Distributed channel allocation simulation

Distributed discrete event simulations can be used to solve many real world problems. The obvious benefit in solving large complex problems using parallel versus serial techniques is speed.In this thesis, I will examine a solution to the problem of channel allocation with specific applications to cellular communication or mobile computing. The distributed simulation, based on conservative simulation techniques, utilizes a deadlock avoidance approach. The distributed simulation is performed on a BBN multiprocessor computer. The simulation utilizes seven independent processors which tessellate into the geometric shape of a hexagon. The simulation is a true distributed message passing application whereby a communication layer is used to pass messages between processors.The channel allocation algorithms examined are based on fixed channel allocation using channel borrowing techniques. The channel reorganization algorithms are done to free channels within the network for requesting calls.Empirical results are examined in terms of efficiency of the simulation, grid size and the inclusion of channel reorganization. Results show that the NULL message approach to deadlock avoidance is favorable when the message density to the network is high. At high message densities as little as five percent of the total messages processed are NULL. At low message densities upwards of 90 percent of messages processed are NULL.The fixed channel allocation using channel borrowing techniques along with channel reorganization showed favorable results. By using channel reorganization, upwards of 20 percent of calls that were initially not connected using channel borrowing schemes were connected after reorganization