Utility based cross-layer collaboration for speech enhancement in wireless acoustic sensor networks

A wireless acoustic sensor network is considered that is used to estimate a desired speech signal that has been corrupted by noise. The application layer of the WASN derives an optimal filter in a linear MMSE sense. A utility function is then used in conjunction with the MMSE estimate in order to evaluate the most significant signal components from each node in the system. The utility values are used as a cross-layer link between the application layer and the network layer so the nodes transmit the signal components that are deemed most relevant to the estimate while adhering to the power constraints of the system. The simulation results show that a high signal-to-error and signal-to-noise ratio is still achievable while transmitting a subset of signal components.