Adaptive Channel Equalization : Multicore Processor Implementation
Kempi, Ilia (2015)
Metropolia Ammattikorkeakoulu
2015
All rights reserved
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-201504043922
https://urn.fi/URN:NBN:fi:amk-201504043922
Tiivistelmä
Advanced telecommunication techniques are more and more relying on the digital processing power that is needed for the signal demodulation and data recovery. In order to assess the practicability of concurrent signal processing in the field of communication, this project was focused on developing the adaptive signal filtering application on a distributed processor system.
To prove and evaluate this concept, different versions of the fundamental and robust Least Mean Squares adaptive filter were implemented on a multi-core digital signal processor. Developed applications were tested by streaming the simulated digital transmission data to the device via Ethernet. Algorithm benchmark results are compared in terms of adaptation rate and execution speed.
The parallelized version of the adaptive algorithm has shown promising results in terms of convergence and the workload distribution while its execution time is still inferior to the single-processor applications. Considering that only the most essential device functionality was used, there is a big room for improvement and optimization. The algorithm evaluation system established in the project can be reused for other concepts and related teaching.
To prove and evaluate this concept, different versions of the fundamental and robust Least Mean Squares adaptive filter were implemented on a multi-core digital signal processor. Developed applications were tested by streaming the simulated digital transmission data to the device via Ethernet. Algorithm benchmark results are compared in terms of adaptation rate and execution speed.
The parallelized version of the adaptive algorithm has shown promising results in terms of convergence and the workload distribution while its execution time is still inferior to the single-processor applications. Considering that only the most essential device functionality was used, there is a big room for improvement and optimization. The algorithm evaluation system established in the project can be reused for other concepts and related teaching.