Thesis (Ph.D.)--Chulalongkorn University, 2001

This dissertation is concerned with designing a comprehensive strategy for supporting the ABR traffic class in ATM networks. The ABR service class has been defined for reliable data service support in high-speed networks. Since data traffic is relatively delay tolerant, the use of feedback control schemes is appropriate. Moreover, it is necessary that the proposed algorithm provides fair resource distribution among competing connections and exhibits good scalability to many different network scenarios. Although a number of ABR congestion control proposals have appeared, some important and critical issues still require further investigation. Some of these include operation in dynamic environments, containment of transient congestion effects and provision of MCR guarantees. From our investigation, performance of the well-known explicit rate congestion avoidance algorithms-the ERICA+ (Explicit Rate Indication Congestion Avoidance) and E-FMMRA (Enhanced Fast Max Min Rate Allocation)-depend significantly upon algorithm parameters selected which are affected by network conditions. The network conditions include a large number of traversing ABR sessions, network scenarios, traffic characteristics and system propagation delay. This limits both algorithms from working in the real environments that the network conditions change continually. Consequently, there is still a need for designing improved algorithm that addresses the above issues effectively. The proposed FRACA (Fast Rate Allocation Congestion Avoidance) algorithm is designed to address limitations of current rate allocation algorithms. The proposed algorithm complies with the ATM Forum guidelines and implements the MCR plus equal share bandwidth fairness criterion. Simulation results indicate that the algorithm performs well in many scenarios. Scalability is good and queue sizes are also properly controlled. The performance also tolerates to continually change of the network conditions. Hence, the proposed algorithm should be appropriate to be implemented in the ATM switches working under real conditions. In addition, our proposed FRACA algorithm significantly improves the performance of point-to-multipoint ABR connections. As the result, the performance of a simple consolidation algorithm over the proposed FRACA algorithm is acceptable. The convergence behavior of the FRACA algorithm is also analyzed mathematically. The analytical results indicate the the algorithm converges to the max-min fairness for the case that all sources are persistent.