Cooperative transmission over relay networks holds great promise for improving the network performance to meet the needs for a wide range of applications, including sensor, vehicle and energy-efficient networks. Performance gains through cooperation have been well studied, and there is a gamut great amount of results demonstrating the coverage, diversity, multiplexing and/or energy gains of cooperative transmissions. In parallel, cooperation has also been studied extensively under the guise of distributed data compression, where correlation among signals at distant terminals is exploited to reduce the required compression rates. The main focus of this paper is to exploit cooperation in both domains, via the novel technique of cooperative joint source-channel coding.
Today sensors and sensor networks are becoming pervasive, users ubiquitously take photos and record videos at events, share and exchange these over social networks, and thousand of users interact through virtual reality worlds. The data generated in all these applications are highly correlated, yet current network architectures have no means to exploit this correlation. Cooperative transmission schemes generally focus on the channel coding and networking aspects, and correlation among the data is largely ignored. In this paper, the authors have outlined a joint source-channel cooperative transmission scheme for large relay networks, in order to exploit correlated side information at the relay terminals together with the classical channel cooperation gains.
In large relay networks, exploiting the availability of correlated data at the relay nodes is challenging. The quality of the side information at the relays can be different, and is not necessarily aligned with the quality of their received signals. In this paper, the authors have considered the decode-and-forward scheme, in which the source signal is decoded by all the cooperating relay terminals. The authors have introduced two different source-channel cooperation schemes, based on sliding-window decoding and backward decoding, respectively. While the two achieve the same performance in terms of the source samples that can be transmitted per channel use, i.e., the source-channel code rate, they have different trade-offs in terms of delay and complexity. While backward decoding relies on separate source and channel coding/decoding, sliding-window decoding scheme uses joint decoding.