This research involves the study of communications systems consisting of multiple sources transmitting correlated information. The final objective is to exploit the correlation among the sources to reduce the total transmitted energy. The techniques developed in this research may have a broad impact in the area of sensor networks, making it possible to develop sensors requiring less transmitted power. For example, a possible applications could be the development of less-power-invasive biosensors.
In order to reduce the total transmitted energy, the research will focus on the application of turbo-like codes, and specifically of linear codes with low-density generator matrix to i) achieve compression rates close to theoretical limits for a wide range of multi-terminal correlated sources (discrete or continuous, with and without memory) and ii) perform joint source-channel coding for a wide range of multi-terminal correlated sources and different channel environments. The proposed system is characterized by its simplicity at the encoder site. This simplicity facilitates the implementation of the proposed scheme in practical applications. Moreover, the encoding process is completely independent for each source, and it does not require knowledge about the correlation model or source statistics, which is of great interest in practical applications such as sensor networks and image/video compression. In order to achieve a performance close to the theoretical limits for source and joint source-channel coding, the statistics of the sources are estimated and exploited at the decoder in an adaptive fashion.