Path Prediction for Resource Reservation Between Mobile Nodes

End-to-end QoS can be achieved through resource reservations along the communication path between end points. Per flow or per aggregate signaling can be used to set up reservation states in routers or domains on the path. In a mobile environment this reservation process can be very cumbersome, given that (i) the time to set up the resources in the new path is not neglectable and (ii) the resources may not be available during the handover process. Advance reservations is a possible solution for this problem. This paper describes a procedure to compute a set of considered paths in a resource reservation, during a communication session when both hosts are mobile. We depart from a semi Markov user mobility model extracted from regular user movements. We combine both user models to generate a time evolution equation that allows computing the probability to be in a future path from a current path in a given time. We use this function to build an optimisation problem in a binary integer programming formulation. A target probability of lacking resources is used as a constraint to select a subset of paths based on a cost metric.