Group-ordered SPRT for distributed detection

We consider the problem of distributed detection in a large wireless sensor network. An adaptive data fusion scheme, group-ordered sequential probability ratio test (GO-SPRT), is proposed. This scheme groups sensors according to the informativeness of their data. Fusion center collects sensor data sequentially, starting from the most informative data and terminates the process when the target performance is reached. To analyze the average sample number, we establish the asymptotic equivalence between GO-SPRT, a multinomial experiment, and a normal experiment. Closed-form approximates are obtained. Our analysis and simulations show that, compared with fixed sample size test and traditional sequential probability ratio test (SPRT), the proposed scheme achieves significant savings in the cost of data fusion.