Preliminary results in B-spline enhanced source-tracking
R.A. Green, John W. Pierre · 2002
Source-tracking algorithms typically require estimation of an array's unknown correlation matrix. Time-averaging forms the basis of estimating the correlation matrix in most applications. This paper provides a preliminary investigation in the use of B-splines rather than time-averages to estimate the correlation matrix. The smooth, continuous character of splines can accurately match the behaviour of an array's true correlation matrix. Preliminary results indicate tracking estimates based on the spline-fit correlation matrix improve smoothness over time-averaging methods and can have comparable or better accuracy. Furthermore, B-splines are computationally efficient and are easily adapted to near-real-time processing.