As a predominant technique to estimate the cause and effect of Quality of Service (QoS) criteria in computer networks, simulations provide insight into how to most efficiently configure protocols to maximize the usage and to estimate the criteria for acceptable performance of network applications. This paper investigates the simulation of Genetic Algorithm-based network routing using Open Shortest Path First (OSPF) protocols. The simulator is implemented using LEDA (Library of Efficient Data types and Algorithms) [1] and the applicability of the library is examined in the context of Genetic Algorithms. OSPF is the most widely used routing protocol. OSPF's popularity comes from its scalability, efficiency, and its self-sufficient nature. In supporting hierarchies and autonomous areas, OSPF allows for networks to grow and still be easily maintainable. By allowing the administrators to set the cost value for a network link between two routers, OSPF can intelligently route packets along links the administrators deem to be better. By broadcasting Link-State Advertisements (LSAs), routers can be added and removed, and OSPF would regenerate the routing tables automatically to reflect the changes in the network. OSPF was designed specifically to use Dijkstra's Shortest Path First algorithm. Dijkstra's algorithm determines the shortest routes to all of the routers in the area using the LSA database. By adjusting the costs (weights) of the links one can alter the routes that Dijkstra's algorithm will calculate.
展开▼
