Description
TitleRobustness in ad hoc networks
Date Created2017
Other Date2017-10 (degree)
Extent1 online resource (xv, 138 p. : ill.)
DescriptionInvestigating the resilience of wireless network is critical since a network is susceptible to many types disturbances, ranging from natural (e.g. wind, building, hill, mobility) to adversarial (e.g. jamming, eavesdropping, denial of service). The notion of network resilience describes a network's ability to recover its structure should damages occur when one or more terminals are shut down or whenever sensitive information is compromised due to impersonating adversaries. This thesis addresses the following three problems related to the reduction of robustness in an ad hoc network: (1) The backtracking problem in overlay networking, which occurs during packet delivery. This condition forces a packet to traverse an extended distance due to the lack of network-layer information being fully shared with the overlay routing functions. With the presence of backtracking, the packet delivery error rate increases in relation with the traversal distance; (2) The black hole problem, which is characterized by the network being punctured, such as is caused by a jammer emitting format-compliant packets in an attempt to overflow a legitimate node's buffer. As a consequence, targeted nodes are segregated from the rest of network; (3) The reduced connectivity problem, which results from a repetitive jamming attack where the weakest link or node is selected by the jammer to be attacked. In this thesis, the first problem is solved by redesigning an overlay routing protocol to be aware of the geographical location of cluster heads. The overlay hash table is re-organized in an order based on distance. The second problem is solved in two different ways: First, a method is presented that locates the jammer's approximate position. Reinforcement learning is applied at each node to learn the possible position of the jammer. The node's judgment is propagated by the routing protocol to interactively exchange the belief about the attacker's position. Second, a power control approach that is being used by wireless nodes is presented that adjusts the transmission power dynamically according to the Fiedler value, which is inferred from topology information carried inherently in the Optimized Link State Routing (OLSR) Protocol. By varying the radio range, packet flows can bypass the jamming area after topology transformation, making the network more resilient. The third problem is approached by developing a better mathematical understanding of connectivity when facing interference, which is achieved by analyzing the achievable throughput based on connectivity. A metric that measures the throughput and weighted throughput connectivity is introduced. Then, the connectivity issue under a jamming attack is solved by applying stochastic game theory to analyze the utility of connectivity and the interactive behaviors of both the jammer and the scanner. An optimal scanning strategy is designed to defend against repetitive jamming attacks involving different intents, and to strengthen the network connectivity at the same time.
NotePh.D.
NoteIncludes bibliographical references
Noteby Ying Liu
Genretheses, ETD doctoral
Languageeng
CollectionSchool of Graduate Studies Electronic Theses and Dissertations
Organization NameRutgers, The State University of New Jersey
RightsThe author owns the copyright to this work.