The 5G Revolution is Empowered by AI and Automation in the Telecommunications Sector

The convergence of artificial Intelligence (AI) and automation has ushered a transformative era in the telecom sector.

In the telecoms industry, the sheer power and performance of 5G arguably takes us to that tipping point. The scope and scale of 5G is transformational but the resulting networks are too complex for people to operate cost-effectively without the use of AI, automation tools and cognitive technologies.

Yes, 5G affords endless opportunities for revenue, productivity and experience upsides and broader impact too, for example sustainability – but telecoms players will never make the very most out of 5G without automation, prediction, and orchestration tools. And especially so in a context of rising pressures and challenges, from rising data storage costs and the complexity of wireless/wireline convergence across multiple domains, to managing legacy systems and growing capital intensity needs, alongside addressing data quality, accessibility, literacy and data harmonization at individual and organisational levels.

Indeed, recent CSP research by Nokia available here finds that only 6% of respondents believe they are at the most-advanced level of automation, that reliant on AI and Machine Learning. No wonder then that research by SkyQuest finds that AI within telecoms is expected to attain a $10.4 billion valuation by 2030 and that collaborations are pivotal to underpin these advances, an excellent example being the recent partnership between MediaTek and NVIDIA – one poised to foster the next-generation of always-connected intelligent vehicles, embedded with the latest AI and computing capabilities.

Clearly, it is no longer a question of whether to adopt AI – but where and how to get started.

What Does AI Offer for 5G Networks?

The fifth generation of cellular networks (5G) is significantly more complex than its predecessors due to factors such as increased cell density, differentiated service requirements, and coexistence with legacy networks. Traditional operations and management solutions, which heavily rely on human intervention therefore become increasingly infeasible to support such complex networks at a reasonable and efficient operating expense.

AI can be applied at all stages of a telecom providers operations. AI simplifies deployment and network management by automating processes, improving network planning and forecasting, and optimizing resource allocation. It enhances service quality by prioritizing critical applications and services, ensuring ‘friction-free’ user experiences, especially in demanding contexts such as highly immersive and interactive metaverse scenarios. Further, AI contributes to higher network efficiency by analysing traffic patterns and device usage, enabling intelligent resource distribution based on real-time demand, whilst also helping to minimize the volume of data generated by client devices.

AI-driven analytics also improve network security by detecting and mitigating threats in real-time, while predictive maintenance reduces downtime and enhances reliability. Further still, AI can support more intelligent operations for sustainability benefits which can range from RAN energy reduction, optimised and reduced site expansions through AI driven capacity planning, and a significant reduction in CO2 via virtual drive testing approaches. It’s clearly a very broad set of benefits - let’s explore some key examples now in more detail.

AI and Automation to Boost 5G Profitability and Value

Of all the ‘tech pairings’ within today’s ‘Age of Convergence’, I believe the combination of AI and Automation is the leading catalyst for innovation, especially within telecoms. Automation improves customer service levels and enhances customer experience by executing and streamlining routine pre-defined tasks and reducing the need for manual intervention to improve network quality - and combined with AI - to offer increasingly personalized services too, further enhanced by superior insights into customer behaviours and preferences. This optimizes ‘Quality of Service’ by identifying and prioritizing critical applications and services, ensuring smooth user experiences even during high-demand scenarios. Compliance navigation is another growing use case.

Furthermore, AI algorithms can analyse and learn from massive amounts of multidimensional data, including cross domain in nature and in real-time to advance data-driven inference / prediction, decision-making and predictive analytics, for example automatically predicting traffic patterns and optimizing network resources. This dynamic adjustment of network parameters leads to improved network performance. Together, better customer service and improved network performance aids in both retaining and attracting customers. Given rising operational challenges such as tool sprawl, alert overload, burnout and talent shortages, I believe this improvement can equally support employee retention and onboarding too. Indeed, new research by Webex has shown the benefits of AI for both communication and wellbeing, more on this here.

AI can also help reduce capital expenditure by optimizing network performance including ensuring observability, coverage auto-optimizing and improving resource allocation. By analysing traffic patterns, device usage, engineering parameters and other factors, AI can automatically adjust network settings and reroute traffic in real-time, ensuring efficient utilization of network capacity – while pointing network planners to the best routes for upgrade and expansion.

Overall, AI supports more efficient management of network capacity leading to reduced infrastructure costs. When AI is applied to network management it also opens doors to new revenue streams by maximizing network capacity. This maximized capacity enables the provision of value-added services, such as personalized offerings to improve customer experience alongside advances such as zero-tech optimization and operations, with continual and real-time performance improvements, empowered by AI’s self-learning at scale.

How Does Performance and Optimization with AI Work? – Examples in Practice

AI is employed for intelligent beamforming in massive MIMO systems, improving signal quality and coverage, particularly in densely populated urban areas. AI can also increase the number of services operating concurrently whilst helping to prevent service disruption. When AI-powered predictive maintenance is applied to 5G infrastructure, operators can proactively identify potential issues before they cause network outages, thereby reducing downtime and mean-time-to-repair, enhancing service delivery and reliability, whilst all helping to ensure a frictionless user experience.

Additionally, Network Slicing in 5G technology enables the segmentation of the network into distinct, isolated portions, each tailored to specific needs. Artificial intelligence affords 5G ‘smart slicing’ - this by dynamically and automatically managing and optimizing these slices according to the specific applications and services they support. This close and adaptive management ensures optimal performance and resource allocation for each slice, and each with its own characteristics and service level agreements.

Finally, AI can also support discovery – beyond areas such as consumer behaviours, preferences, pain points and segmentation to broader use cases such as smart manufacturing, massive IoT and ultra-reliable low-latency communication, whilst also evaluating the feasibility and profitability of each!

AI to Boost 5G Network Security

Cybersecurity is afforded a huge boost with AI, as algorithms analyse network traffic patterns to identify anomalies and potential security threats, such as malware. This real-time analysis enables network operators to detect and respond to security threats quickly, reducing the risk of network breaches and protecting users' data, something ever more critical given the rising diversification in threat approaches as cyberattacks continue to escalate in scope, scale and sophistication - and with the associated costs brought into sharp focus here (research by Secureworks).

AI and Machine Learning algorithms continuously detect and mitigate security threats in real-time, making 5G networks more secure against various forms of cyberattacks. That includes through behavioural analysis: AI-driven behavioural analysis helps detect unusual patterns in user activity or network behaviour, signalling potential security breaches.

By monitoring and analysing user behaviour, automated security can identify suspicious activities and alert network operators to take appropriate action. Indeed, AI-based analytics tools can detect more subtle nuanced patterns in network traffic, helping mobile network operators (MNOs) identify and respond to sophisticated cyber threats.

These advanced analytics tools go beyond traditional rules-based detection systems, providing a more robust and proactive defence against AI-based malware and other emerging threats.

360 Degrees of AI-Powered Automation Benefits… on the Edge

Network operations, known to be one of the most complex aspects of a MNOs business, are crucial for a network’s success. These operations demand coordination across large business units, but the sheer capabilities of 5G have made network operations tough for humans to run.

AI simplifies these tasks, removing the complexity, by enhancing functions everywhere – from in-store experiences to call centre efficiency to network management. Telcos are increasingly adopting AI to optimize these service operations, and much of the action is at the edge: out in the field.

Indeed, edge AI is a significant focus for MediaTek, a global leader in semiconductor technology, emphasizing local AI processing in devices rather than relying on cloud-connected support: just the right fit for 5G networks that are by definition highly distributed. Thinking about smartphones specifically, with some 5 billion in use already, I believe the rise of on-device AI is set to transform these most ubiquitous of devices.

And thanks to MediaTek chips, MNOs can achieve real-time processing all at low power consumption – at the extremes of the edge. Edge AI from MediaTek includes Deep Learning Accelerators, Visual Processing Units, Multicore Scheduler, and Software Development Kits. In addition, MediaTek has recently announced it’s leveraging of Meta ’s Llama 2 Large Language Model (LLM) alongside its advanced APUs and NeuroPilot AI Platform.

This paves the way for on-device processing of Generative AI applications with the intent to build a complete edge computing ecosystem with the acceleration of AI application development embedded by design - across smartphones, smart-home, IoT, automotive and other edge devices.  And in respect to vehicles specifically, you can also explore news of the partnership between MediaTek and NVIDIA here – a collaboration set to advance the next-generation of always-connected intelligent vehicles, and with the latest AI and computing capabilities.

Final Thoughts

The complexity of 5G networks necessitates the ‘power pairing’ of AI and automation for efficient, cost-effective and sustainable operations. By simplifying management, enhancing forecasting, optimizing resources, bolstering proactive security, minimizing downtime, and improving the customer and employee experience alike, AI unlocks the full promise of 5G for telecoms while boosting revenues — making its adoption imperative. In addition, the benefits of AI can exist across a continuum of compute: from on-device, to cloud, edge cloud, multi-access edge compute (MEC) or datacentre.

And with Edge-AI specifically, as exemplified by MediaTek’s continuing innovation at the edge, I believe this is raising the game in the augmentation of human intelligence and the enablement of hyper-personalization, privacy, security, and responsiveness – by design.