From the network management approach, the term network efficiency signifies the effective utilization of network resources. The critical aspect of managing the Multi-Protocol Label Switch (MPLS) networks is to compute the best routes across the network that guarantees the cohesive traffic flow with the effective use of network resources. Considering the optimal routes in multiple switching based MPLS networks, comprised of multiple domains, serves as a complex and challenging task. Technically, the computation of optimal routes that can be depending on multiple objectives along with constraints introduces the concept of multi-objective subject to multiple constraints (MCOP) optimization problem in the field of optimization, which is considered as a computationally complex optimization problem.Metaheuristic optimization algorithms have raised as a mainstream approach for solving MCOP based complex optimization problems. However, metaheuristic algorithms can generate sub-optimal solutions because rooted problems within algorithms that badly disturbs the algorithm`s performance. Therefore, extensive research on the improvement of algorithms has become necessary. This thesis investigates the particle swarm optimization (PSO), bat, and dolphin echolocation (DEA) algorithms, highlights the problems in the algorithms and offers novel versions of these algorithms as a proposed methodology for the MPLS optimization problem.For MPLS optimization, the offers the MCOP based optimization models which consist of multiple objective functions and are mathematically formulated for experimental setups. For the considered optimization problems, the new metaheuristic algorithms are suggested as the modified and hybrid versions of PSO, Bat, and DEA algorithms. The numbers of experiments are conducted along with extensive results analysis, which demonstrates the performance of presented algorithms for MPLS optimization, and to validate these algorithm performances, an exclusive comparative analysis is established with other familiar metaheuristic algorithms.
|Date of Award||28 Jul 2020|
- University Of Strathclyde
|Sponsors||University of Strathclyde|
|Supervisor||Ivan Glesk (Supervisor) & Ivan Andonovic (Supervisor)|