Source and channel coding for homogeneous sensor networks with partial cooperation

A sensor networks in which two nodes communicate a remote measurement to an access point is investigated (CEO problem). The focus is on assessing the performance advantages of cooperative encoding strategies as enabled by out-of-band and finite-capacity communication links between the sensors. The analysis assumes Gaussian source and observation noises, a quadratic (MSE) distortion metric and homogeneous sensors. With rate-constrained links between each sensor and the access point, an achievable rate-distortion trade-off is derived that reduces to known rate-distortion characterizations in the cases without and with perfect cooperation. This result is then extended to a Gaussian multiple access channel scenario by deriving achievable distortions with separate or joint source-channel coding. It is concluded that, for both scenarios, even modest values of the capacity of the inter-sensor links enable the optimal performance with full sensor cooperation to be approached.