Research On ADMM Method For PAPR Reduction Of OFDM Signals

Orthogonal frequency-division multiplexing(OFDM) technique is widely used in wireless communication systems with its high bandwidth efficiency, powerful ability to resist the effects of multi-path fading and simple implement. However, one major drawback of OFDM signals is that its time domain waveforms are with high peak-to-average power ratio(PAPR), which limits the development of OFDM techniques, for high PAPR will make the linear range requirements to D/A(Digital-to-Analog converter) and high power amplifier strict. Besides, it will also result in the phenomenon of low power consumption, high costs for mobile networks and nonlinear distortion for signals. So it's a key topic to research on PAPR reduction of OFDM signals.Recently, convex optimization techniques have been exploited to reduce PAPR values of OFDM signals. Though they enjoy better PAPR reduction performance in comparison with traditional approaches, high computational complexity limits their wider applications in practice. As a result, it is necessary to find a way with lower computational complexity that can reduce the PAPR effectively at the same time.In this paper, we focus on this issue of high PAPR and try to propose low complexity convex optimization method for PAPR reduction of OFDM signals. The main contributions of this paper are as follows:1) Three convex optimization models involving ‘clipping' constraints are formulated. By using convex relaxation technique, we relax the PAPR constraint which is non-convex to convex clipping constraint, which is a kind of linear approach and is easy to implement.2) The customized solving algorithms are developed for the models via alternating direction method of multipliers(ADMM) technique, which is gradually mature in big data producing and machine learning areas. ADMM can transform the problems into several sub-problems based on different variables, which can be solved in parallel and separately with simpler analytic solution, so it is easier to solve the original problem.3) We show that their computational costs in each iteration of the customized solving algorithms for these models are dominated by fast Fourier transform(FFT).So they have low computational complexity of(7)(8)22l Nlog l N. Moreover, since FFT twiddle factor matrices related to OFDM signals are orthogonal, the proposed ADMM solving algorithm can be performed distributedly.4) At last, the simulation results of our approach and the comparison with other existing methods demonstrate that the proposed ADMM approaches reduce the PAPR of OFDM signals effectively with much lower computational complexity.