The Research And Improvement Of Congestion Control Algorithm In Resilient Packet Rings

Resilient Packet Ring (RPR) is a ring based network for high-speed metropolitan area networks which has properties of fault tolerance and high bandwidth utilization. In congestion control of RPR, the issues of fairness, stability, and convergence time are important. In this thesis, we propose an effective Fuzzy Congestion Control (FCC) algorithm to achieve above considerations. FCC generates a precise local fair rate by fuzzy logic to suppress upstream traffic to mitigate congestion and achieve global fairness. FCC is composed of two components: Smooth Fair rate Calculator (SFC) and Fuzzy Fair rate Calculator (FFC). SFC uses the key idea in DBA and applies moving average technique to produces a primary fair rate; FFC first use STQ occupancy and arrival traffic rate to STQ as input to inference fuzzy congestion degree which are some grades of membership for linguistic terms of fuzzy sets, then it adopts the fair rate caculated by SFC as a basis, adjust it according to the fuzzy congestion degree using fuzzy control method to generate a local fair rate which reflects the real traffic condition. Since SFC and fuzzy control method are applied, the smaller convergence time than other fairness algorithms is obtained. Furthermore, even in different scales of congestion domain, FCC still has an excellent performance as compared with other fairness algorithms. Simulation results show that the proposed FCC has an balanced performance on fairness, stability, and convergence time in different testing scenarios. Consequently, we can conclude that the FCC is a feasible and effective design for congestion control in RPR.