Optimization Group

Entwicklung und Implementierung effizienter Decodieralgorithmen für lineare Blockcodes

The research project EIDelB is funded by the DFG from 01st January 2016 to 31st December 2018.

Motivation

In today's world, being connected to the internet is an important part of everybody's life. Without noting it explicitly, the amount of data shared is constantly increasing. For example, there are about 66,000 Google searches per second, 2,700,000 emails are sent per second, and overall, 60,000 GB of data are shared via the internet in one second. To deal with this large amount of data, communication needs to be fast and reliable. During data transmission, e.g. while streaming a video on the smartphone, data can become perturbed due to weather conditions, obstacles, or interfering signals. Therefore, it is important to make data robust to possible disturbances by adding redundant information to the sent message. This redundancy allows the receiver to detect and correct errors that occurred during transmission. 

Innovation

A so-called parity check matrix is used to detect and correct errors in a message. The error correcting works on a so-called Tanner graph which is induced by the matrix. There are two undesirable events: The algorithm may take too long or it may output an error. In the scenario of video streaming, an occurrence of the former case means that the video frequently stops to load. In the second case, the video can be too noisy to have fun watching.

Our approach is to alter the matrix and obtain a simplified Tanner graph. It had turned out that a reduced number of edges increases the algorithm speed and a reduced number of short cycles increases the error performance. The goal is to develop algorithms and heuristics to find optimal matrices with respect to the number of edges and cycles.

Partners 

  • Technische Universität Kaiserslautern
    • AG Optimierung (Prof. Dr. Stefan Ruzika, Tobias Dietz)
    • Fachbereich Elektrotechnik und Informationstechnik (Prof. Dr. Norbert Wehn, Kira Kraft)
  • Universität Koblenz Landau
    • Fachbereich Mathematik (Florian Gensheimer)

Funding

Zum Seitenanfang