Why AI and automation are critical for assuring smart services

Industrial sectors, including manufacturing, mining and energy production, are undergoing digital transformation to improve decision-making and to automate processes. Communications service providers (CSPs) and cloud hyperscalers are underpinning this transformation by providing connectivity and infrastructure services that enable automation. At the same time, CSPs must automate their own infrastructure and services to support their enterprise customers.

Automation in smart industries

Industry’s foremost objective for digital transformation is to augment productivity through software for industrial process efficiency, monitoring and automation. The digital transformation is highly driven by connected devices providing the analytics for AI-based IoT applications. Examples are:

• Factory floor robotics where wireless robots reconfigure production lines, logistics and warehousing.
• Production-critical networks, such as electricity distribution grids using metering sensors.
• Primary industries, such as mining and oil/gas that use automated machinery combining sensors and applications.

Smart connectivity

Advanced or smart connectivity is critical for both indoor/outdoor and static/mobile devices to enable the industrial IoT applications and automation applications for on-campus or off-campus communications. Many industrial applications require robustness, latency, and data rate guarantees. Connectivity needs to offer high capacity because of the density of devices and the demand of IoT applications that use video streaming-based AI/analytics to guide the robotics applications. In addition, the security and privacy requirements are critical for placement of workload choices, on-campus or at the macro-network edge.

How CSPs can offer cost-efficient smart connectivity

Within the IT and telecom ecosystem, the technologies that enable a flexible, mission critical, high capacity, secured and cost-effective digital transformation are:

• Mobile Edge Computing (MEC), which pushes workloads to the edge (public MEC) or on-premises (private MEC) to meet security, latency, robustness, and capacity requirements.
• 5G and Open RAN provide multi-tenant or slices connectivity in a cost-effective manner as network functions can be placed within MEC infrastructure. For this, the sharing of radio spectrum enables cost-efficient and customized private 4G/5G networks.
• Cloudification is essential for flexible placement and management through Infrastructure as a Service (IaaS) model for compute, network, and storage intensive IoT applications.

The challenges of smart connectivity

Whether industrial enterprises go for pure private network approaches or use CSP or hyperscaler-based private/public networks and managed services, the challenges to implement the technology enablers are:

• Complexity resulting from additional abstraction layers and an end-to-end architecture where multi-tenancy and several compute, storage, and networking third parties are combined into complex value chains.
• High scalability to manage the explosion of sensors, managed elements and the IT workflows and infrastructure.
• The diversity in terms of open source and vendor proprietary technologies as the list of technology enablers grows.

Automation and AI to the rescue

To resolve the connectivity challenges, automation and AI go hand in hand; they assure the reliability and underlying smart connectivity of industrial applications. Workflow and process automation is key to enabling sequencing and processing of software-based tasks, while AI technologies reduce complexity and offer knowledge discovery. The TMF Autonomous Framework has defined the move to higher autonomy, i.e., from policy-based approach enabling automated networks to intent-based approach enabling autonomous networks, which is achieved through a hierarchy of autonomous loops. Intent is defined as the business goal that can be unambiguously understood by both humans and machines. However, until intent-based approach is fully achieved, automation workflows will continue to be written as rules or policies generated by humans for a policy-based automation. It’s important to note that automation assisted by AI will together provide the intent-based approach, much needed for smart connectivity by smart industries. Going forward, the role of a Service Assurance system is to offer AI and automation such that CSPs and enterprises can assure the connectivity and offer guaranteed QoS keeping in mind the end-to-end industrial applications. For this, Service Assurance systems would need to normalize multi-technology data from a CSP network, enabling high-quality modelling and monitoring of intent. And a policy-based workflow automation engine would be required to reinforce the business goals.