“Comparison of TCP Congestion Control Algorithms: Harnessing the power of Traditional Hybrid and Machine Learning Fusion.”

Modern evolution of communication technologies have resulted in complexity of network communications. In addition increasing diversity of preferred applications with large number of end users have further exaggerated this complexity. As a result the traditional rule-based Transmission Control Protocol (TCP) congestion control approach have now become absolute to efficiently handle this new type of intricacy. Presently machine learning approach finds a major role in solving TCP congestion issue. Both the traditional TCP congestion control and congestion control by machine learning approach have their own merits and demerits. The traditional TCP have less overhead, faster convergence, and show fairness and hence are practical but not adaptable and flexible to the dynamic environment of the networks[1]. On the other hand the machine learning algorithms are good at adaptability and flexibility but are not practical. This paper presents overview on hybrid modes of TCP congestion control namely (i) Combining more than one traditional rule-based approach, (ii) Combining Traditional rule-based and machine learning-based approach and (iii) The combination of more than one machine learning based approach, in order to overcome the present intricacy in communication networks. The main focus of this overview is to highlight that hybrid mode of TCP with machine learning can be an efficient solution for the present TCP congestion issues.