Proactive Congestion Management in Next-Generation Wireless Networks: Evaluating Traditional Approaches Against Machine Learning Models
John Justin Thangaraj John Justin Thangaraj, T. Mohanraj, S Caleb, J Srija, R Latha, Pushpalatha Rajendran · 2024
sensible congestion control approaches continue to be imperative in attaining the highest capability of the next-generation facilities for wireless networks. To this end, the objective of the present research is to establish the effectiveness of the study conducted with classical congestion management techniques to that of the machine models. The proposed work environments a novel and controlled setup of the next-generation wireless network and applies, assesses, and optimizes the traditional QoS and traffic-shaping methodologies, and generic AI forms of the supervised learning and reinforcement learning. Other standards that were utilized and incorporated in the efficient aspect included the latency, through put and packet loss with the view of evaluating efficiency levels of the employed methods. This simply measures that the use of ML is more superior to the conventional methods because it relieves latency and packet loss while at times enhancing through put. The analysis of the two states’ comparison proof that congestion recovery problems are more fit for the application of ML algorithms in modern transport systems due to better flexibility and online decision-making chances. Nevertheless, it discusses some problems that exist and they include; The major challenge of how to integrate the ML model in the system and the need to obtain massive training data. Therefore, one can state that the approach of applying more preventive ML models in controlling congestion also appears to be wiser when it comes to deploying lesions over wireless networks from the viewpoints of efficiency and scalability. Future work includes: working with real data; further work on the Advanced ML section; Security and Compliance.