Peak to average power ratio reduction for OFDM system

Abstract

Wireless communication systems at high data rates and low power consumptions are currently in high demand. Power efficiency becomes a critical issue for technology that will be used in 3G and 4G areas. Orthogonal frequency-division multiplexing (OFDM) is a technique using discrete multi-tone modulation with each subcarrier modulating in a conventional modulation scheme. OFDM offers many advantages for multi-carrier transmission at high date rates, particularly in mobile applications. The primary advantage of OFDM over single-carrier schemes is its ability to cope with severe channel conditions without complex equalization filters. However, the high time domain peak-to-average power ratio (PAR) that limits the transmitter power efficiency has been a major obstacle in the implementation of power efficient transmitter. An OFDM system with high PAR requires a power amplifier (PA) with large dynamic range, or alternatively, a perfect linearized saturating PA must back off its maximum output power by approximately the PAR of the input data for distortion-less transmission. While PA is non-linear, additional back-off is necessary. Therefore, the PAR reduction is of viral importance for OFDM systems with power efficient transmitter. In this thesis, we present a new active constellation extension (ACE) based convex optimization algorithm which reduces PAR through convex programming. Interior point methods, also referred to as the barrier methods, are a class of algorithms to solve linear and non-linear convex optimization problems. These algorithms, in contract to other optimization methods, reach the optimal solution by traversing the interior of the feasible region. They consist of self-concordant barrier functions used to encode the convex sets. The ACE constrained IPM (ACE-IPM) shows that the OFDM signal with globally minimum PAR, subject to the constraint on the allowable constellation region, can be efficiently computed using convex programming. Moreover, our method can be implemented by discarding parts of the time-domain signal samples and frequency constellation variables to reduce the complexity with only slight performance degradation. Experimental results show that our proposed algorithm outperforms other ACE-based algorithms with a symbol-error-rate (SER) performance comparable to existing convex programming methods. We also give a convergence analysis of the ACE-IPM and the simplified ACE-IPM as well as the suboptimality bound. Furthermore, the ACE-IPM combined with other techniques are proposed. The first one is non-linear mapping based ACE-IPM (NM-ACE-IPM). For OFDM with low order QAM, the ACE-IPM is very effective. However, the PAR reduction capability of the ACE-IPM will be limited for OFDM with higher order QAM. The NM-ACE-IPM maps the constellation non-linear to provide more variables for PAR reduction. The second one is to use data subcarriers combined with free subcarriers and out-of-band subcarriers to reduce PAR, which gives a general frame work of PAR reduction. For data subcarrier, they need to satisfy ACE constraint so as to guarantee the BER performance. While for free and out-of-band subcarriers, they satisfy spectrum flatness requirement and power spectrum mask constraint, respectively. Finally, three PAR reduction methods, ACE-IPM, error vector magnitude (EVM) constrained IPM (EVM-IPM), and repeated clipping and filtering (RCF) based PAR reduction method (ACE-SGP) are compared and analyzed. Analysis shows that ACE-IPM guarantees the minimum constellation distance, which explains why the ACE-IPM yields better SER performance than the EVM-IPM. On the other hand, the ACE-SGP cannot achieve the optimal solution, so the PAR reduction performance is not satisfactory. In conclusion, we propose a general frame work for OFDM which can achieve better PAR reduction performance than some existing methods with comparable computational complexity.

Date of Award	2 Oct 2008
Original language	English
Awarding Institution	City University of Hong Kong
Supervisor	Shu Hung LEUNG (Supervisor)