I hereby declare that I am the sole author of this thesis. This is a true copy of the thesis, including any required final revisions, as accepted by my examiners. public. I understand that my thesis may be made electronically available to the ii Multi-hop wireless networks based on 802.11 are being used more widely as an alternative technology for last-mile broadband Internet access. Their benefits include ease of deployment and lower cost. Such networks are not without problems. Current research on such networks aims at a number of challenges, including overcoming capacity limitation and poor fairness. The focus of our research is for achieving fairness in multi-channel multi-hop wireless networks. First, we review the literature for different methods for representing link-contention areas, and the existing single-channel fairness computational model. Second, we generalize the fairness constraints applied to each link-contention area, defined in the existing single-channel fairness reference model, to multi-channel models. Third, by adopting the concepts of link-usage matrix and medium-usage matrix to represent network topology and flow status, and using Collision Domain theory and Clique Graph theory to represent link-contention area, we develop a computational model to com-pute optimal MAC-layer bandwidth allocated to each flow in a multi-channel multi-hop WMN. We simulate various network configurations to evaluate the performance of the fairness algorithm based on the above computational iii model in different scenarios. We have found that in the multi-channel envi-ronment, our extension to the Collision Domain model generally provides a more accurate estimation of network capacity. Based on this model, we have extended the source-rate-limiting mechanism, which limits the flow rate to its fair share computed by the computational model. Experimental results that validate these findings are presented in this thesis. iv

Traditional analog video surveillance systems technology has recently become inadequate to face the massive demand of security systems consisting of hundreds and sometimes thousands of cameras often deployed in hostile environments tens of miles far away from the control room. During the last few years, the rapid growth of the digital technology has produced sophisticated cameras which can directly record high-definition digital videos. The packetized video stream can be straightforwardly conveyed to the control room relaying on common IP network infrastructures. This solution results extremely flexi-ble as the network infrastructure can be built over a wide variety of heterogeneous network technologies from the traditional Ethernet-based Local Area Networks (LANs) to the recently proposed Wireless Mesh Networks (WMNs). However, the widespread adoption of IP-based solutions for video surveil-lance poses serious problems in terms of required bandwidth, processing power, network security and system dependability. In this paper, we first investigate the advantages of the IP-based video surveil-lance systems over the traditional analog ones. Then, we describe the technical challenges and the open research issues which still lack an ultimate solution which permits to completely abandon the traditional analog technology. Finally, we propose and verify, by means of a case study, a methodology to address the design of video surveillance systems in real deployment.