نشریه مهندسی برق و الکترونیک ایران الکترونیک ایران

Quality of Service (QoS) refers to a set of rules or techniques that help the network administrators use the available network resources optimally to manage the effects of congestion and to treat the applications according to their needs. The differentiated services architecture (DiffServ) allows providing quality of service to users. The major DiffServ premise is that individual flows with similar QoS requirements can be aggregated in larger traffic sets and identified as classes. Relative service differentiation is a simple and easily deployed approach compared to the absolute differentiation service. In this paper, we use fuzzy logic systems to design a novel algorithm for queue management and scheduling of classified packets in the differentiated service IP networks. The proposed model consists of two parts. The first part of the proposed model is used for absolute differentiated services and tries to optimize QoS parameters and to share sources between different requests fairly. The second part of the proposed fuzzy system is dedicated to relative Diffserv model. As one of the famous existing algorithms for both buffer management and scheduling in the relative differentiated service is Jobs algorithm, in the second parts of the proposed system, we modify the traditional Jobs algorithm and proposed a fuzzy based modification of existing Jobs algorithm. The proposed algorithm uses different fuzzy logic controllers to differentiate the delay of traffic classes. Simulation results confirm that the proposed fuzzy system, can provide better delay differentiated than the traditional algorithms.