Files
SastreFerrandez_22531400_2015.pdf
Open access - Adobe PDF
- 1.36 MB
Details
- Supervisors
- Faculty
- Degree label
- Abstract
- TCP (Transport Control Protocol) is widely known as a communication protocol between devices that guarantees a reliable, ordered and error-checked exchange of information flow between them. As research progresses, changes in it’s design and architecture are made, new features are added or modified, etc.. This translates into the existence of a large variety of different implementations of the same. Every new Linux Kernel version comes on the heels of representative changes inside TCP that may incur an en hancement or a disadvantage when it comes to use one or another in certain specific network scenario. Measuring performance of all distinct versions of the protocol turns into an impossible task to carry out in the ample variety of heterogeneous real environments. This thesis provides a set of simulations of different TCP implementations under diverse traffic scenarios, all performed in several network topologies considered representative to a range of real cases. To achieve this goal, simulations are done within the ns-3 discrete event network simulator along with the DCE(Direct Code Execution) framework. Obviously taking into account every potential scenario is impossible. Hence, what this work offers is a representative subset of tests that may be useful to later researchers/sysadmins as a reference when it comes to deploy one or other version in a real environment.