Due to mobility and frequent node failure, the topology of a mobile ad hoc network (MANET) is highly dynamic. Routing protocols should adapt to such dynamism, and continue to maintain connection between the source and the destination. A hybrid computational intelligence-based multipath routing algorithm is presented in this chapter. The proposed method employs Hopfield neural network (HNN) as a disjoint path set selection tool for choosing disjoint paths that maximise the network reliability. The parameters of Hopfield model are also optimised by particle swarm optimisation (PSO) algorithm. This method selects disjoint paths in such a way that the network reliability is maximised. For this purpose, each node in the network is equipped with an HNN. Simulation results show that the proposed PSO-optimised HNN-based routing algorithm has better performance as the reliability of multiple paths is increased while the number of algorithm iterations is reduced as compared with the non-optimised HNN multipath routing. In addition, the PSO-optimised HNN-based routing algorithm shows better performance in terms of reliability and number of paths when compared with the backup path set selection (BPS) algorithm.

Chapter Contents:

• 10.1 Introduction
• 10.2 Multipath routing protocols in MANETs
• 10.3 Investigated computational intelligence algorithms
• 10.3.1 Hopfield neural network
• 10.3.2 Particle swarm optimisation algorithm
• 10.4 Proposed approach
• 10.4.1 Model and assumptions
• 10.4.2 Path set reliability computation method
• 10.4.3 Disjoint path set types
• 10.4.4 HNN-based multipath routing
• 10.4.4.1 Route discovery algorithm
• 10.4.4.2 Path set selection by neural network model
• 10.4.5 Path set selection by PSO-optimised HNN
• 10.5 Implementation details and parameters tuning
• 10.5.1 Ad hoc network model
• 10.5.2 Initialisation of HNN model
• 10.5.3 Selecting network parameters by PSO algorithm
• 10.6 Simulation results
• 10.7 Conclusion
• References

Inspec keywords: routing protocols; particle swarm optimisation; telecommunication network topology; mobile ad hoc networks; mobile computing; multipath channels; Hopfield neural nets; telecommunication network reliability

Other keywords: MANET topology; BPS algorithm; hybrid computational intelligence algorithms; routing protocols; mobile ad hoc networks; PSO-optimised HNN-based algorithm; network reliability; disjoint path set selection tool; non-optimised multipath routing; particle swarm optimisation algorithm; Hopfield neural network; backup path set selection algorithm

Subjects: Computer networks and techniques; Optimisation techniques; Neural computing techniques; Communication network design, planning and routing; Optimisation techniques; Mobile radio systems; Reliability; Protocols; Computer communications