Date of Award
Doctor of Philosophy in Electrical Engineering - (Ph.D.)
Electrical and Computer Engineering
Alex R. Dytso
Blind source separation (BSS) is performed on frequency hopping (FH) sources. These radio frequency (RF) signals are observed by a uniform linear array (ULA) over a Spatial Channel Model (SCM) in four different propagation environments: (i) line-of-sight (LOS), (ii) single-cluster, (iii) multiple-cluster, and (iv) LOS with interference. The sources are spatially sparse, and their activity is intermittent and assumed to follow a hidden Markov model (HMM). BSS is achieved by utilizing direction of arrival (DOA) of the sources and clusters. A sparse detection framework is applied to obtain estimates of the sources' FH and DOA patterns. The solutions are binned according to a frequency grid and a DOA dictionary. A method is proposed to reduce the effect of falsely detected active sources and mitigate the effects of interference, by leveraging the activity model of the intermittent sources. The proposed method is a state filtering technique, referred to as hidden state filtering (HSF), and is used to improve BSS performance. Multiple activity patterns associated with different DOAs are considered "similar" if they match over a prescribed fraction of the time samples. A method pairing DOA and FH estimates associates the FH patterns to specific sources via their estimated DOAs. Numerical results demonstrate that the proposed algorithm is capable of separating multiple spatially sparse FH sources with intermittent activity, by providing estimates of their FH patterns and DOA.
Ghosh, Anushreya, "Sparse methods for blind source separation of frequency hopping rf sources" (2023). Dissertations. 1662.