Signal fusion network for spatial diversity MIMO radar

This paper concerns target detection algorithms of spatial diversity MIMO radar. The GLRT algorithm is developed, under the assumption that clutters in different spatial diversity channels follow independent Gaussian distributions with different covariance matrix. The global test statistics turns out to be a production of the local test statistics. Three signal fusion networks are introduced, in which each local site can detect targets with the CFAR property. The CFAR thresholds for the fusion centre and the local sensors can be estimated by offline numerical experiments. It is shown that the signal fusion network yields a better detection performance than that of the decision fusion network.