Fuzzy Logic Design for Packet Switched Network

The recent widespread of packet-switched network leads to better offered services. However, this incurs the cost of introducing additional complexity into existing networks. To combat such pitfall, countless of controlling methods have been proposed. Out of the diverse controlling techniques, fuzzy logic seems to be the favoured choice. In practice, the technique is integrated into various areas of packet-switched network such as congestion control, buffer management, packet routing and etc. This dissertation aimed to demonstrate better routing decision by introducing hybrid metric in TCP/IP routing network. Generally, the hybrid metric is the crossbreed of conventional routing metrics with the aid of Fuzzy Inference System (FIS). Such composite metric is deemed to solve one of the current issues in Internet routing. This dissertation intends to investigate the fuzzy logic-based packet routing in the Internet. The area of routing is limited to the Interior Gateway Protocol (IGP), where only intra-autonomous system (intra-AS) parameters are concerned. The system is developed using MATLAB R2006a to simulate different sizes of routing network within an autonomous system (AS). The network size is determined by the amount of routers and router-to-router links. Design methodology and explanations of simulation results are included in the paper.For route discovery purpose, one of the shortest path algorithms, namely Dijkstra's algorithm is used. For fuzzy system optimisation, trial-and-error approach is chosen due to its simplicity.