د.ابوعجيلة المصري دغمان2026-06-072026-06-07https://dspace.academy.edu.ly/handle/123456789/2169This thesis develops an artificial intelligence (AI)-augmented comparative framework for evaluating the performance of Multiprotocol Label Switching (MPLS) and Software-Defined Wide Area Networks (SD-WAN) in multimedia transmission contexts, with a focus on Quality of Service (QoS) metrics relevant to voice and video applications. A virtual simulation environment was established in GNS3 to model MPLS and SD-WAN architectures under controlled conditions. Using UDP-based traffic generators, the experiment captured three core performance indicators—delay, jitter, and packet loss—across both voice and video traffic streams. The comparative analysis employed AI techniques—fuzzy logic, linear regression, random forest, and multi-layer perceptron neural networks—alongside Euclidean distance metrics to quantify performance variations and model non-linear QoS relationships. Results demonstrate that SD-WAN provides superior jitter control and greater packet stability for voice transmissions, whereas MPLS exhibits marginally lower delay in specific video streaming scenarios. Under network stress conditions, SD-WAN demonstrates more consistent QoS and temporal stability, making it particularly effective for delay- and jitter-sensitive applications such as video conferencing and Voice over IP. The proposed framework yields practical guidelines for mitigating packet loss in SD-WAN environments and optimising delay and jitter management in MPLS infrastructures. Furthermore, the integration of AI-driven predictive monitoring enhances proactive network management. This research advances AI-enabled QoS assessment methodologies and provides practical guidance for infrastructure planning aligned with contemporary multimedia service demands. Keywords: Multiprotocol Label Switching (MPLS), Software-Defined Wide Area Network (SD-WAN), Quality of Service (QoS), Multimedia Traffic, Artificial Intelligence (AI).This thesis develops an artificial intelligence (AI)-augmented comparative framework for evaluating the performance of Multiprotocol Label Switching (MPLS) and Software-Defined Wide Area Networks (SD-WAN) in multimedia transmission contexts, with a focus on Quality of Service (QoS) metrics relevant to voice and video applications. A virtual simulation environment was established in GNS3 to model MPLS and SD-WAN architectures under controlled conditions. Using UDP-based traffic generators, the experiment captured three core performance indicators—delay, jitter, and packet loss—across both voice and video traffic streams. The comparative analysis employed AI techniques—fuzzy logic, linear regression, random forest, and multi-layer perceptron neural networks—alongside Euclidean distance metrics to quantify performance variations and model non-linear QoS relationships. Results demonstrate that SD-WAN provides superior jitter control and greater packet stability for voice transmissions, whereas MPLS exhibits marginally lower delay in specific video streaming scenarios. Under network stress conditions, SD-WAN demonstrates more consistent QoS and temporal stability, making it particularly effective for delay- and jitter-sensitive applications such as video conferencing and Voice over IP. The proposed framework yields practical guidelines for mitigating packet loss in SD-WAN environments and optimising delay and jitter management in MPLS infrastructures. Furthermore, the integration of AI-driven predictive monitoring enhances proactive network management. This research advances AI-enabled QoS assessment methodologies and provides practical guidance for infrastructure planning aligned with contemporary multimedia service demands. Keywords: Multiprotocol Label Switching (MPLS), Software-Defined Wide Area Network (SD-WAN), Quality of Service (QoS), Multimedia Traffic, Artificial Intelligence (AI).هندسة كهربائية