Algoritmo distribuido para el control de congestión con asignación explícita de ancho de banda y QoS

Abstract

Abstract: It presents: In distributed environments, such as computer networks, managing and optimizing network traffic is a major challenge. In these environments, network resources are limited and shared by a large number of users. Congestion can easily occur if traffic is not properly managed. The Distributed Utility Function Algorithm (D-AFU), which is a significant advance in the management and optimization of network traffic in distributed environments. Derived from the utility function principle, the D-AFU adapts resource allocation according to network demands and conditions dynamically, ensuring optimal performance and enhanced user experience. Unlike its centralized predecessor, the D-AFU operates under a distributed scheme, enabling more efficient and robust scalability. In addition, this distributed approach results in improved resilience to failures and overloads, for performance evaluation of the D AFU algorithm, tests are performed in different scenarios with the NS-3 simulator. To determine the effectiveness of the algorithm in these contexts, three key metrics were established: the allocated transmission rate, the utility achieved per session, and the level of fairness, which was measured through the Gini coefficient. The results of the simulations indicated a remarkable superiority of D-AFU, versus a scheme without congestion control and obtaining similar results to centralized AFU that served as the basis for our algorithm, especially in real-time applications, which are critical in terms of QoS and latency, showed optimized performance with the use of the D-AFU algorithm in all situations analyzed. This observation supports the proposal of D-AFU as a viable and efficient solution to manage and improve data transmission for applications in complex networks.