No abstract

No abstract

Quality of service is an important issue in Delay Tolerant MANETs. This work aims at increasing the QoS in such networks by relying on spanning forests (DA-GRS). The existing algorithms are improved by introducing the notion of trust into the spanning forest and choosing the most robust (trustable) spanning trees among existing opportunities. The robustness/quality of the tree can be assessed based on two cost functions. Two heuristics are proposed 'G-TRUST' (a greedy-based heuristic) and 'G-TRUST BREAK' (including an automatic tree reconfiguration heuristic) and evaluated.

Abstract-Delay Tolerant Networks (DTNs) are an extension of Mobile Ad-Hoc Networks (MANETs). Global knowledge in DTNs cannot be obtained or guaranteed due to their dynamicity, decentralized nature and non-permanent structure. Managing such networks optimally is very difficult, if not impossible. Trust management in such networks receives much attention recently due to their potential application. One solution for managing information within DTNs lies in constructing and maintaining spanning forests. DA-GRS is a local computation based model for the description of decentralized algorithms designed for dynamically distributed environments like Delay-Tolerant MANETs (DTMs). DA-GRS proposes a framework for constructing and maintaining a spanning forest in such an environment. This work introduces the notion of trust into DA-GRS resulting in T-DA-GRS algorithm. The goal of the proposed algorithms is to construct and maintain robust trusted spanning tree where less trustable nodes are leaves. Three cost functions are suggested as means to assess the robustness of trusted spanning trees. T-DA-GRS is also further improved by incorporating greedy algorithm to become T-GDA-GRS. These algorithms were tested with four different networks generated by a DTM simulator known as Madhoc. Efficiency of these algorithms is compared with optimal values.

No abstract