Tawfik, A. (2018). Compressed sensing for graph signals [Diploma Thesis, Technische Universität Wien]. reposiTUm. https://doi.org/10.34726/hss.2018.47380
In this thesis we are going to tackle the problem of estimating a sparse graph signal with an unknown frequency support set of known size K from a sampled noisy version. Not knowing the location of the non-zero Fourier coefficients and by taking a number M of samples larger than K will give us a compressed sensing problem. The Bayesian Approximate Message Passing (BAMP) algorithm is used to solve the compressed sensing problem and recover the original signal from few coefficients.
