Enhanced TCP'S Congestion Control Mechanisms Over Long Term Evolution Networks
Wireless technologies have experienced intensive improvements in the past two decades. New technologies, expensive applications, and new specifications have been rapidly introduced. The latest were the Long Term Evolution (LTE), LTE-Advanced, and WiMAX-Advanced marketed as 4G. One other is the Trans...
Saved in:
| Summary: | Wireless technologies have experienced intensive improvements in the past two decades. New technologies, expensive applications, and new specifications have been rapidly introduced. The latest were the Long Term Evolution (LTE), LTE-Advanced, and WiMAX-Advanced marketed as 4G. One other is the Transmission Control Protocol (TCP), which is the most widely used transport protocol in today's networks. The TCP was designed as a reliable transport protocol over wired links, where losses rarely occur but generally indicate congestion. However, in wireless networks, losses are often caused by the wireless transmission medium, and include fading, interference, and node mobility. Unfortunately, TCP congestion control mechanisms fail to support the new wireless technologies. Significant unfairness and low performance over wireless networks have been reported. Traditional TCP congestion control mechanisms such as Tahoe, Reno, and NewReno incorrectly assume congestion to be the case for every loss event, and unnecessarily reduce transmission rate. Therefore, intensive research has been conducted to combat this harmful failure, and many modifications have been introduced to enhance TCP congestion control mechanisms over wireless networks. Even though these modifications share the same principle of recognizing the cause of the loss, they use different mechanisms to do this incurable task. This means that new modifications and enhancements are still required. In this study, we introduce a new End-to-End congestion control mechanism to enhance TCP performance over low-latency and wide-bandwidth networks such as LTE networks. The new modification uses a modified TCP Westwood's bandwidth estimation mechanism to preserve the best link utilization, and includes three modifications to the standard TCP congestion control mechanisms. Firstly, we proposed two modifications to the slow start phase. The first is a new mechanism to properly setup the Initial Slow Start Threshold value, and the second proposes a faster start phase to accelerate the Congestion Window growth over the poor utilized links. Secondly, we propose two new modifications to the Fast Retransmission and Fast Recovery mechanism: the faster retransmission and the best recovery. Thirdly, we modified the timeout function to prevent unnecessary congestion avoidance procedures when every timeout occurs. The proposed modifications were integrated into a new TCP congestion control implementation called PETRA using Network Simulator-3 (ns-3). Intensive simulation experiments were conducted to evaluate the new implementation performance over an LTE simulated network. The validation method includes comparing PETRA, NewReno, and TCPW for congestion window behaviour, throughput, average delay, packet loss ratio, jitters, and flow fairness. Simulation results showed significant improvements of the new modification compared to other implementations over long and short latency links, high bit error links, high mobility user equipment, and wide-bandwidth links. Therefore, the new modification is efficient for use as a transport protocol over low-latency and wide-bandwidth networks like LTE networks. |
|---|