A Low-Complexity Practical Quantize-and-Forward Scheme for Two-hop Relay Systems Duong T. Tran School of Computer and Communication Sciences ´ Ecole Polytechnique F´ ed´ erale de Lausanne Lausanne 1015, Switzerland Email: duong.tran@epfl.ch Sumei Sun, Ernest Kurniawan Modulation and Coding Department Institute for Infocomm Research, A*STAR 1 Fusionopolis Way #21-01, Singapore 138632 Email: {sunsm,ekurniawan}@i2r.a-star.edu.sg Abstract—We study a low-complexity practical quantize-and- forward (QF) scheme, a special case of the compress-and-forward (CF) transmission for half-duplex relay systems. By adjusting the number of quantization levels, we show that significant performance improvement can be achieved even though the relay node does not exploit any correlation information with the signal received at the destination. We demonstrate the performance of our scheme using 16QAM as the modulation and Turbo Code as the channel coding, and compare its performance against the decode-and-forward (DF) scheme. An approximately 2 dB gain is shown to be achievable at a target error rate of 10 -4 in the scenario where the relay is close to destination. I. I NTRODUCTION Recently, relaying has been utilized as an efficient technique to provide cooperative diversity. The most fundamental model is the three-node cooperative relay channel which includes a source (S), a destination (D) and a relay (R). There are some foundation papers that focused on deriving the capacity bounds of three-node relay channel in full-duplex operating mode [1, 2]. Motivated by practical implementation, Madsen et. al. [3] studied half-duplex relaying and derived the capacity bounds when the relay node operates in a time-division manner. They proved the achievable rate for the two notable relay schemes: decode-and-forward (DF) and compress-and-forward (CF). CF scheme is known to outperform DF scheme when the relay is close to the destination [3]. In this scenario, the S-R link is weak, the relay hence cannot decode the transmitted signal from the source efficiently. While the DF seems to be in vain, the relay node in CF scheme can forward some quantized/compressed knowledge of its received signal to the destination through the strong R-D link to enhance the decoding at the destination. In that way, CF does not suffer as much loss of information at the relay as DF. The existing implementations of CF scheme can be divided into two categories, namely those which utilize the Wyner-Ziv coding [4] to quantize and compress the signal at the relay exploiting the correlation with the signal at the destination [5– 7], and those which ignore this correlation to simplify the relay processing [8]. Our work will adopt the later approach and no side information from the destination is used to compress the estimate (quantized value) of the signal received at the relay Y r1 . We show that as long as the estimation of Y r1 can be recovered reliably at the destination, the decoder of the S D R Phase 1 S D R Phase 2 d 1-d Fig. 1. The three-node relay channel quantize-and-forward (QF) scheme can significantly outper- form the DF. Instead of a complicated Wyner-Ziv coding, the relay processing is therefore simplified to include only a low complexity quantizer, a source encoder and a channel encoder. No requirement for the correlation information between the received signals at the destination and the relay nodes has also eliminated the information exchange over the network, hence making the scheme more practically viable. We demonstrate our proposed QF scheme using bandwidth- efficient 16QAM modulation (4 bits/2 dimensions) instead of BPSK modulation as in most of existing works, and apply the quantization on both in-phase (I) and quadrature (Q) components independently. Turbo Code is used as the channel code to get close to the theoretically achievable performance. Compared to other schemes such as the S → D direct transmission and the DF scheme with the same modulation and channel code, our QF scheme is shown to be superior in terms of the error rate performance. The rest of the paper is organized as follows. Section II introduces the system model. Analysis of the achievable rate is described in Section III. The proposed scheme is presented in Section IV, and in Section V we give the details of the quantization and source coding process. The numerical results are given in Section VI, and finally, concluding remarks are drawn in Section VII. II. THE SYSTEM MODEL We consider a three-node relay channel operating in half- duplex mode as in Fig. 1. In our work, we divide the time equally between the two phases so the ratio of either phase 1 or 978-1-4673-0990-5/12/$31.00 ©2012 IEEE