One of the issues in mobility management is to support handoff. When the mobile user moves from one location to another, the network should ensure that all ongoing connections are rerouted to another access point in a seamless manner. Part of our work focuses on connection rerouting due to inter-switch handoff in wireless ATM networks. Although fast local connection rerouting minimizes handoff delay, the end-to-end path after rerouting may become "suboptimal", which implies an inefficient use of network resources. Path optimization may be necessary afterwards. Our research begins with the following question:​ "How often should path optimization be performed?" To this end, we propose three path optimization schemes (namely:​ exponential, periodic, and Bernoulli), which are simple to implement. Closed-form solutions of the optimal operating point are derived for each scheme.

We further investigate this problem and propose a stochastic model to determine the optimal time to initiate path optimization. Link cost and signaling cost functions are introduced to capture the trade-off between the network resources utilized by a connection and the signaling and processing load incurred on the network. Results indicate that the optimal policy derived from our model has a better performance compared to other heuristics.

Another issue in mobility management is to track the location of the users between call arrivals. Although it has been shown that the distance-based location update algorithm has a better performance than the movement and timer based schemes, the determination of the optimal distance threshold is often based on certain unrealistic assumptions. We propose a stochastic model to study the distance-based update algorithm. Our model is applicable to arbitrary cell topologies and the cell residence time can follow general distributions. We consider Markovian movement patterns in which the probability that the mobile user moves to a particular neighboring cell can depend on the location of the current cell or a list of cells recently visited. Results indicate that the distance thresholds determined from our model have a better performance than those derived from a hexagonal cell configuration with random walk movement pattern.