Accelerating LTE Simulation

by vishy on August 8, 2011

in Case Studies,Jacket

Simulation in MATLAB is a driving force in several research projects. However, the accompanying long simulation times can tend to be a drag in many of these  projects. In this article, we shall bring up the example of the work on 3GPP LTE System Simulation by Yuan Gao et al (from Tsinghua University, Beijing) and demonstrate how the use of Jacket can significantly improve the simulator performance and lead to faster validation times in simulation projects.

3GPP’s LTE (Long Term Evolution) and LTE-Advanced are important telecommunication standards pertaining to 3G and 4G communication networks. With networks worldwide beginning to adopt them for consumer usage,  a great need has come up for several novel link and system-level communication techniques developed by researchers to validated on these systems.  Needless to say, such a need implies fast and reliable simulation.

Researchers from Tsinghua University have found that the MATLAB-based LTE System simulator they were using was of “low efficiency” and “insufficient to support further extension of the standardization process”.  As a result,  a part of the simulator was re-engineered using Jacket to speed up the process. The figure below describes the LTE system-simulator. The parts in yellow were targeted for parallelization: The Base Station Cycle block (labeled as BS Cycle) performs adaptive modulation and coding (AMC), scheduling,  and mapping from system level to link level (SL-LL mapping) for several users, each independently. The parallelism inherent here was exploited to generate a faster simulator using Jacket.

Design of LTE System-level Simulator

The graph below shows benchmark timings (in seconds) as a function of the number of users participating in the network. The Jacket-based simulator beats the previous MATLAB based simulator by a speed-factor of 3X!

 

Performance of GPU-based LTE System Simulator

 

The results were generated using the much-older Jacket 1.3 against MATLAB R2010a (64-bit). The hardware used involved an Intel Core i7-860 CPU, 4GB memory and a GeForce GTX 260 GPU. We would encourage that this experiment be retried with a superior GPU and the latest version of Jacket to take advantage of the latest improvements in GPU computing and enjoy much higher speedups.

We would like to thank the researchers from Tsinghua University for taking the time to explore Jacket to speed up this simulator, which we are sure, will help several other research projects involving LTE run faster in turn! To find more details about this work, please refer to this publication.

 

Comments on this entry are closed.

Previous post:

Next post: