Lessons from ANRW ’24: AI and Cloud in 5G/6G Systems

The ACM, IRTF & ISOC Applied Networking Research Workshops (ANRW) offer a vibrant forum for researchers, vendors, network operators, and the Internet standards community to exchange emerging results in applied networking research. To foster collaboration across these diverse groups, ANRW events are co-located with IETF standards meetings, typically held annually in July. These workshops prioritise interactive discussions and engagement, complementing traditional paper presentations.

ANRW '24, held on 23 July 2024 at the Hyatt Regency Vancouver, brought together industry leaders and academics to share insights on advancing networking technologies. Among the standout sessions was a keynote presentation by Sharad Agarwal, Senior Principal Researcher at Microsoft. His keynote titled, "Lessons I Learned in Leveraging AI+ML for 5G/6G Systems", highlighted pivotal themes influencing telecom and networking.

Sharad distilled his experiences into three key lessons, each underscored by examples of research and systems developed to address specific challenges in the telecom industry:

1. Leverage Cloud Scale to Overcome Limitations of Deployed Protocols: He emphasised that the scale of cloud computing is critical to managing the massive demands of modern telecom networks. For instance, systems like TIPSY (Traffic Ingress Prediction SYstem) demonstrate how AI and ML can predict traffic ingress points across thousands of peering links, helping to avoid bottlenecks and ensure optimal traffic distribution.
2. Custom Learning Algorithms vs. Off-the-Shelf Solutions: While bespoke algorithms offer higher precision for niche applications, their complexity and deployment challenges often outweigh their benefits. Sharad argued for balancing innovation with practicality, advocating for leveraging pre-built AI and ML models wherever possible to streamline integration.
3. Mitigate Risks of AI Hallucinations through Careful System Design: Acknowledging the risks posed by unreliable AI outputs, he stressed the importance of robust system design. Using LLexus, an AI-driven incident management system, as an example, Sharad highlighted techniques like iterative plan generation, validation rules, and human auditing as essential safeguards against AI errors.

The talk also delved into broader trends shaping the telecom landscape:

• Cloudification of Telecom Infrastructure: The shift from hardware-based to software-based network functions, underpinned by cloud-native principles, has revolutionised telco infrastructure. This transformation facilitates rapid upgrades, reduces costs, and introduces new opportunities for AI-driven analytics.
• Challenges in Performance and Reliability: Ensuring high throughput, low latency, and carrier-grade reliability in cloudified networks remains a significant hurdle. Innovations like PAINTER and LLexus demonstrate how AI and ML are being applied to optimise these aspects.
• Emerging Business Models and Private Deployments: The integration of new radio technologies and virtualised network functions is driving novel revenue streams, such as private 5G/6G networks for mission-critical applications like factory automation.

Finally, Sharad’s keynote underscored how AI, ML, and cloud computing are reshaping the telecom industry, particularly in the era of 5G and the forthcoming 6G. By leveraging the scale of cloud infrastructure, balancing algorithmic complexity, and designing systems with resilience against AI pitfalls, the industry is poised to meet its ambitious goals of high bandwidth, low latency, and unparalleled reliability.

The video of his talk is embedded below and the slides are available here:

Related Posts:

• Free 6G Training - SK Telecom 6G White Paper: View on Future AI Telco Infrastructure
• Telecoms Infrastructure Blog: SK Telecom's Vision for Future Telco Infrastructure in the AI Era
• Telecoms Infrastructure Blog - Data Centers At Meta: Heterogeneous Integration Driven By AI/ML And Network Applications 

AI/ML and Other ICT Industry Trends in the coming decades

At the Brooklyn 6G Summit (B6GS) 2023, top tier economist Dr. Jeff Shen from BlackRock, presented a talk from the industry perspective of AI (Artificial Intelligence) and investment. Jeff Shen, PhD, Managing Director, is Co-CIO and Co-Head of Systematic Active Equity (SAE) at BlackRock. He is a member of the BlackRock Global Operating Committee, BlackRock Systematic (BSYS) Management Committee and the BlackRock Asian Middle Eastern & Allies Network (AMP) Executive Committee.

How are firms reacting to AI? From 'ICT Industry Trends in the coming decades' by Jeff Shen, BlackRock at #B6GS - I am guessing the peak of mobile mention was the peak 5G hype in 2015/16. Hopefully the industry is now much wiser and won't repeat the 6G hype. pic.twitter.com/xjEiDtmSUZ

— Zahid Ghadialy (@zahidtg) November 5, 2023

In his talk he covered the history of how and where AI has been traditionally used and how the thinking around AI has changed over the last few decades. He then presented his view on if AI is just a fad or it's more than that. To illustrate the fact, he provided an example of how Generative AI market is expected to grow from $40 Billion in 2022 to $1.3 Trillion in 2032. 

There are many challenges that AI faces that one should be aware of; namely regulation, cyber threats and ethical concerns. In the US, AI touches the entire economy, from legal to healthcare. In their quarterly reporting, firms are now discussing AI and the larger tech companies are not afraid to grow inorganically in order to get more exposure to the trend. 

You can watch the whole of his talk embedded below, courtesy of IEEE Tv.

Related Posts: 

• Free 6G Training: 'Impact of 6G on Metaverse' from Brooklyn 6G Summit (B6GS) 2023
• Free 6G Training: Qualcomm Keynote on 'Advancing Wireless Innovations towards 6G — from a Device’s Viewpoint' from B6GS 2023
• Operator Watch Blog: Jio 5G - Enabling India to go Digital
• Operator Watch Blog: UScellular 5G Progress and 6G Plans
• The 3G4G Blog: Presentations from 2nd IEEE Open RAN Summit
• The 3G4G Blog: AI/ML Enhancements in 5G-Advanced for Intelligent Network Automation
• The 3G4G Blog: What Is the Role of AI and ML in the Open RAN and 5G Future? 

Presentations from 2nd IEEE Open RAN Summit

The second IEEE SA (Standards Association) Open RAN summit, hosted by the Johns Hopkins University Applied Physics Lab, took place on 9-10 Aug 2023. It covered the topics related to the standardization of Open RAN including O-RAN Alliance, 3GPP, IEEE, various deployment scenarios, testing and integration, Open RAN security, RAN slicing, and RAN optimization among others. 

The videos of the presentations can be viewed on the summit page here or though the video playlist here.

The talk from Dr. Chih-Lin I, O-RAN Alliance TSC Co-Chair and CMCC Chief Scientist, Wireless Technologies on 'AI/ML impact, from 5.5G to 6G' is embedded below:

Related Posts: 

• The 3G4G Blog: Open RAN Terminology and Players
• The 3G4G Blog: O-RAN Introduction for Beginners
• The 3G4G Blog: Accelerating O-RAN Adoption Through Open Source
• The 3G4G Blog: 5G RAN Functional Splits
• The 3G4G Blog: Open RAN Explanation, Videos, White papers and Other Resources 

What Is the Role of AI and ML in the Open RAN and 5G Future?

Artificial Intelligence and Machine Learning have moved on from just being buzzwords to bringing much needed optimization and intelligence in devices, networks and infrastructure; whether on site, on the edge or in the cloud.

New Study in 3GPP RAN3 on Artificial Intelligence and Machine Learning in NG-RAN - https://t.co/o4o4kq6Wpp via @3GPPLive #Free5Gtraining #3GPP #5G #ArtificialIntelligence #MachineLearning #AIML #RAN #Algorithms #SON #5GC #5GS pic.twitter.com/pvPlG9HpPC

— Free 5G Training (@5Gtraining) October 15, 2021

Qualcomm has been very active in talking about AI/ML in webinars and on their site. A detailed blog post looking at 'What’s the role of artificial intelligence in the future of 5G and beyond?' is available here. It was posted in time for a Light Reading webinar where Gabriel Brown, Principal Analyst – Mobile Networks and 5G, Heavy Reading and Tingfang Ji, Senior Director, Engineering - Wireless R&D, Qualcomm discuss the topic. The video is embedded below and slide deck is available here.

Standardizing AI/ML in cellular communication systems. From #Qualcomm presentation, The essential role of AI in the 5G future - How machine learning is accelerating wireless innovations in the new decade and beyond - https://t.co/fwxtoKIZKR#Free5Gtraining #3G4G5G #5G #AI #ML pic.twitter.com/YS5w8KXQuw

— Free 5G Training (@5Gtraining) September 27, 2021

Louis Scialabba, Senior Director of Marketing at Mavenir, looking at AI and Analytics spoke at Layer 123 conference on the topic, 'AI/ML for Next Gen 5G Mobile Networks'. His talk is embedded below and a blog post by him on the topic, 'The RIC Opens a New World of Opportunities for CSPs' is available here.

Related Posts: 

• Free 6G Training: Qualcomm Pushing the boundaries of AI research
• The 3G4G Blog: 3GPP Release-18 Work Moves Into Focus as Release-17 Reaches Maturity
• The 3G4G Blog: 5G-Advanced Flagship Features
• Connectivity Technology Blog: 5G Indoor Precise Positioning
• The 3G4G Blog: ITU Standardization Bureau on Machine Learning for 5G
• The 3G4G Blog: AI your Slice to 5G Perfection
• The 3G4G Blog: Qualcomm Explains 5G Millimeter Wave (mmWave) Future & Integrated Access and Backhaul (IAB)
• The 3G4G Blog: Qualcomm Demoes Sub-band Half Duplex (SBHD)
• Connectivity Technology Blog: Super Uplink to improve 5G Coverage and Speeds
• Free 6G Training: One XR Device to Rule Them All! 

ITU Standardization Bureau on Machine Learning for 5G

Last year I blogged about Global ITU AI/ML 5G Challenge on the theme “How to apply ITU's ML architecture in 5G networks".  The grand challenge finale happened in December. All the recording and presentations are available here.

Back in October, Bilel Jamoussi from ITU presented a keynote to the 2020 IEEE 5G World Forum plenary session where he addressed the challenges of applying machine learning in networks, ITU’s ML toolkit, and ITU’s AI/ML in 5G Competition. IEEE Tv shared the presentation only in April so the competition part is a bit outdated. It does nevertheless an interesting 20 minute talk.

ITU Recommendation Y.3174, Framework for data handling to enable machine learning in future networks including IMT-2020 is available here.

Related Posts: 

• The 3G4G Blog: Artificial Intelligence (AI) / Machine Learning (ML) in 5G Challenge by ITU
• The 3G4G Blog: IMT-2020 (5G) Requirements 

Anomaly Detection and other AI Algorithms in RAN Optimization

Yesterday I watched this very inspiring live chat that I would like to recommend to anyone who is interested in how machine learning techniques (aka "AI") can help to optimize and troubleshoot the Radio Access Network.

 [The real contents of in the video starts at approx. 42:00 min] 

My key takeaways from this fireside chat are: 

Verizon Wireless has enough data (100… 500 time series KPIs per cell) that they use to feed anomaly detection ML algorithms and this generates a huge number of alarms, but only a few actionable outputs. 

The “big elephant” (Nick Feamster) is to identify if these alarms indicating real problems that can and have to be fixed or if they just indicate a new behavior of e.g. a new handset or a SW version that was not present in the training phase of the ML algorithm and hence, its pattern is detected as a new “anomaly”. 

For Bryan Larish (Director Wireless AI Innovation, Verizon) the “big open problem” is “that it is not clear what the labels are” and “no standard training sets exist”. 

[For more details watch the video section between 52.00 min and 57:32 min and listen to Bryan’s experience!] 

In most cases Verizon seems to need subject matter experts to classify and label these anomaly alarms due to “the huge diversity” in data pattern. 

According to Bryan only for very few selected use cases it is possible to build an automated loop to fix the issue. Especially the root causes of radio interference are often mechanical or cabling issues that need manual work to get fixed. 

All in all it is my personal impression at the end of the session that anomaly detection is currently a bit overhyped and that the real challenges and problems to be resolved start after anomalies are detected.

Nevertheless, as Bryan summarizes: “ML is a very, very powerful tool.” 

However, strategically he seems not to see a lot of value in anomaly detection by itself, but rather: “Can we use machine learning (results) to change how we build networks in the future?”