Today, we are excited to announce that Arm RAN Acceleration Library (RAL), a key building block of most complex parts of RAN software, is moving to an open-source code base under the permissive BSD license. This means that you can now use, modify, and distribute RAL code without any restrictions beyond acknowledgment. You can also contribute to the development and expansion of Arm RAL by upstreaming your changes and adding new functionality.
But before we dive into the details of this announcement, let’s recap what Arm RAL is.
Announced in 2020, Arm RAL is a software library that provides optimized signal processing and related math functions for enabling 5G RAN deployments. It leverages the efficient vector engines of Arm Neoverse cores to accelerate 5G NR and LTE signal processing workloads, such as vector/matrix manipulation, channel coding, modulation, FFT. By using Arm RAL, you can achieve significant performance improvements over alternative legacy software implementations (e.g. 20x for Cyclic Redundancy Check and 4x in Polar Encoding).
Arm RAL is designed to address the challenges of developing and deploying RAN software on general-purpose CPUs. Conventionally, DSP cores were deployed for the L1 processing, but they have several drawbacks, such as vendor lock-in, limited scalability, and lack of cloud-native benefits. On the other hand, the CPU programming model is mainstream and offers more flexibility, portability, and efficiency. However, CPU cores also require specialized knowledge and skills to exploit their SIMD/Vector capabilities for RAN workloads.
That Is where Arm RAL comes in. It provides building blocks (functions or kernels) required by RAN L1 that run on CPU (and use SIMD). It abstracts away the complexity of vector programming and offers a simple API that can be easily integrated into your L1 stack. It also supports multiple architectures (Neon, SVE, SVE2 …) and operating systems (Linux and RTOS).
Arm RAL has been widely adopted by the 5G ecosystem partners since its introduction in 2020. Today, more than 100 companies have so far evaluated Arm RAL and multiple L1 vendors have used Arm RAL kernels in their Arm-based L1 implementations. This includes OEMs, ISVs, ODMs, and system integrators who are developing 5G infrastructure solutions using Arm RAL around the world.
Arm RAL has progressed significantly since its inception and now covers Low-L1, High-L1 and L2 functions for both 5G and LTE. Some recent features are:
We plan to add more building blocks in future, such as channel estimation and channel equalization. You can find the full list of features and functions in the Arm RAL documentation.
We believe that open-sourcing Arm RAL will bring several benefits to the 5G RAN community:
We are grateful to our partners who have supported us throughout this journey and have committed to contributing code to RAL. We look forward to working with them and the broader 5G RAN community to improve Arm RAL.
If you are interested in using or contributing to Arm RAL, you can find all the resources you need on our GitLab repository. There you can access the source code, documentation, examples, tutorials, and more. You can also report issues, request features, or submit pull requests.
You can also join our Arm Developer Community to ask questions, share feedback, or connect with other users and developers of Arm RAL.
We hope you are as excited as we are about this announcement. We can’t wait to see what you will create with Arm RAL.
Get Started with Arm RAL
Astella/Vicinity
“Arm RAL has optimized kernels to accelerate the development of commercial 5G infrastructure software”, said Dr. Eric Tsang, CEOs of Astella Technologies Limited and Vicinity Technologies Limited, “Astella and Vicinity are excited to collaborate with Arm to enable 5G infrastructures with other industry partners to provide commercial and competitive 5G products.”
ArrayComm
“Thanks to Arm for making Arm RAN Acceleration Library (RAL) available as open-source software. Our PHY development time can be reduced with the use of various modules such as LDPC and Turbo coding.” said Xin Huang, CEO of ArrayComm. “It is great to work with Arm to verify the steady improvement in performance and cycle counts in the RAL modules. RAL is a great addition to the low power, high performance Arm Neoverse Platform that will encourage innovation in the Arm 5G RAN ecosystem.”
NXP Semiconductors
“We are delighted to see Arm open-sourcing its RAN Acceleration Library, which provides a comprehensive set of Software Defined Radio enablement libraries for 5G RAN applications.” said Wim Rouwet, Senior Principal Engineer at NXP® Semiconductors. “Arm RAL has helped us to optimize our L1 software stack for our Layerscape® processors, which are based on Arm cores. By using Arm RAL, we have achieved significant performance gains and reduced our development time and cost. We believe that this initiative will benefit the whole Arm ecosystem and foster more collaboration and innovation in the 5G community.”
Phluido
"As users of the Arm RAN Acceleration Library (RAL), we are excited to support the announcement of the library being open sourced,” said Dario Fertonani, CTO of Phluido. “Arm RAL is a valuable resource for the 5G RAN ecosystem, particularly because of the Layer-1 kernels highly optimized for Arm processors. We are confident that Arm RAL will help accelerate the development of 5G Radio Access Networks.”
PureSoftware
“PureSoftware is proud to be partners with Arm in upstreaming RAL for OEM’s and ODM’s to strengthen 5G infrastructure worldwide.” said Noy Kucuk, Senior Vice President at PureSoftware. “The decision by Arm to make RAL Open-Source will facilitate increased collaboration and innovation, leading to faster deployment, upgrades, and reduced fragmentation in the 5G ecosystem.”
Software Radio Systems (SRS)
“The Arm RAN Acceleration Library provides new entrants like SRS with a key technical enabler to expand the OpenRAN ecosystem.” said Paul Sutton, Founder and CEO at Software Radio Systems (SRS). “The physical layer (L1) is the most challenging element of the RAN software stack, currently presenting a critical bottleneck for the entire industry. Arm RAL helps to eliminate that bottleneck, allowing OpenRAN to realize its full potential.”
4G/5G Consultants Pvt Ltd
“4G5G Consultants and Arm have collaborated to create an Open Source 5G ORAN DU solution, enabling simpler and lower cost proof of concept testing for our customers. 4G5G Consultants has ported, optimized, and integrated the Arm RAN Acceleration Library, which provides optimized signal processing blocks for building the necessary 5G Physical Layer channels in a DU.” said Raman Odugoudar, CTO at 4G5G Consultants Pvt Ltd. “The DU solution supports multiple Arm architectures including Neon, SVE & SVE2. This solution also features hardened Arm RAN Acceleration Library and 7.2x ORAN compliant Front Haul interface. The collaboration will further expand to include 4G5G Consultant's Innovative Core-Thread Pool framework based on power efficient multi-core capabilities delivered by Arm architectures as well as further contributions to the Arm RAN Acceleration Library.”