| Popis: |
Mobile Ad Hoc Networks (MANETs) have wide applications in practice, such as rescue, emergency operations, and battlefield communications. MANETs consist of a group of mobile nodes that dynamically exchange data among themselves without the reliance on any fixed infrastructure or centralized administration. Nevertheless, mobility and self-organizing characteristics of MANETs cause the change of topology in an unpredictable way. Due to the limited transmission range, each node has to seek assistance of its neighboring nodes to help relay the data to transmit to other nodes. The reliable neighboring nodes are important for successful data transmission from a source node to a destination node. The early works providing reliable routing mainly rely on cryptographic tools to prevent malicious nodes from injecting false information into network. However, these reliable routing schemes are always coupled with prerequisites, such as authentication center, key management mechanism, which are not feasible in MANETs. Furthermore, cryptographic operations are considered computationally expensive on resource con-strained mobile nodes. The most important issue is that these schemes are incapable of tracking internal attackers who are authenticated to participate in the network but misbehaved during the routing process, e.g. black hole and grey-hole attackers. Thus, there is a need to look for a lightweight approach to provide a reliable ad hoc routing. In this thesis, we present two trust-based routing schemes to prevent different routing disruption attacks, such as active black hole, passive black hole, and grey-hole nodes. The most attractive feature is that our proposed schemes are not only capable of preventing those well-known routing disruption attacks, but also immune to collusion attack. Even the internal attackers might know how our proposed trust-based security mechanism works; they still cannot attack the system successfully. In the first scheme, we propose a neighborhood connectivity based trust scheme targeting on the active black hole attacks. In this scheme, every node is required to exchange neighbor information before route discovery, and then uses collected neighbor information to verify each received route reply message. By utilizing previously collected neighbor information, we can greatly increase the robustness of the designed system against internal attackers. In the second scheme, a trust-based routing scheme that emulates human cognitive process is proposed. The mobile nodes will exchange trust-level values and evaluate these values based on their own cognitive judgment. Eventually, each node can dynamically evolve its cognition to exclude malicious entities, such as passive black holes and selfish nodes. The distributed nature of our design allows both of the proposed schemes to be easily implemented on any existing routing protocols to handle routing disruption attacks. We have integrated our proposed trust-based routing schemes into dynamic source routing (DSR) protocol and studied their performances under different adversarial network environments. Simulation results affirm that our proposed trust-based routing schemes promote network scalability and ensure the effective routing under the presence of various routing attackers in MANETs. DOCTOR OF PHILOSOPHY (EEE) |
