Resource Assignments in Network Data Transport

Abstract

Communication networks have come a long way since their invention and adoption in the modern society. They have fundamentally changed how society functions and the speed at which activities are conducted. Colossal amount of data is transported through today's networks to convey information between various end points. The Internet is the largest information and communication network shared by billions of end points as a common medium of data transfer today. The amount of data that can be transported through these networks within a given time period, also known as bandwidth, is the principal resource shared between the different end points. And, as is the case with many other finite resources available to modern society, scarcity of resources has led to the development of various schemes of resource distribution. The Transmission Control Protocol (TCP) is the dominant mechanism for transporting data between end points in the Internet. It provides various functionalities such as ensuring that the data is transported reliably between the end points and preventing the receiver and the network from being overwhelmed by too much data sent. TCP implements congestion control mechanisms so that multiple flows can share the network capacity. This thesis analyses the sharing of network resources in communication networks, using a customized TCP variant. Traditionally, TCP treats all flows as equal regardless of their end utility and attempts to provide equal share of the network bandwidth to each flow. This research analyses a scenario without that assumption. Specifically, short and long data transfers are considered unequal. As a manifestation of such concept, the research work developed a TCP congestion control variant which was able to provide different share of network capacity to different flow types. The performance of the protocol was then tested under multiple settings to investigate the notion further and evaluate the implications for deployment at large. Solutions to the resource assignment problem merit not just technical but also socio-economic analysis. The research explores the development of different resource assignment schemes in society at large and finds analogous schemes in communication networks. By comprehending the various notions of fairness prevalent in the society, several concerns to be addressed while developing resource assignment schemes are identified.