In this paper the sampled signal reconstruction problem is formulated and solved as the sampled-data H2 smoothing problem. Both infinite (non-causal reconstructor) and finite (reconstructor with relaxed causality) preview cases are considered. The optimal reconstructors are in the form of the cascade of a discrete-time smoother and a generalized hold (interpolator). In the particular case of reconstructing polynomial signals with infinite preview, the proposed procedure recovers the cardinal B-spline reconstructors.
|Publisher||IEEE Signal Processing Society|