Disconnection prediction in mobile ad hoc networks for supporting cooperative work

A predictive technique, which serves as the basic layer of an innovative pervasive architecture for cooperative work and activity coordination in mobile ad hoc networks (MANET), was investigated. The predictive algorithm, called as Mobile Gambler's Ruin (MGR) algorithm, is based on the Markov chain model of the well-known gambler's ruin problem. The approach used for the investigation combines local connection management among devices with global management of both network topology and task assignment. The results show that the coordinator applies algorithms for choosing a bridge and/or executes workflow task reassignment when needed.