Robust Estimation and Filtering in the Presence of Unknown but Bounded Noise
In this paper optimal algorithms for robust estimation and filtering are constructed.
No statistical assumption is supposed available or used and the noise is considered a deterministic variable unknown but bounded belonging to a set described by a norm. Previous results obtained for complete (one-to-one) and approximate information  are now extended to partial and approximate information. This information seems useful in dealing with dynamic systems not completely identifiable and/or with two different sources of noise, for example process and measurement noise. For different norms characterizing the noise, optimal algorithms (in a min-max sense) are shown. In particular for Hilbert norms a linear optimal algorithm is the well-known minimum variance estimator. For 1₀ ₀ and 1₁ norms optimal algorithms, computable by linear programming, are presented. Applications to time series prediction and parameter estimation of nonidentifiable dynamic systems are shown. State estimation is formalized in the context of the general theory. Assuming an exponential smoothing of the bounds of the noise it is proved that, for stable systems, the uncertainty of the state is aymptotically bounded. Then the results of the previous sections provide computable algorithms for this problem. Two application examples are shown: Leontief models and Markov chains.
- CUCS-223A-86.pdf application/pdf 745 KB Download File
More About This Work
- Academic Units
- Computer Science
- Department of Computer Science, Columbia University
- Columbia University Computer Science Technical Reports, CUCS-223-86
- Published Here
- August 7, 2013