Dissertation Defense, Gang Xiong

Title: Distributed Synchronization and Spectrum Sensing in Cognitive Radio Networks


As an emerging and promising technology, cognitive radio has been recently proposed to alleviate spectrum scarcity by allowing unlicensed (secondary) users to coexist with licensed (primary) users while not causing harmful interference. In this work, we study two important components in constructing cognitive radio networks: distributed time synchronization and cooperative spectrum sensing.

First, we focus on the task of synchronizing distributed cognitive radios to the same timing reference, so that they may effectively communicate over a common control channel and conduct network tasks, e.g., cooperative spectrum sensing, distributed spectrum allocation, etc.. Although presented here in the context of cognitive radio network formation, distributed timing synchronization is critical in all distributed network scenarios. In this dissertation, we propose a novel discrete time second- and high-order distributed consensus time synchronization (DCTS) algorithm for ad hoc networks and examine their convergence properties. We claim that the optimal convergence rate of the second- and high-order DCTS algorithm is superior to that of the first-order DCTS algorithm under an appropriate algorithm design. Furthermore, we extend our study on the convergence of the DCTS algorithm when both deterministic and uncertain time delays impact local pair-wise time information exchange. Specifically, we model random delay between secondary users using a Gaussian approximation and determine the resulting asymptotic behavior of global synchronization error.

In the second topic, we study cooperative spectrum sensing in cost constrained cognitive radio networks with a centralized fusion center. Specifically, we examine the case when cognitive radios forward local spectrum statistic to the fusion center over two channel scenarios: parallel access channel (PAC) and multiple access channel (MAC). For both channel scenarios, we aim to maximize the global detection probability of available spectrum subject to a system level cost constraint. (1) In PAC scenario, our objective is to choose appropriate number of energy samples that must be collected at each secondary user and appropriate amplifier gain that each secondary user must use to forward its statistics to the fusion center. When jointly designing these two parameters, we demonstrate that only one secondary user needs to be active, i.e., collecting local energy samples and transmitting energy statistic to fusion center. (2) In MAC scenario, our objective is to choose appropriate beamforming weights subject to a global transmit power constraint. Under correlated lognormal shadowing, we derive closed-form expressions of optimal beamforming weights and claim that global detection probability increases as the number of secondary users increases for a simplified linear array network.

Committee Member:

Prof. Shalinee Kishore (Advisor)
Prof. Tiffany Jing Li
Prof. Parv Venkitasubramaniam
Prof. Aylin Yener (Pennsylvania State University)