Analog distributed source-channel coding using sinusoids

In a wireless sensor network, it is likely that the measurements of the sensors are correlated. Distributed source coding can be used to reduce transmission rate or mitigate the effects of the channel noise in the case of analog transmission. In this paper, we propose a novel scheme for implementing distributed source-channel coding based on analog mappings. We assume that an analog source is to be transmitted to a receiver that has access to correlated side information, as in the Wyner-Ziv problem. From the Cramer-Rao lower bound, we observe general properties of analog distributed source-channel mappings. It is especially clear how the stretch factor influences the performance. From this observation we propose two different mappings based on sinusoidal waveforms. The proposed transmission scheme is numerically evaluated and shown to perform well, particularly in the low-SNR regime. Furthermore, it requires no encoding or decoding delay, making it suitable for delay-critical applications in wireless sensor networks.