Independent Component Analysis Assisted Efficient PAPR Reduction and Symbol Recovery in OFDM Systems Zhongqiang Luo, Lidong Zhu, Chengjie Li National Key Lab. of Science and Technology on Communications University of Electronic Science and Technology of China CHINA No.2006, Xiyuan Ave, West Hi-Tech Zone, 611731,Chengdu, Sichuan, P.R.China Email: luozhongqiang@126.com Abstract: In order to coordinate the performance loss and PAPR reduction due to the effect of the conventional PAPR reduction scheme, a new peak-to-average power ratio (PAPR) reduction and symbol recovery scheme in OFDM system is proposed assisted by precoding and independent component analysis (ICA). The proposed scheme is formulated as the model of the mixture of the mutual statistical independence of subcarrier’s signals of OFDM, which can use ICA for strengthening the signal recovery capabilities. Simulation results illustrate that the precoding in transmitter combined with ICA based blind equalization not only provides PAPR reduction role, but also has a better bit error rate (BER) performance, which makes a good trade off between bandwidth and energy requirement. Key-Words: independent component analysis; orthogonal frequency division multiplexing; peak-to-average power ratio; blind equalization 1 Introduction Orthogonal frequency division multiplexing (OFDM) is an attractive multicarrier technique for high-bit-rate transmission [1]. By dividing wideband frequency selective fading channel into parallel narrowband flat fading sub-channels, OFDM can effectively offer high spectral efficiency and high power efficiency, immune to the multipath delay and frequency selective fading. Due to these merits, OFDM system has various applications including the digital audio, digital TV, and broadband satellite communication, and so on. Particularly, a lot of wireless standards (IEEE 802.11a, LTE and Wi-Max) have adopted OFDM technology as a method to improve wireless communication in the future [1, 2]. However, one of the main drawbacks of OFDM systems is the high peak-to-average power ratio (PAPR) of OFDM signals [1-3]. The transmit signals in OFDM system can have high peak values in the time domain since subcarrier components are added via an inverse fast Fourier transformation (IFFT) operation. In this case the high PAPR problem is caused. The high PAPR is one of the most detrimental aspects in OFDM system as it decreases the signal-to-quantization noise ratio (SQNR) of the analog-digital convertor (ADC) and digital-analog convertor (DAC) while degrading the efficiency of the power amplifier in the transmitter. This gives rise to non-linear distortion which changes the superposition of the signal spectrum resulting in performance degradation. So far, several techniques have been proposed to reduce the PAPR of an OFDM signal, including clipping and filtering, selected mapping (SLM), partial transmit sequences (PTS), tone reservation, tone injection and Reed-Muller codes, etc. These techniques can be organized into three classes [1, 2]: signal distortion, signal scrambling, and block coding. The signal distortion is the simplest class of techniques to reduce the PAPR, including clipping and peak windows, and peak cancellation, etc. These techniques reduce peaks directly by distorting the signal prior to amplification. To clip the signal, the peak amplitude is limited to some desired maximum level. It can give a good PAPR, but at the expense of some performance degradation, including in-band and out-of-band interference. The second class of techniques is signal scrambing, including SLM and PTS technique. The basic idea of signal scrambling is to introduce some limited redundancy, and to send the OFDM signal with the minimum PAPR. The goal is not to eliminate the peaks, but only to achieve lower probabilistic occurrence of peak values. However, for the SLM and PTS, to recover the data, the used side information must be transmitted to the receiver, resulting in some loss of band efficiency. The third class of techniques is block coding, including typically m-sequences, complementary Golay sequences, and Reed-Muller codes. This class of techniques limits the set of possible signals that can be transmitted, and only those signals with peak amplitude below some threshold are chosen. WSEAS TRANSACTIONS on COMMUNICATIONS Zhongqiang Luo, Lidong Zhu, Chengjie Li E-ISSN: 2224-2864 204 Volume 14, 2015