The rapid deployment of the Mobile Ad Hoc Networks features topologies that are frequently associated with highly variable network density and mobility conditions. For these networks, the provision of fundamentalnetworking functions (such as self-organization or routing) becomes a difficult task, because the network nodes must take decisions based on several sources of context, related to the local environment and to the
characteristics of individual neighbor nodes (e.g., their position, velocity, energy resources, cooperation status, etc).
The thesis fills this gap, by introducing quantities expressed in terms of time, that appropriately “translate” the environmental conditions and nodal characteristics (which may be appropriately adjusted, to also reflect cooperation or other social aspects) in a form suitable for taking the decision in question. To accommodate this, the thesis develops a novel and suitable notion in the context of routing, which is the “retaining time”, namely the time that a node holds a message before forwarding it. This notion of time has been successfully exploited towards dynamically self-adjustable routing in diverse mobile topologies by the proposed MAD
(Maximum Advance Decision) protocol.
The retaining time is a particular instance of the more general notion of “Decision-Related Event Occurrence Time” (DREOT), which refers to the time duration up to the occurrence of an event linked to the decision. The
thesis contributes novel probabilistic reasoning methodology and results for DREOT calculations, which are applicable beyond a particular protocol or mobility model. Furthermore, these abstract DREOT-related techniques are employed towards model-based calculations for the value of the retaining time.
Another important complementary issue is the operational aspect of the signaling required for a realistic implementation of the routing protocol. Towards this end, the thesis studies relevant on-demand beaconing
techniques and suggesting a generic analysis and policies for information exchange between the involved nodes.
Finally, all these aspects are compared and verified through a rich set of numerical and simulation results. These demonstrate the efficient and effective nature of the proposed techniques and notions and provide
additional useful insights and guidelines as to how the decision policy of different operations can self-adapt in diverse mobile environments.