Abstract
It is demonstrated that when the system to be identified is blockwise decomposable, additional advantages can be obtained by blockwise identification with recursive least-squares (RLS) identification algorithms. The following aspects are considered: (a) reduced computational burden; (b) lessened memory storage; (c) relaxed severity of the richness requirement on the exogenous input signals. The price paid for all these is the additional number of sensors used to tap the intermediate system states.