A Review on Machine Learning based Loss Discrimination Algorithm for Wireless TCP Congestion Control
Pallavi Bhomle, Shrikant V. Sonekar · Zenodo (CERN European Organization for Nuclear Research) · 2020
In the incorporated condition of wired and wireless networks, different issues can bring about bundle misfortunes, for example, congestion and wireless parcel misfortune. The current TCP senders treat the parcel misfortunes as the loss of bundles in the transmission line which are brought about by network congestion, coming about to the TCP execution corruption. In this paper, the network congestion and wireless parcel misfortune are segregated dependent on machine learning algorithms with one concealed layer. In the event that the evaluated outcome is chosen as bundle misfortune from congestion, at that point congestion window and ssthresh is diminished to half, in any case, those qualities are kept in the event that it is chosen as a wireless mistake. The proposed algorithm utilizes essential TCP NewReno, yet if there should arise an occurrence of bundle misfortune, it adjusts congestion control plans dependent on the pre-learned machine-learning algorithm. The reproduction results show the improved TCP execution as contrasted and different existing TCP algorithms.