A Greedy Ant Colony Optimization for routing in delay tolerant networks

This paper proposes a new prediction-based routing protocol for Delay Tolerant Networks (DTNs) called Greedy Ant (GrAnt). GrAnt uses a greedy transition rule of the Ant Colony Optimization (ACO) metaheuristic to provide the exploitation of good previous solutions, when available, and to forward the messages to the most promising node(s). By making use of pheromone concentration and heuristic function, GrAnt protocol provides three modules: scheduling, routing, and buffer management. Simulation results show that in a community-based movement model GrAnt is able to achieve higher successfully message delivery and lower bytes overhead than Epidemic and PROPHET. In a buffer size of 5 MB, for example, GrAnt delivers almost 79 percent of the messages against 39 percent of messages delivered by PROPHET and 45 percent of messages delivered by Epidemic.