Optimization of TCP NewReno Algorithm Based on Available Bandwidth Estimation

With the development of new network structures, the current network environment has become complex and diverse. Traditional TCP NewReno algorithm does not have a packet loss judgment mechanism, it cannot distinguish the types of data packet loss. The noise packet loss is erroneously interpreted as the congestion packet loss, resulting in frequent jitter and long delay. We propose a new method, namely TCP NewRenoBw, to address the available bandwidth estimation. By judging the type of packet loss, it can effectively distinguish noise loss and congestion loss. Our simulation results show that TCP NewRenoBw can resist noise interference and make the transmission per unit time more stable at the same time. The throughput of TCP NewRenoBw is significantly higher than that of TCP NewReno, which achieves better bandwidth utilization and competition fairness, and improves the transmission performance of TCP in the low bandwidth and high packet loss rate network.