We consider the problem of blind adaptive filtering in high- dimension vector spaces with limited data sample support. The focus application is joint space-time adaptive processing for DS/CDMA communications systems with antenna arrays. The objective is the development of fast, low complexity optimization procedures that exhibit superior disturbance (multiple-access- interference and channel noise) suppression characteristics in small data support situations. The low optimization complexity objective implies that matrix inversion and/or eigen decomposition operations are highly undesirable. In this context, the core subject of this investigation is the development and analysis of inductive, conditional optimization procedures in the form of a sequence of weighted auxiliary vectors that are orthonormal to each other and to the joint space-time vector direction of interest.
