Optimizing a Single-Hop Virtual Path System Using a Genetic Algorithm
Surat Tanterdtid, Watit Benjapolakul, W. Steanputtanagul · International Journal of Computers and Applications · 2000
The virtual paths (VPs) system has been proposed as a key technique in order to utilize the asynchronous transfer mode (ATM)-based transport network. This article addresses the problem of how to assign the VPs system-based ATM network. This problem can be classified as a combinatorial and NP-hard complete optimization problem in which we want to minimize the designed VPs system assignment objective subject to the network constraints. A genetic algorithm (GA) is introduced as a new approach to solve this problem. Using the ON-OFF source and the fluid flow concepts, the equivalent bandwidth is formulated as the effective transmission rate of each heterogeneous traffic VP which guarantees both quality of service (QoS) in the cell level and the grade of service (GoS) in the connection level. The multi-objective function which comprises the average blocking rates (ABR) and the average packet delay (APD) is regarded as our designed VPs system assignment objective. A new coding technique in which each chromosome represents one of the VP system configurations and is also created associating with the designing delay bound of each VP is proposed. The concept of past solution is modified to suit the dynamic VPs system control technique that requires fast calculation speed. Numerical results show that by using our proposed method, it gives a good solution. Faster calculation speed than offered in previous literature, can be obtained by using our new coding technique. Moreover, comparative studies between this Genetic Algorithm (GA) application and conventional exhaustive search method which has been used as a reference algorithm to evaluate the quality of solution for global optimum provides very similar results.