Realizing Quality-Of-Service With Link Lifetime Prediction In Mobile Ad Hoc Networks
Recent research in mobile ad hoc networks (MANETs) has been studying the feasibility of provisioning the Quality-of-Service in such a network. One major factor that makes this a difficult task is the node mobility, which induces a dynamic network topology and makes a multi-hop path susceptible to abrupt breakage when any of its constituent links breaks. It is therefore desirable to acquire predictive knowledge of the path lifetime, which reflects its reliability and is closely associated with the lifetime of each constituent link. We first study how node mobility impacts the lifetime of a path in the MANET, and propose three algorithms that employ link age as the decision parameter to select the best path from all available paths between the sender (source) and recipient (destination) of data for transmissions. Performance evaluation results show that these path-selection algorithms achieve a robust performance in choosing a path that meets a specified path-lifetime requirement. We next propose the Mobile-projected Trajectory (MPT) algorithm that employs four periodically measured distances between two nodes of a link to compute the relative movement trajectory and the remaining lifetime of a link. This algorithm is based on linear curve fitting, and does not require any knowledge of node position, speed, and direction. To account for the frequent occurrences of velocity change during the link lifetime, we propose a simple velocity-change detection test that is augmented to the MPT. The new algorithm, Mobile-Projected Trajectory with Velocity Change Detection, improves the performance of the algorithm by recomputing the remaining link lifetime more accurately after a velocity change is detected. We also propose a novel link lifetime prediction algorithm based on the Unscented Kalman Filter, which casts the link lifetime as a non-linear dynamic system model, and recursively estimates its states to compute the remaining link lifetime. Performance evaluation of the proposed algorithm demonstrates robust performance to estimate the link lifetime while nodes move along various trajectories induced by velocity changes.
dissertation or thesis