On the value of information in system identification - Bounded noise case

i Abstract-Assu.ming. inst~nta.neous bounds on the noise: sysistic. A practical example is the truncation noise in te~ par.ameter Identification ~s formu.lated as membership set digital signal processing. Note that the distributestimation problem. Sequential algorIthms are constructed to. . ... . estimate the membership sets of the parameters which are Ion of the nOise, even In this simple case, IS " cons.istent with the measurements and the noise. cons~raint~. debatable. Another important example of Th.e.lmpo~tant new feature of the propose~ algorIthms IS t~elr significant practical importance is the problem of ability to Ignore redundant data. The efficient data extraction property of the new algorithms is achieved with small fault diagnosis, where the 'noise' corresponds to -' computational effort and with impro.ved performance when fluctuation of the component values around their compar~d to the leas.t squa.re al~on~~m.. The convergen~e nominal values (Saeks and Cambell 1979. Salka propertIes and the notion of Identifiability In the set theoretic. ' , context are also studied. and Tnck, 1979). The approach of this paper is to find a membership set in the parameter space which is compat