spectrum over the whole OFDM band. Moreover, the clipping noise at the end of several clipping and filtering iterations is shown to be approximately proportional to that generated in the first iteration. The constant of proportionality is estimated via the level-crossing theory for high clipping thresholds. Using this analysis, a constant-scaling algorithm and an adaptive-scaling algorithm are proposed for tone reservation. These algorithms scale the filtered first-iteration clipping noise to compensate for peaks that are above the threshold. The simulation results show that the proposed algorithms achieve a larger peak reduction and lower complexity than the active-set algorithm.

I. Introduction

Orthogonal Frequency Division Multiplexing (OFDM), although used in standards such as IEEE 802.11a/g, IEEE 802.16, high-performance radio LAN version 2, and digital video broadcasting [2], suffers from the high peak-to-average power ratio (PAR) [2]. A large PAR requires a linear high power amplifier (HPA), which is inefficient. Moreover, the combination of an insufficiently linear HPA range and a large PAR leads to in-band distortion and out-of-band radiation [2]. Various PAR reduction techniques have, therefore, been proposed, including clipping and filtering [3]–[6], tone reservation [1], [7], [8], multiple signal representation [9]–[11], and coding [12]–[14]. The clipping and filtering technique causes bit-error-rate (BER) degradation [15]–[17]. Although the degradation is small for high clipping thresholds, clipping noise cancellation techniques are required to lower the BER degradation due to low clipping thresholds [18]–[21].