Game Theory Based Load Balancing Algorithms Over Multiple Virtual Tunnels

Load Balancing (LB) of data traffic over several channels connecting two gateway network nodes is a common and important issue in many networks and communication scenarios. The LB operations can be useful especially in the case of wireless networks when an Access Point (AP) has heterogenous connectivity and the access links have capacity limitations. The paper proposes several Game Theory (GT) based load balancing algorithms (auction based and selfish routing based algorithms) for distributing the traffic received by an AP node over several virtual tunnels instantiated in a heterogeneous network. The proposed LB algorithms were evaluated by computer simulations and were compared with "classical" LB algorithms. The performed simulations show that the auction based algorithms have the best performance in terms of delay suffered by the transmitted data packets, while the selfish routing based algorithm has weak performance, even behaving poorly than some of the non-GT based LB algorithms.