Featured Story
Ericsson rewrites sales pitch in face of slowing traffic growth
Ericsson substitutes value for volumes in its patter after recognizing a slowdown in traffic growth – but it still bets AI will have a massive impact on the network.
Startup DeepSig said its new research highlights the role that artificial intelligence (AI) and machine learning (ML) can play in 5G. #pressrelease
May 16, 2022
ARLINGTON, Va. – DeepSig—experts in artificial intelligence (AI) and machine learning (ML) for wireless communications—today published "Amplifying 5G vRAN Performance with AI & Deep Learning." Co-authored with a leading industry partner, this white paper explores how mobile operators can leverage AI and ML to amplify critical air interface performance metrics and value while deploying open virtualized radio access network (Open vRAN) architectures.
Open vRAN's standard interfaces between primary network elements allow public and private network operators to select distinct, best-in-class products and enable breakthrough technologies to improve network performance and lower costs.
The new white paper explains how machine learning can be used to perform 5G radio baseband functions in an AI-Native method, which learns from radio data and the local radio environment and produces a more efficient and performant "neural receiver." The paper goes on to explain key outcomes of AI/ML in the physical layer including:
Maximizing link capacity, coverage, and performance in Multiple-Input Multiple-Output (MIMO) radio systems.
Minimizing OpEx, CapEx, and latency through improved processing efficiency by reducing core count, power consumption and processing time in the L1.
The white paper describes how AI and ML improve Open vRAN stack software to improve automation, enhance performance, and more. DeepSig's OmniPHY 5G AI software combines Deep Learning with FlexRAN and deep neural networks to improve wireless performance and resource utilization in the upper PHY running in L1 DU software in 5G Open vRAN systems.
This unique deep learning technology in Layer 1 improves power consumption, optimizes the computational load of the distributed unit (DU) baseband processing, and maximizes spectral efficiency, coverage area, and link margins. These findings are based on tests conducted in DeepSig's 5G Wireless AI Lab, which uses commercially available products to construct an end-to-end 5G standalone (SA) network based on Open vRAN architecture. The 5G test lab uses a mid-band FCC experimental license to conduct 5G NR over-the-air (OTA) tests with commercial 5G Handsets to validate model performance in a live environment across a range of operating configurations.
"Performance and cost are top concerns operators have when it comes to deploying 5G networks," said James Shea, DeepSig CEO. "Our technology continues to show machine learning provides leading vRAN performance that's robust and operational for addressing both concerns. This new white paper is a must-read for operators who have– or plan to implement–5G vRAN and want to lower opex and capex."
You May Also Like