Relay Switching Aided Turbo Coded Hybrid-ARQ for Correlated Fading Channel Hoang Anh Ngo, Robert G. Maunder and Lajos Hanzo School of ECS, University of Southampton, SO17 1BJ, United Kingdom. Email: {han08r, rm, lh}@ecs.soton.ac.uk, http://www-mobile.ecs.soton.ac.uk Abstract—Hybrid-Automatic-Repeat-reQuest (HARQ) has be- come an indispensable technique in reliable communications systems. However, its performance is inevitably affected by the channel’s fading correlation. In this paper, we proposed a novel relay-switching aided HARQ scheme in order to mitigate the detrimental effects of correlated fading without unduly increasing the system’s complexity and delay. Our results show that the proposed relay-switching regime operates efficiently in correlated channels, hence significantly reduces the error floor of turbo- coded HARQ. Additionally, a HARQ scheme using Segment Selective Repeat (SSR) is incorporated in the relay-switching scheme for achieving further improvements. Quantitatively, the proposed relay-switching aided turbo-coded HARQ scheme using SSR may achieve an approximately 2 dB gain, compared to the conventional amplify-and-forward aided turbo coded HARQ arrangement using Chase Combining. Index Terms - Relay switching, correlated fading channel, Hybrid-ARQ, turbo codes, chase combining, incremental redun- dancy, selective segment repeat. I. I NTRODUCTION Automatic Repeat reQuest (ARQ) is a powerful technique of mitigating errors incurred in packet data services. In wireless communications the transmit signal is subjected to fading effects, which corrupt the received signal. Therefore, ARQ is typically combined with powerful channel coding schemes, such as turbo codes [1] or Low Density Parity Check (LDPC) codes [2], in order to improve the robustness of data trans- mission over fading channels. The resultant technique, also known as Hybrid-ARQ (HARQ), was introduced in the 1960s by Wozencraft and Horstein [3] [4], which relies on a combi- nation of both error detection and error correction combined with retransmission requests. Their system is now known as Type-I HARQ. Naturally, the combination of Forward- Error-Correction (FEC) codes and the classic ARQ protocol is capable of improving the achievable throughput and of reducing the number of retransmissions, hence the delay. An improved version of this system, known as the Type-II HARQ, was invented by Lin and Yu [5], which is widely employed in contemporary communication systems, such as the Universal Mobile Telecommunications System (UMTS) and the 3GPP Long Term Evolution (LTE) standards [6] or in the IEEE 802.16 mobile WiMAX system [7]. The transmitter of the classic Type-I HARQ scheme typ- ically retransmits all the information and parity bits of cor- rupted packets, when a negative acknowledgement (ACK) is received, while the receiver simply drops erroneous pack- ets [8]. In the HARQ Type II scheme, the information received during the consecutive transmission attempts are combined be- fore each repeated decoding attempt. Clearly, Type-II HARQ is capable of outperforming Type-I HARQ, at the cost of a higher complexity. The channel characteristics substantially affect the attain- able system performance. For instance, correlated fading may corrupt consecutive retransmissions, especially when using Chase Combining, since the transmitted replica may also experience a deep fade. In this scenario the spatial diversity gain of multiple-input-multiple-output (MIMO) systems may mitigate this problem. However, employing the classic co- located MIMO elements at the mobile station, which has compact physical dimensions, may also become ineffective in the presence of spatially correlated fading. To overcome this potential drawback, distributed MIMOs relying on cooperative relaying are proposed. We exploit the flexibility of the cooper- ative networks [9], [10] by advocating a novel relay-switching regime, where the specific relay activated is changed after each transmission attempt, in order to overcome the spatial correlation effects. Additionally, we combine the proposed relay-switching regime with HARQ relying on the Segment Selective Repeat (SSR) technique of [11], where not all, but only the most error-infested segments are retransmitted, in order to further improve the overall performance of the system. The paper is organized as follows. In Section II, the pro- posed relay-switching aided ARQ protocol is described, along with its capacity expressions. Section III introduces the relay- switching regime into the Turbo Coded HARQ (TC-HARQ) arrangementconsidered in conjunction with different diversity- combining techniques. Section IV provides our numerical results, followed by our concluding remarks. II. RELAY SWITCHING AIDED ARQ A. Channel Model We will consider an ARQ scenario, where the source station (SS) broadcast its data to both the relay station (RS) and the destination station (DS) in the first time slot and then the RS amplifies and forwards the data to the DS. It is assumed that the Source-to-Relay (SR), Source-to-Destination (SD) and Relay-to-Destination (RD) channels suffer from correlated fading, but there is no correlation among the three channels, owing to their substantial physical separation. The signal received at the RS may be expressed as y R [i]= G SR h SR [i]x[i]+ n SR [i], (1) IEEE WCNC 2011 - PHY 978-1-61284-253-0/11/$26.00 ©2011 IEEE 477