Load balancing and congestion control using fuzzy logic control method in computer networks

In this thesis, we use fuzzy logic control to improve computer networks performance by balancing loads among computers and controlling the congestion status for computer networks. We build a new model and devise sender-initiated and receiver-initiated protocols that use fuzzy logic control for load transfer. We compare the system performances for the systems using sender-initiated and receiver-initiated protocols respectively. The simulation results show that the performances of our protocols are better than a conventional method and that these protocols are more stable than the method without using the fuzzy logic control. The system using sender-initiated protocol with fuzzy logic control exhibits a better performance than the system using receiver-initiated protocol.; Since a network may be congested because of too many request messages, we introduce a new implicit congestion control method for loading balancing in computer networks. Our approach uses fuzzy logic control to reduce network congestion by changing membership functions of each node when applying load-balancing algorithms. We use the new model to device a node locating service protocol that uses fuzzy logic control for load transfer and investigate the problems of load balancing in the context of network congestion. Our simulation results show that the packet drop rate is significantly reduced when the new congestion method is used. This suggests that network congestion can be improved when the new congestion mechanism is applied.; In this research, we also design a load service protocol in wireless ad hoc networks. Because hosts in an ad hoc network moves dynamically, it is very important for a wireless host to find other hosts for load transferring purpose. Our protocol uses the database concept in directory agents to save the information provided by load-server agents and dynamically finds available hosts for load services when hosts moves to other regions. We also use fuzzy logic control concepts to transfer loads instead of fixed threshold levels.