Wide-Sense Stationarity in Generalized Graph Signal Processing

We consider statistical graph signal processing (GSP) in a generalized framework where each vertex of is associated with an element from Hilbert space. This general model encompasses various signals such as the traditional scalar-valued signal, multichannel and discrete- continuous-time signals, allowing us to build unified theory random processes. introduce notion joint wide-sense stationarity this GSP framework, which allows characterize process combination uncorrelated oscillation modes across both space domains. elucidate relationship between notions different domains, derive Wiener filters for denoising completion under framework. Numerical experiments on real synthetic datasets demonstrate utility our approach achieving better estimation performance compared or time-vertex