Low complexity bandwidth compression mappings for sensor networks

Compressive (2 : 1) joint source-channel coding using direct mappings from source to channel symbol space is considered. To enable the use of prior information due to e.g. correlated samples at the receiver, or statistical knowledge of the source, minimum mean square error decoding is considered. The prior information is incorporated in the form of the a-priori distribution in the decoding. Four mapping methods are presented and evaluated using the generic Bayesian minimum mean square error estimator. The schemes are evaluated for transmitting a memoryless Gaussian source over additive white Gaussian noise channel with a quadratic distortion measure. The simplicity of implementation and applicability to a wider variety of sources is discussed.