Optimal Fractional Fourier Filtering for Graph Signals

Graph signal processing has recently received considerable attention. Several concepts, tools, and applications in such as filtering, transforming, sampling have been extended to graph processing. One extension is the optimal filtering problem. The minimum mean-squared error estimate of an original can be obtained from its distorted noisy version. However, best separation noise, thus least error, not always achieved ordinary Fourier domain, but rather a fractional domain. In this work, problem for signals domains, theoretical analysis solution proposed are provided along with computational cost considerations. Numerical results presented illustrate benefits domains.