Joint source-channel coding for distributed sensor networks

In a decentralized detection system, a number of peripheral nodes gather observations from an environment and communicate relevant information to a fusion center for collective decision making. In systems employing tetherless transmission between the sensor nodes and the fusion center, the presence of communication channels need to be taken into consideration in the design of local sensor signaling schemes. Motivated by the delay and resource constraints omnipresent in most wireless sensor network applications, we design optimal zero-memory joint source-channel codes (JSCC) for a canonical decentralized detection system. Aimed at minimizing the error probability of the fusion center output, we derive necessary conditions for optimal sensors signaling. The conditions enable us to develop an iterative algorithm to construct optimum distributed encoder at local sensors. The JSCC approach is shown to offer optimal performance compared with various alternatives, including the separate source-quantization and channel-encoding. It also exhibits the inherent adaptivity in resource (bit) allocation in response to varying channel conditions.