Abstract
The paper presents a technique for designing a self-optimizing filter to combine noisy measurements of the same physical quantity as measured by different instruments into a single time-series representative of the measured quantity (signal). The procedure minimizes the output power (sum of the squared amplitudes) of a filter that is run over the data, subject to a set of constraints on filter coefficients to characterize signal information. Practical applications of such a data-adaptive weighting algorithm are also illustrated.
