Blind system identification using multiple sensors

This paper presents a study of blind system identification using measurements of multiple sensors. The mathematical model we consider here consists of multiple FIR channels driven by an unknown common source. We first show an orthogonal complement system matrix which is very useful in understanding the multi-channel system and developing efficient identification techniques. We then show a fast maximum likelihood method, its relation to a cross-relation method, and a new method utilizing a minimum noise subspace. We also report a study of strict identifiability of the multi-channel system and its relation to existing concepts