THE APPLICATION OF STATISTICAL FILTER DESIGN TO A PARTICULAR PROBLEM
Abstract
A particular solution of the optimum filter design is presented for the class of problems where the noise is normally distributed and the data are described by a normal distribution. This is an application of the general results obtained in previous memos. This particular problem lends itself to an analytic solution and in fact turns out to yield a linear filter. This linear filter has long been used in tracking problems but it is believed that this is the first rigorous derivation. Three examples are given which represent the complete range of characteristics of the filter and one is able from them to make statements about the necessary 'memory' of the filter. One is also able to draw conclusions about how often one should sample a function based upon the irregularity of the function and on the noise that contaminates the measurements.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 21, 1953
- Accession Number
- AD0020805
Entities
People
- W. I. Wells
Organizations
- Massachusetts Institute of Technology