As peoples lives today accelerate by a significant rate day after day, so does the quantity of the information and data produced by us. While our infocommunications networks are doing a steady job to keep up with the pace of this kind of evolution, what can we say about the current situation of the transportation layer in the network stack?
The TCP protocol had been invented at least 30 years ago now, even though it still remains as the most valuable keystone for our communications needs. If anyone happens to be in need of trustworthy exchange of information, there's quite a good chance that TCP is involved in it. In recent years, there have been signs of improvement in the field of some of the well known performance related problems sorrounding TCP, therefore new congestion control algorithms were created and are being put to good use – but the problem that still remains is about the level of efficiency which could be obtained by controlling the dynamics of dataflows only from the edge's of the communications network.
During my work, I have studied the so-called high speed versions of TCP's congestion control algorithms, which emerged in the recent years. I have designed and implemented a measurement topology within a well known and flexible research testbed, so that I may be able to study the behavior of each TCP version, both separately and by examining the effects of what other algorithms can have on them when running side by side on the network. To achieve all this, I have used a relatively new software measurement tool, in the hope of that the produced results are going to be somewhat more precise and solid than the ones already created by long time existing tools.
I finish my diploma with the summary of the conducted experiments, showcasing the resulted data on various graphs and by giving a syllabus about the goals which have been procured.