Data-Driven Passive Localization With Non-Cooperative Radiation Sources via Mutually Inverse Networks

This letter proposes a passive localization method with non-cooperative radiation sources and distributed receivers. In this letter, a pair of mutually inverse networks is put forward. The inverse network aims to generate more labeled data, and the original network is trained with the generated data to learn the mapping from the observed data to the target location. Experimentally, the results show the proposed method can achieve better performance with one-sixteenth the amount of training data than the network in same structure. The proposed method has the potential in outdoor passive localization with communication base stations.