print page

Bachelor's Thesis

ROFL - RatiOnal FLoat

Motivation

On sensor nodes, computations with non-integer numbers are usually inefficient and waste a lot of precious space for program code. This is due to floating point computations being implemented in software rather than hardware since the CPUs of sensor nodes usually do not include a floating unit. We propose a promising approach for representing and working with floating point numbers. More information will be given to the applicant in person.

Work description

The student will implement a class, preferably in C++, that implements the new concept for floating point numbers. The aim is to prove the practicability of our approach. Furthermore, an evaluation is expected that compares our approach to the classical software floating point emulation and to verify the sanity in different aspects. After evaluation, the student might find that the proposed idea is unpromising, this will not have any influence on the overall appreciation of the submitted bachelor thesis.

Start date 30. November -1
End date 20. March 2013
Documents Flyer
Projects ToleranceZone | Self-Stabilization
Supervisor Dr. rer. nat. Sven Köhler
Dr.-Ing. Gerry Siegemund