A Fuzzy LSP Regulator for Preemption Control in a DiffServ-Aware MPLS Internet N. M Din & H.Z. Abidin Department of Electrical Engineering, College of Engineering, Universiti Tenaga Nasional, KM 7, Jalan Kajang-Puchong, 43009 Kajang, Selangor, Malaysia {norashidah, husna)@uniten. edu. my Abstract This paper describes a proposal for a fuzzy logic LSP regulator for preemption control in a DiffServ- aware MPLS Internet network based on the Russian Doll bandwidth allocation scheme. The fuzzy logic regulator regulates the incoming traffic based on class by either rerouting or resizing existing less prioritized traffic or rejecting the admission request. Three DiffServ traffic class types were used in the DiffServ-aware MPLS network under study, i.e. premium, assured and best effort traffic. The fuzzy logic regulator uses the premium traffic utilization value, the assured and best effort traffic amount of borrowed bandwidth information as inputs and the preemption decision as output. A simulation study was performed for the fuzzy logic regulator using ns-2. A comparative analysis was conducted with preemption, no preemption and with fuzzy logic regulator preemption. The results show that the fuzzy logic pre-emption scheme gave better performance in terms of connection blocking probability and throughput. Index Terms- DifJServ, Fu-y Logic, MPLS, Preemption UINTRODUCTION The DiffServ architecture [1] combined with MPLS [2] has become a promising mechanism of achieving QoS with the scalability required by the Internet. MPLS achieves fast forvarding of packets through its connection-oriented framework. The MPLS edge router assigns packets with labels. Label switched routers in the MPLS core network direct the packets based on their labels. Labels have local significance, and because of this their actual values can change as a packet traverses an MPLS network. However, the packet's path through the network is completely determined by the initial label the ingress LSR assigns it. This path is known as label switched path (LSP). Each distinct flow traversing the LSP is known as a Forwarding Equivalent Class (FEC). The FEC of a packet can be determined by a number of parameters, i.e. DiffServ Code Point, IP protocol ID, Source and/or destination IP addresses or netwvork addresses. MPLS imposes a connection-oriented framework on an IP-based S.F. A. Rahman & N. Fisal Telekom s Laboratory Faculty of Electrical Engineering Universiti Teknologi Malaysia 813 10 Skudai, Johor Bahru Johor, Malaysia intemet and thus provides the foundations for reliable traffic contracts allowing traffic engineering to take place. What can be seen from the DiffServ and MPLS architectures are that they both provide support for service differentiation which complement each other. The PHBs in DiffServ and FECs in MPLS can be mapped to provide QoS.This paper describes work on a fuzzy logic LSP regulator for preemption decision making in a DiffServ- aware MPLS domain. Section II describes related work in literature whereas Section III describes the fuzzy logic regulator. Section IV discusses the simulation work and Section V the conclusion. II.RELATED WORK A.Fu zy Logic in Traffic Contr-ol Fuzzy logic-based traffic control had been proposed by several earlier works mainly addressing ATM networks, such as in [3, 4, 5]. In [3], a congestion management scheme to support ATM's Available Bit Rate traffic to utilize available bandwidth without causing congestion was proposed. A fuzzy inference engine was used to determine the explicit rate for the source of an ATM network whereas a neural fuzzy mechanism is used for generating fuzzy rules for predicting future cell loss. Work by [4] proposed an admission control mechanism that uses a type-2 fuzzy logic system where a group decision making comprising 30 subject matter experts is captured into the inference mechanism instead of taking from only one expert. A type-2 fuzzy set is characterized by a fuzzy membership function, i.e. the membership value for each element of a set is a fuzzy set in [0,1] unlike a type-I where the membership value is a crisp value in [0,1]. Research conducted by [5] uses fuzzy logic to implement admission control and usage parameter control (UPC) for an ATM switch. A fuzzy switch controller for queue size management used in admission control was introduced. Three types of traffic flow were accommodated, i.e. voice, video and data. The fuwzzy input was source transmission rate for each of the source type and the output is the queue 1-4244-0000-7/05/$20.00 ©2005 IEEE. 415