Home > Teaching > Bachelor Thesis Topics 2010

Bachelor Thesis Topics 2010

October 12th, 2009

Flowy Performance Evaluation and Improvements

Network routers can export information about network flows they have been forwarding. Flow collectors receive and record these flow records for further processing. The CNDS group has designed a stream-based query language to search for patterns in recorded flow datasets. A first prototype implementation called Flowy, mostly written in Python, has just been completed.

The goal of this assignment is to evaluate the performance of Flowy by comparing it to other tools and identifying bottlenecks of the current implementation and perhaps addressing them subsequently.

The qualification required is very good knowledge of Unix system interfaces and strong C / Python programming skills.

Flow Signatures for Popular Applications

Flow records summarize the network traffic seen at a network metering point (usually located in a router, but might be also located in end systems). Our hypothesis is that many popular
applications generate specific flow signatures, due to probing mechanisms to work around firewalls or due to automatic software update mechanisms or user registration features for example. The
goal of this assignment is to take a number of popular applications and to identify their flow signatures. If our hypothesis is correct, we should later be able to identify such applications in larger flow datasets.

The qualification required is an interest in studying in detail the network traffic generated by popular applications and to (reverse) engineer the logic generating the network traffic. Ideally, the
signatures would be simple and only contain significant features. General scripting and strong data analysis skill will be required.

Visualization of Network Flows

The analysis of large network traces and network flow datasets requires tools to visualize properties of the data. Last year, a tool called snam was created to visualize traces. It is implemented in C++ using OpenGL. The goal of the assignment is to extend the tool with additional visualizations.

The qualification required are excellent C++ programming skills and ideally knowledge of OpenGL and an interest in generating a visualization user interface using an agile development approach that meets the requirements of the users of the visualization tool.


6LoWPAN is a protocol layer enabling IPv6 over IEEE 802.15.4 wireless links. This radio technology is popular for low power and resource contrained devices (8/16-bit microcontroller, 4/8 KByte of
RAM). The goal of this assignment is to implement SNMP over 6LoWPAN with a particular focus on how to make such an implementation resource efficient in terms of CPU and memory usage.

This assignment requires very good C programming skills and the interest to learn about coding in embedded operating systems such as Contiki or TinyOS.

Energy Conservation Through Radio Transmit Power Management

Modern low-power radios such as CC2420 allow dynamic setting of transmission power in order to change the transmission range. Higher transmission powers lead to greater ranges, however, using lower transmission powers can enable more spatial reuse of frequencies. Taking this approach may also reduce the energy consumption. Measuring this energy saving through transmit power management could be very useful since it could lead to energy conservation being built in to the MAC protocols.

The purpose of the project is to investigate the state of the art in radio transmit power management for energy conservation purposes. It would also be useful to create a sensor network using TinyOS and/or Contiki to evaluate the effects of such methods on different network setups. Further, evaluating and comparing existing state of the art would be a meaningful result as well. It would also be interesting to evaluate efficiency of such energy conservation mechanisms on different types of broadcast and multi-hop networks.

TCP Congestion Control Measurements

While there is a standard for TCP congestion control, there are meanwhile many non-standard congestion control mechanisms in use on the Internet. The goal of this project is to further develop a tool that can detect the congestion control algorithm of a TCP server through active probing and to do a larger scale experiment over the Interent.

The ideal candidate has good knowledge about TCP and congestion control, good C programming skills, and strong organizational talents needed for carrying out a larger measurement experiment.

Comments are closed.