Studentische Arbeiten

Titel: Enhancing Iterative Decoding of Polar Codes
  • Forschungsarbeit
Status: abgeschlossen


Polar codes are proven to be capacity achieving under successive cancellation (SC) decoding for infinite block lengths. However, for short lengths, SC decoding shows a weak performance compared to state-of-the-art LDPC codes. A major breakthrough in polar decoding for short length codes was achieved by Tal and Vardy with a successive cancellation list (SCL) decoder. SCL decoding renders polar codes into a powerful coding scheme whenever short block lengths are required, as for example for the internet of things (IoT) or very low latency applications, both cornerstones of the upcoming 5G standard. Besides their excellent decoding performance, the code structure of polar codes is inherently given by the concept of channel polarization, making them attractive for upcoming communication standards. Additionally, the code rate can be freely chosen by appropriately fixing a fraction of frozen bits.

However, the issue of the high SCL decoding complexity needs to be solved before polar codes can become competitive for practical applications. Thus, an alternative approach shall be evaluated in this work.


•    Improve BER performance of polar codes under iterative decoding
•    Convergence behavior not (well) visualized
•    Current (experimental) decoder shows low throughputs


•    Extend existing setup (CUDA)
•    Evaluation of several algorithms
•    Analyze convergence behavior of decoder (CUDA & Matlab)
•    [optional] Implement own ideas
•    Thesis in GERMAN or ENGLISH language

•    CUDA (GPU) programming
•    MATLAB programming
•    Bit error rate (BER) simulations (extend existing simulation framework)


•    Basic knowledge about channel coding
•    Programming experience in CUDA (or C)
•    Matlab
•    Could be extended towards a master thesis


•    Basic knowledge about polar codes (which might be a hot candidate for next generation communication standards)
•    GPU computing