A Concurrent Ant Colony Optimization Multipath Forwarding Algorithm in IP Networks

The rapid development of communication technology and programming router in the last few years have given rise to a strong research interest in Concurrent Multi-Path (CMP). At the same time, much of the researches of ant colony optimization (ACO) focus on wireless networks such as MANET (Mobile Ad-hoc Networks) and WSN (Wireless Sensor Networks), on the other hand, there are little researches about ACO on wired networks such as Internet. Combining with the advantages of CMP and ACO while based on the disadvantages of conventional single-path OSPF (Open Shortest Path First), ECMP (Equal Cost MultiPath) and random CMP, this paper proposes a novel CMP forwarding algorithm aimed to increase the scalability and decrease the route overhead to existing ant-based routing algorithm. The proposed algorithm, which is called Concurrent ACO CMP Forward (CACF), makes full use of the positive pheromone feedback of forward ants and backward ants to update the state of different paths. Forward ants probe the current condition of the networks and backward ants update the routing information of each path and adjust forwarding granularity for different availability bandwidth. CACF can adapt to the changing of link performance dynamically and utilize multiple paths to transmit simultaneously. We validate CACF algorithm through theoretical analysis and NS2 simulation driven by real-world traffic traces and networks topologies. Results prove that CACF algorithm reduces the degradation of performance of different paths, obtains significant performances of throughput, end-to-end delay and jitter.