Robust Dynamic Congestion Control Protocol for Mobile Networks (TCP DCM+)
Derar Sameeh Abdel-Aziz Khader
ضرار سميح عبد العزيز خضر
Data networks are considered as a critical corner of data transmission between the different hosts wherever they exist. In the last few years, the wireless and mobile networks become more important for daily use and are their spread is increasing for personal and commercial use. The main difference between wired and wireless networks is the large number of lost packets during the data transmission. The packet losses are a result of errors on the data transmission channel. These errors are due to external noise, interference and mobility of the wireless devices that results in deep fading. The mentioned problems earlier are the reasons that the throughput of wireless, mobile and mobile adhoc networks is less than wired networks, which does not suffer such problems. Old traditional transmission control protocols like (Standard TCP) behave extremely hard when they detect any data packet losses. They drop the congestion window to the half though the transmission channel capacity is not exhausted. This high drop results in low throughput, hence longer time to finish the transmission. Most traditional TCP protocols lack the use of appropriate techniques to estimate the available channel capacity, which are known as bandwidth estimation (BWE) techniques. In 2004, TCP Westwood+ protocol proposed a technique for estimating the available channel capacity. It uses a first-order low-pass filter to find the available bandwidth. TCP Westwood+ has largely improved the throughput of TCP connections, however, the problem of window drops is still existing, which makes it less appropriate for use in networks, that include mobility, i.e. MANETs. Hence, it is desired to modify the TCP protocol behavior to eliminate these drops, which are the results of congestion events or channel problems. If the congestion events are eliminated, then we can detect the times at which the transmission channel problems occur. The proposed approach in this thesis is called TCP DCM+. It is the abbreviation for “Dynamic Congestion Control for Wireless and Mobile Networks”. The transfer of data with different sizes has been simulated with different packet error rates, which should simulate the existence of wireless channel for large packet error rates (1e-3 to 5e-2). We executed hundreds of simulations for cases with different parameters like error rates, MTU sizes, bandwidth of both bottleneck (link) and destination (access), protocol type and the size of sent data. We found that DCM+ performs better than the other approaches, especially if the error rates are large. We used the usual performance metrics like throughput, average delay and packet losses to measure how well our approach performs. Additionally, we introduced two new metrics to measure the total time needed to finish the transmission, and also to measure the robustness and stability of the transmission. Our conclusion is, that DCM+ is minimizing congestion events, hence, transmits data much faster, shows stable behavior and is highly robust compared with other approaches.