Aggregation and compression of distributed binary decisions in a wireless sensor network

An aggregation and compression architecture for wireless sensor networks without a fusion center is proposed. In this architecture, sensors are grouped into pairs whose local binary decisions are combined to form half as many new one-bit decisions. The new decisions are paired and combined until a single one-bit global decision is reached. A simple recursive algorithm is presented and proved to find the minimum size of the network required to achieve a given global error probability.