AccelerComm develops LDPC Decoder for Intel’s FlexRAN

1 min read

AccelerComm, the company specialising in optimisation and latency reduction IP, has announced they have developed a highly optimised LDPC software decoder in collaboration with Intel.

The solution is integrated into Intel’s FlexRAN Reference Software resulting in increased throughput by up to 3 times over alternate implementations. The LDPC decoder has been designed with a combination of optimised architectural and algorithmic enhancements and use of the powerful Intel Architecture and Intel Advanced Vector Instructions 512 (Intel AVX512) instruction sets.

The AccelerComm LDPC implementation takes advantage of these capabilities on Intel Xeon Scalable processor and with fewer cores delivers performance within 0.1 dB of Intel’s SDK module. FlexRAN is a reference architecture developed by Intel to implement virtualised Radio Access Networks which can be deployed on any part of the wireless infrastructure from edge to core.

The LDPC software decoder can operate as an upgrade to FlexRAN or standalone and is available directly from AccelerComm.

Channel coding, which is also known as forward error correction, is used to correct transmission errors in mobile communications caused by noise, interference and poor signal strength. While 3G and 4G used Convolutional and Turbo codes for the control and data channels, 5G uses the much more sophisticated Polar and LDPC codes and requires the industry to look afresh at how to address error correction.

If channel coding is not working well the impact on mobile networks is poor capacity, poor data rates, poor coverage and poor quality of service.

“5G promises a revolution in the way that mobile networks are deployed and their performance,” said Tom Cronk, CEO of AccelerComm. “To deliver on these expectations the industry has to increase efficiency at every level in order to deliver the best possible potential consumer experience and maximum business potential. This combination of AccelerComm’s pioneering Channel Coding IP and Intel technology delivers significant performance and flexibility gains.”