Received January 10, 2016, accepted January 26, 2016, date of publication March 2, 2016, date of current version May 11, 2016. Digital Object Identifier 10.1109/ACCESS.2016.2537739 Distributed Source Coding and Its Applications in Relaying-Based Transmission ABDULAH JEZA ALJOHANI, SOON XIN NG, (Senior Member, IEEE), AND LAJOS HANZO, (Fellow, IEEE) University of Southampton, Southampton SO17 1BJ, U.K. Corresponding author: L. Hanzo ([email protected]) This work was supported in part by the European Research Council within the BeamMeUp Project, in part by the U.K. Engineering and Physical Sciences Research Council under Grant EP/Noo4558/1 and Grant EP/L018659/1, in part by the Saudi Ministry of Higher Education and in part by the Royal Society's Wolfson Research Merit Award. ABSTRACT Distributed source coding (DSC) schemes rely on separate encoding but joint decoding of statistically dependent sources, which exhibit correlation. DSC has numerous promising applications ranging from reduced-complexity handheld video communications to onboard hyperspectral image coding under computational limitations. The concept of separate encoding at the first sight compromises the attainable encoding performance. However, the DSC theory proves that independent encoding can in fact be designed as efficiently as joint encoding, as long as joint decoding is allowed. More specifically, distributed joint source- channel coding (DJSC) is associated with the scenario, where the correlated source signals are transmitted through a noisy channel. In this paper, we present a concise historic background of DSC concerning both its theory and its practical aspects. In addition, a series of turbo trellis-coded modulation (TTCM)-aided DJSC-based cooperative transmission schemes are proposed. DJSC scheme is conceived for the transmission of a pair of correlated sources to a destination node (DN). The first source sequence is TTCM encoded, and then, it is compressed before it is transmitted both over a Rayleigh fading channel, where the second source signal is assumed to be perfectly decoded side-information at the DN for the sake of improving the achievable decoding performance of the first source. The proposed scheme is capable of performing reliable communications for various levels of correlation near to the theoretical Slepian–Wolf/Shannon (SW/S) limit. Pursuing our objective of designing practical DJSC schemes, we further extended the above-mentioned arrangement to a more realistic cooperative communication system, where the pair of correlated sources are transmitted to a DN with the aid of a relay node (RN). Explicitly, the pair of correlated source sequences are TTCM encoded and compressed before transmission over a Rayleigh fading multiple access channel, where our proposed scheme is capable of operating within 0:55 dB from the SW/S limit for a correlation coefficient value of ρ D 0:8, and within 0:07 bits of the minimum SW compression rate. The RN transmits both users' signal with the aid of a powerful superposition modulation technique that judiciously allocates the transmit power between the two signals. The correlation is beneficially exploited at both the RN and the DN using our powerful iterative joint decoder, which is optimized using extrinsic information transfer characteristics charts. INDEX TERMS Distributed joint source coding, distributed source coding, Slepian-Wolf Coding, joint source-channel decoding, TTCM, superposition modulation. I. INTRODUCTION both satellites, which are referred to as STEREO Ahead Let us commence by considering the stereographic images and STEREO Behind, is shown in Fig. 2. On the 20th of of the Sun in Fig. 1, which were captured using a pair September 2013 they were more than 50 million km away of satellites that are part of NASA's Solar Terrestrial from each other. In such a scenario, is it not readily fea- Relations Observatory (STEREO) project.1 The location of sible for them to communicate with each other. Similarly, each satellite has very limited communications with planet 1The STEREO project aims for providing revolutionary stereoscopic imaging of the sun in order to reveal the solar surface activities, such as the Earth [1], [3]. Thus, the employment of source compression Coronal Mass Ejection (CME) [1], [2]. is desirable, but the encoding of images has to be carried 2169-3536 2016 IEEE. Translations and content mining are permitted for academic research only. 1940 Personal use is also permitted, but republication/redistribution requires IEEE permission. VOLUME 4, 2016 See http://www.ieee.org/publications_standards/publications/rights/index.html for more information. A. J. Aljohani et al.: DSC and Its Applications in Relaying-Based Transmission FIGURE 1. Stereographic solar images taking by NASA's both STEREO satellites Ahead and Behind c NASA [1]. (a) STEREO Ahead solar image. (b) STEREO Behind solar image. FIGURE 3. Decoded stereographic images of Fig. 1 after transmission over uncorrelated Rayleigh fading channels at SNR D 1:0 dB. (a) Ahead view, DJSC of Sec. V-B. (b) Behind view, DJSC of Sec. V-B. (c) Ahead view, non-DJSC benchmark of Sec. V-B. (d) Behind view, non-DJSC benchmark of Sec. V-B. attaining perfect images recovery. By contrast, for the same amount of transmit power, both the Ahead and Behind images were severely corrupted upon using exactly the same coding FIGURE 2. Positions of both STEREO satellites Ahead and Behind scheme dispensing with joint decoding. Thus, exploiting the recorded on the 20th September 2013 c NASA [1]. correlation between the images by jointly decoding them has lead to a significant power reduction, while maintaining reliable communication. This room for improvement can be out separately at the satellites, while decoding may be car- also utilised for further source sequence compression, as it is ried out jointly at the Earth station, where both the power going to be illustrated in the subsequent sections. and computational constraints are relaxed. Intuitively, sepa- rate encoding would only allow separate compression of the II. DISTRIBUTED SOURCE CODING images of the distant satellites even though there is substantial DSC refers to the problem of compressing several phys- correlation between their images. However, the Distributed ically separated, but correlated sources, which are unable Source Coding (DSC) theorem of [4] states that separate to communicate with each other by exploiting that the encoding may be invoked instead of joint encoding without receiver can perform joint decoding of the encoded any loss of compression efficiency as long as the correlation signals [3], [4], [7]–[9]. However, Distributed Joint Source- among the sources is preserved throughout their transmission Channel coding (DJSC) is specific to the case, when the to the receiver, provided that they are jointly decoded [4]–[6]. correlated sources signals are transmitted over noisy chan- To offer a glimpse on the benefits of applying the DSC nels [8], [10]–[17]. More explicitly, a single channel code technique, we have separately encoded both satellite views of is employed for both source compression (via Slepian-Wolf Fig. 1 before their transmission over a uncorrelated Rayleigh Coding (SWC)) and channel error protection. Typically, the 2 fading channel. As Fig. 3 illustrates, when joint decoding is channel code in DJSC schemes is jointly designed to per- activated by invoking our decoder of Sec. V-B.1, no higher form both source compression as well as error protection. than 1:0 dB Signal-to-Noise Ratio (SNR) is required for Intuitively, the joint source-channel coding approaches would be less powerful than their separate counterparts [10], [11]. 2Both images were encoded separately using 1=2-rate Turbo However, this deficiency could be compensated through Trellis-Coded Modulation (TTCM) and decoded jointly using the decoder of Sec. V-B which is illustrated in Fig. 26, readers can have a quick look at exploiting the correlation between the sources at the joint Table 8 for main simulations parameters summary. decoder [10], [11]. VOLUME 4, 2016 1941 A. J. Aljohani et al.: DSC and Its Applications in Relaying-Based Transmission For both DSC and DJSC schemes the ultimate goal is to exploit the existing correlation for the sake of minimis- ing the transmission energy required by the sources, while maintaining reliable communication. From an architectural perspective, distributed techniques may be categorised into two main families [3], [10], [18], namely the class of oper- ating in the presence of perfect side-information and in the absence of perfect side-information schemes. The schematic of the former [3], [9], [18] is shown in Fig. 4, where the source sequence fb1g is compressed before its transmission, f g FIGURE 5. Schematic diagram of dispensing with perfect while the correlated source signal b2 is assumed to be side-information DSC. flawlessly available at the decoder, but not at the source fb1g. By contrast, in the scenario dispensing with perfect side- information both sources are compressed at a rate lower than their corresponding entropy rates, in which any point between the locations A and B of Fig. 7 can be reached. A special case of the latter scenario, when both users are compressed at the same rate, represented by point C in Fig. 7. The encoder has to compress fb1g without knowing fb2g, yet the decoder is capable of exploiting the knowledge of fb2g for recovering fb1g. FIGURE 4. Schematic diagram of relying on perfect side-information DSC. The rest of the paper is organised as seen in Fig. 6. Explicitly, the main principles as well as the historical back- ground concerning both the theory and practice of DSC are presented in Sec. II. Then, our source correlation model is illustrated in Sec. III. Next, in Sec. IV we discuss our DJSC scheme under the idealized simplifying assumption of having FIGURE 6. Paper structure. perfect side information, before conceiving our cooperative DJSC scheme in Sec. V. Finally, we conclude our discourse in Sec.