Master of Science in Electrical Engineering (MSEE)
Electrical and Computer Engineering
The recently developed paradigm of cognitive radio wireless devices has been developed with the goal of achieving more customizable and efficient spectrum utilization of commonly used wireless frequency bands. The primary focus of such spectrum utilization approaches has been to discern occupancies and vacancies over portions of the wireless spectrum without necessarily identifying how specific radio frequency (RF) devices contribute to the temporal dynamics of these occupancy patterns within the spectrum. The aim of this thesis is to utilize a hidden semi-Markov model (HSMM) statistical analysis to infer the individual occupancy patterns of specific users from wireless RF observation traces. It is proposed that the HSMM approach for RF device characterization over time may act as a first step towards performing a more complete characterization of the RF spectrum in which the inferred traffic patterns may demonstrate the coexistence of multiple networks, the specific devices comprising each distinct network, and the level of mutual interference between the component networks resultant from such coexistence. The first main portion of this thesis is the development of a Bayesian learning framework for HSMM characterization of the wireless RF observations, with occupancy periods and each individual RF device being classified as distinct states in the HSMM. The traditional HSMM approach is supplemented with the concept of the hierarchical Dirichlet random process to achieve a minimal number of states needed to effectively capture each distinct device, without the need for strong a priori assumptions regarding the number of devices seen in the RF trace prior to computational analysis. The second portion of the thesis utilizes user-programmed cognitive radios to construct a real-time software-defined RF network environment emulation testbed to assess the accuracy of the HSMM characterization. Finally, the HSMM algorithm is tested on wireless devices operating under an actual implementation of the ubiquitous IEEE 802.11 wireless standard.
Document Availability at the Time of Submission
Release the entire work immediately for access worldwide.
Carroll, Benjamin Elliott, "Learning and identification of wireless network internode dynamics using software defined radio" (2013). LSU Master's Theses. 2509.