An improved tone reservation method for PAPR reduction in OFDM systems

Orthogonal frequency division multiplexing (OFDM) systems often suffer from a high peak-to-average power ratio (PAPR), which is essentially due to the phase shift and superposition in the IFFT process. Tone reservation (TR) is a popular PAR reduction technique that uses a set of reserved tones to generate a peak cancelling signal. The usage of the time domain kernel signal has been proposed for TR. However, so far the methods based on the kernel signal have not produced sufficiently good PAPR reduction performance for the lack of overall optimization considerations. In this paper we propose a new method that utilizes the least squares approximation (LSA) and optimal time domain kernel signal based on variance minimization in order to generate the optimal peak cancelling signal. Simulation results show that the proposed approach has lower complexity and better performance than typical other kernel-based approaches such as clipping control method [14] and fast convergence LSA-TR scheme [15].