—We propose a novel Self-Balancing Supply/Demand (SBSD) reactive MANET protocol. Our algorithm dynamically discovers publications matching specified subscriptions by controlled forwarding of subscriptions and publications throughout the network. We use a utility function to rank subscriptions, balancing subscription demand, age, and replication against the negative congestive effects of excessive replication on the whole network. We show our model gives highly-demanded subscriptions larger spatio-temporal areas within which publication delivery is likely; further, delivery probabilities decline more slowly with subscriber-publisher distance for such subscription types. SBSD is tested under realistic conditions, including: limited broker memory; high mobility; broadcast collisions; device failures; packet loss; and high subscription to publication ratios. We also show its scalability for large, dense populations even when the network arrival rate of new subscriptions increases with br...
Aris M. Ouksel, Doug Lundquist