Congestion Control for Epidemic Routing in Opportunistic Networks

Flooding-based routing protocols in opportunistic networks like Epidemic provide a simple routing algorithm. While delivery probability is high and delivery latency is low, the generated overhead of distributing every message to each node generates unnecessary load in the network, especially in areas with higher node density and areas with network congestion. Several approaches try to control congestion by utilizing additionally transferred information to coordinate packet transmissions. Due to the nature of an opportunistic network, solutions designed to use a central or global oracle are seldom feasible. Our solution contributes a per-message probability-based decision for congestion control using ephemeral information of a nodes' surroundings. Simulations conducted using the ONE simulator show, that after a certain node density resulting in a congested network area, overhead ratio decreases and delivery probability significantly increases compared to Epidemic routing. By testing scenarios with different movement models, we also foreclose algorithm efficiency based on node interactions.