An overview of blind signal processing

Summary form only given, as follows. Blind signal processing is an emerging area of research. Mathematically speaking, there are two major problems: (1) from the relation x(t)=As(t), the problem is to determine both the matrix A and the original signal vector s(t) from only the received signal vector x(t), and (2) from the relation x(t)=h(t)⊗s(t), the problem is to determine both the impulse response h(t) and s(t) from only x(t), where ⊗ is the convolution operator. A unique feature of this problem is the characterization of its indeterminacy, and finding conditions so that the solution of the above equation is “unique”. Major application areas are: blind signals separation, blind channel identification, array signal processing, and blind identification of medium with multiple-layer structure