Mobile network operators (MNOs) around the world are starting to build out 5G networks that will usher in a new era of gigabit-speed, low-latency, ultra-reliable mobile services. MNOs envision 5G powering a broad range of applications and use cases for consumers and businesses that are simply not possible in today’s 4G/LTE networks. 5G will connect billions of devices via next-generation New Radio (NR) technology operating in new spectrum bands. 5G networks will also feature dense deployments of small cells, facilitating the delivery of high-bandwidth services in distance-limited, high-frequency bands.
All this will be enabled by 5G cloud-native infrastructure that is disaggregated, virtualized and software-defined. Utilizing a cloud-native stack will power digital transformation, allowing 5G MNOs to benefit from the adoption of modern DevOps techniques employed by the leading cloud service providers and cloud-based digital enterprises. The CI/CD approach will enable MNOs to conceive, develop, test, deploy and enhance 5G services at Internet speed and scale, which holds the potential to unleash a wave of unprecedented service innovation in the telecom industry.
But there is a catch.
Overcome 5G Complexity with Machine Intelligence
5G networks are inherently complex and will be deployed at a scale that will ultimately outstrip the ability of human operators to keep on top of network and service operations to satisfy SLAs and ensure subscriber quality of experience. To further complicate matters, the vast majority of devices will not be controlled by human users who might be clever enough to work around connectivity problems. In many cases, devices might not even be capable of reporting a problem.
The 5G rollout will require mobile networks to advance beyond operator-centric network management and service orchestration. Machine intelligence – analytics, machine learning (ML) and AI – will be required to augment the ability of operations staff by generating real-time operational intelligence that directs network automation and orchestration functions, with minimal operator intervention.
Network operations will transition from human operators making decisions and directing workflows utilizing information gleaned from tools and dashboards to analytics-driven closed-loop automation that enables zero-touch network operations.
Machine intelligence is also integral to the operation of cloud-native infrastructure. CI/CD requires continuous feedback about system state that is derived from analytics, ML and AI, and this operational intelligence is used to drive automation and orchestration functions in the cloud-native stack.
3GPP Network Data Analytics Function (NWDAF)
The 3GPP recognized the need for machine intelligence in conceiving its 5G Service-Based Architecture (SBA), which provides the overarching framework for the set of standards which govern the operation of the 5G RAN and 5G Core. In contrast to 5G, the 3GPP standards for 4G/LTE were conceived at a time when real-time streaming analytics, ML and AI were not ready for prime time. The 3GPP 5G SBA explicitly defines data analytics functions that are integral to the operation of 5G networks and that deliver operational intelligence for driving network automation and service orchestration functions in the 5G Core and informing provisioning, performance and fault management functions in the OAM layer.
The Network Data Analytics Function (NWDAF) is the 3GPP standard analytics function that provides machine intelligence in the 5G Core, driving closed-loop network automation, which will be critical for MNOs to successfully deploy and operate 5G networks at scale.
NWDAF continuously collects and analyzes a wide range of data from the 5G network, as well as various application functions and OAM layer functions. NWDAF utilizes real-time analytics, ML and AI to generate insights about the current state of the network and the behavior of attached devices. It then delivers this operational intelligence to specific requesting network functions in the 5G Core, which use this information to execute the appropriate network automation and service orchestration processes.
The Rise of the Machines
The adoption of machine intelligence in 5G networks will follow a logical progression, based on the refinement and application of increasingly powerful ML/AI technologies over time. For example, NWDAF performs statistical analytics to monitor different aspects of network state and detect any significant changes. Trend analysis then builds on this to predict future network state. NWDAF implementations will employ novel techniques to deploy, train and score ML/AI models that generate these predictions.
Network operations staff will be relieved of performing routine but important tasks as they grow to trust automated systems that have proven to be accurate and reliable. More importantly, operators will depend on machine intelligence to automate the highly complex and time-critical tasks that lie beyond the ability of the human mind, with its limited capacity for brute force data ingestion and number crunching.
While the ultimate goal is zero-touch operations, it is unrealistic to expect that machines will replace human operators in a system as vast and complex as a 5G network supporting millions of devices. Instead, machine intelligence will augment human intelligence, automating workflows that are well-defined and can be driven by the output of well-defined data analytics functions. This will free up operations staff to concentrate on the more difficult problems that currently lie beyond the capability of today’s ML/AI technology.
Read more on 5G Analytics & NWDAF:
- Guavus 5G-IQ NWDAF
- 5G Analytics Standards: Who Needs ‘Em?
- NWDAF: Automating the 5G Network with Machine Learning & Data Analytics
- 5G Rollout Challenges & Analytics-Driven Remedies for MNOs
To learn more about Guavus’ 5G NWDAF approach, schedule a meeting with Guavus today.
Image attribution: iStock