ENHANCED BIO-INSPIRED ANT COLONY OPTIMIZATION ALGORITHM FOR FAULT-TOLERANT NETWORKS

Abstract

This this presents a summary of introduction, literature review, methodology, research finding, conclusion and recommendations. Today, computer networks form a very important part of institutions and organizations helping them to run their operations swiftly and on time while keeping them up to date with the current changing technologies in the market. The networks are also used by many people to access resources on the internet and also for communication purposes. This has made the world to be indeed a global village. However, networks may be intentionally deterred by the introduction of forced physical loops in the switches, which is common in institutions of higher learning and organizations. As a result of these network loops, packets may be got up in cycles limiting communication process due to network convergence issues. Secondly, the research sought a way of addressing the convergence issues in trying to solve network problems in ACO. To do this, the study aimed to understand the number of artificial ants needed to find an optimal solution in the search space. MMAS, RankedAS and EliteAS were studied for optimal number of ants needed. It was found that the optimal number depended largely on various factors including 1) type of ACO algorithm in use, 2) The complexity of the problem under study, and 3) The ratio of the number of specialized ants to that of normal ants. Lastly, the study aimed at establishing the functionality of current network systems, evaluating network faults, and developing an enhanced model based on the existing ACO model to help solve these network issues. The new model developed suggests ways of solving packet looping and traffic problems in common computer networks. The study used simulation method to carry out research whereby an enhanced algorithm was developed and used to monitor and control the flow of packets over the computer network. The research employed experimental research design that involved the development of a computer model and data was collected from the model. Packet traffic was monitored by the Cisco Packet Tracer tool. In this tool, a network of four computers, a router and two switches was and used to simulate a real network system. Data collected from the simulated network was analyzed using the ping tool, direct observation of the movement of packets in simulation mode and message delivery status displayed by the Cisco Packet Tracer in the real time mode. In the experiment, a control was used to show the behavior of the network in ideal conditions without varying any parameters. Here, all the packets sent were completely and correctly received. Secondly, when a loop was introduced in the network it was found that the network was adversely affected because none of the packets sent by the computers on the network was delivered due to stagnation. In the third experiment, still, with the loops on, a new Enhanced ACO (EACO) model was introduced in the Cisco Packet Tracer used to simulate the network. In this experiment, all the packets sent were completely and correctly delivered just like in the control experiment. In summary, this research found that when conditions of networks are varied for instance introduction of forced physical loops in computer networks, the computers lose communication. This is a common situation in colleges and universities where some students forcefully introduce loops into networks affecting the communication process. However, with a new EACO model, it was found that the problem can be solved where the looping packets can still be rerouted to their destination hence no effect on the communication process. From this research, we can conclude that EACO model can be applied in computer network systems especially dynamic systems having many users to solve the common problem of network loops. However, this research has some challenges in that the algorithm must be run on all computers on the network for optimal results.