Link longevity Kalman-estimator for ad hoc networks

In ad hoc networks, links are established on the fly as mobile nodes move in and out of each other's transmission range. Due to this mobility, routes can be disrupted while in service. Availability of a good estimation of link longevity between neighboring nodes could permit the selection of a more stable route, thus enabling a better enforcement of quality of service (QoS) contracts. In this paper, we propose three link longevity estimators that could be embedded into mobile nodes. In our approach, link longevity estimates do not require knowledge of the mobility patterns/models of nodes. The foundation of all three designs lies in extended Kalman filtering, where a linear process model is implemented to represent the state transition, while a non-linear measurement model is included to account for the received signal strength indication measurements. We provide with the mathematical background for our estimators and their input/output vectors, and show various performance metrics using extensive simulations. We conclude that our filters provide good estimates for the remaining up-time of wireless links.