Restoring lost data samples with modified low-pass FIR filters
LC Westphal · 1st IASTED International Symposium on Signal Processing and its Applications · 1987
When a sample sequence {x(k)) has been taken from a signal x(t) with a minimum frequency B Hz by sampling at a rate fs > 2B, but occasional samples are missing or unreliable, there is often a desire to reconstruct the missing data points. The usual methods of doing so involve polynomial interpolators of low degree, but, as Schafer and Rablner [1] argued for the related and much-discussed problem of data resampling using interpolation, such interpolators have poorer bandwldths than specialised finite duration impulse response (FIR) filters. Harks [2] described an algorithm for restoring lost samples that in effect digitises an ideal low-pass filter and hence involves infinite sequences of data weighted by sinc( ) functions, but the method requires a large data window for reasonable accuracy.