MAAS: Multibit Adaptive Aggregation Scheme for wireless sensor networks

The problem of data aggregation for binary event detection using wireless sensor networks has been considered in the present work. We propose a novel multibit adaptive aggregation scheme (MAAS) which combines the merits of both multibit and adaptive aggregation schemes. Firstly, being adaptive, the proposed scheme does not require knowledge of the performance indices of sensor nodes as required by the existing optimum decision fusion rule. Therefore, in case of small variations in the topology or performance indices, the proposed scheme needs to just fine tune the weights starting from the previously adapted weights, instead of seeking new values of the parameters or recomputing the initial setup. Secondly, being a multibit aggregation scheme, the performance of the proposed MAAS is higher than the existing single-bit aggregation schemes at a cost of communicating only a few extra bits. This makes it suitable for event detection applications where detection accuracy is more important than the energy drained in communicating a few additional bits. Further, the performance of the proposed MAAS is shown to be consistently superior to single-bit schemes even in the presence of Rayleigh fading channel.