Abstract/Description
In this paper, we present a variation of Genetic Algorithm (GA) for finding the Optimized shortest path of the network. The algorithm finds the optimal path based on the bandwidth and utilization of the network. The main distinguishing element of this work is the use of ldquo2-point over 1-point crossoverrdquo. The population comprises of all chromosomes (feasible and infeasible). Moreover, it is of variable length, so that the algorithm can perform efficiently in all scenarios. Rankbased selection is used for cross-over operation. Therefore, the best chromosomes crossover and give the most suitable offsprings. If the resulting offsprings are least fitted, they are discarded. Mutation operation is used for maintaining the population diversity. We have also performed various experiments for the population selection. The experiments indicate that random selection method is the most optimum. Hence, the population is selected randomly once the generation is developed. In this paper, we have shown the results using a smaller network; however the work for larger network is in progress.
Keywords
Bandwidth, Biological cells, Telecommunication traffic, Traffic control, Genetic algorithms, Genetic mutations, Decision making, Shortest path problem, Routing, Arithmetic
Session Theme
Artificial Intelligence – II
Session Type
Other
Session Chair
Dr. Sharifullah Khan
Start Date
16-8-2009 12:20 PM
End Date
16-8-2009 12:40 PM
Recommended Citation
Sarfraz, M., Sohail, S., Javed, Y., & Anjum, A. (2009). Artificial Intelligence – II: Network path optimization using GA approach. International Conference on Information and Communication Technologies. Retrieved from https://ir.iba.edu.pk/icict/2009/2009/25
Artificial Intelligence – II: Network path optimization using GA approach
In this paper, we present a variation of Genetic Algorithm (GA) for finding the Optimized shortest path of the network. The algorithm finds the optimal path based on the bandwidth and utilization of the network. The main distinguishing element of this work is the use of ldquo2-point over 1-point crossoverrdquo. The population comprises of all chromosomes (feasible and infeasible). Moreover, it is of variable length, so that the algorithm can perform efficiently in all scenarios. Rankbased selection is used for cross-over operation. Therefore, the best chromosomes crossover and give the most suitable offsprings. If the resulting offsprings are least fitted, they are discarded. Mutation operation is used for maintaining the population diversity. We have also performed various experiments for the population selection. The experiments indicate that random selection method is the most optimum. Hence, the population is selected randomly once the generation is developed. In this paper, we have shown the results using a smaller network; however the work for larger network is in progress.
