The revenue and profit of CSPs (Communication Service Providers) has been pressured by increasing traffic, the decline of price per Gigabit (Gb) and external and internal competition. For operators to regain a competitive edge, it’s critical to have clarity as to what is happening in the network.
Artificial Intelligence for IT Operations or AIOps, can play a key role for CSPs, by attaining intricate data visibility and delivering actionable intelligence in real time, without adding hardware costs or increasing the network complexity. Real operational value can be derived from lowered costs, increased revenues and improved customer experience.
The never-ending complexity and pressure on profit in a recent annual industry survey, showed that nearly a quarter of all telecom industry professionals who answered the survey opted for “increased pressure to lower prices and profit margins” as the leading long-term threat to the success of their business. This threat comes from two directions.
From one angle, the advancement of technologies from 10 Gigabit Ethernet to 40 GbE, then 100 GbE or in the mobile world, from 3G, 4G and now 5G, has pushed the volume of data going through the networks to increase exponentially. CSPs have made massive investments to upgrade their networks with the latest and greatest technologies, but the price they are able to charge their customers hasn’t increased much. For many operators, the ARPU value hasn’t altered for years which basically means the revenue per Gb of data or per minute of airtime, has decreased substantially. Competition between CSPs are often, just focused on price and that from outside the coms industry, has vastly affected the status quo. No case is more telling of such disruption that that of instant messages replacing text messages over the last decade.
As a result of the nearly-nothing additional network cost, SMS was a very profitable business for operations for years. With the rise of 4G, multiple OTT messaging services with impressive features, took over SMS in a big way.
What’s probably most difficult for operators to see is that, these service operators have an active part in destroying their value. To combat these revenue and profit losses, operators have been busy rolling new technologies and aiming to gain an advantage over their peers and retake control of the ecosystem. However, in addition to the increasing network costs, it is becoming increasingly difficult to roll out new technologies one after the other. This is because networks are becoming more complex with the generation of legacy networks that co-exit alongside the new additions.
One such consequence of this complexity, is that it is very difficult for operators to have full clarity of what is happening on their networks. This should actually be the foundation for the successful roll out of new services and transformation into new business models. Traditionally, operators have used corporate IT analytical tools like that of deep packet inspection (DPI) technologies which extract network data for monitoring analytics and security purposes. The drawbacks of DPI and other mainstream solutions include, high costs, weak scalability and rigid, outdated policies that are unable to satisfy the needs of CSPs. They require new tools to carry out such tasks more efficiently and effectively.
Artificial Intelligence for IT Operations, or AIOps, was a concept originally developed by Gartner a few years ago. As the name suggests, the system was intended to improve corporate IT support systems. Based on the fundamentals of AI’s big data analytics capabilities, AIOps is supposed to scale the manageability of IT systems to support the corporates’ growth strategies. This can also apply to operators and its application in the world of telecoms.
This is driven by two trends in the telecoms industry with one complimenting the other. The first is the evolution of networks. Previously, there was a clear demarcation inside CSPs between production telecom networks and IT support systems. As the networks become more and more IP-based, especially when we move to 5G, with its strong characteristics of virtualisation and software centricity, there is a requirement for a much stronger integration between networks and the IT systems. The second, is the size and complex nature of data that operators are dealing with. The telecoms systems become more powerful and the sheer size of the data that they generate and handle, continues to grow. Making sense of the correlation between data points and producing actionable business intelligence from massive volumes of data, goes far beyond the abilities of manual work. The telecoms industry has to embrace AI and machine learning to do the job for us.
Such complexity, further to the hybrid nature of the majority of networks with components from different generations, also comes from a shift inside the ecosystem. Since the 4G years and even more so in the era of 5G, telecoms companies have been working hard to not just be ‘connectivity providers’ and to become more of a platform that enables digital services. To succeed in this transition, operators need to be much more proactive in their collaboration strategies. Partners could include OTT service providers and could also factor in industrial customers, manufacturing facilities, autonomous car operators or even local municipalities.
To interact with partners in real time also requires, a high degree of automation from telecoms operators, which is powered by robust AI, machine learning and big data analytics capabilities.
The many advantages of AIOps are evident in their capability to provide comprehensive visibility into the networks. This includes network data, service and application data and customer data. Each category can be broken down in to numerous sub-sections. As an example, network data covers data from classic telecoms networks, virtualised networks and cloud networks. In addition, AIOps is much more agile in following user-defined workflows and answering ad hoc questions. The scalability is much stronger and robust. AIOps is also integration-ready with other network components and also interfaces with partner networks with strong API automation. Last but not least, AIOps is able to tie network traffic to business identifiers (customer’s applications and locations) as opposed to more conventional IT tools.
When it comes to boosting telecom operators’ competitive edge, the contribution from AIOPs can be evaluated in three dimensions:
- Growing top line business revenues
- Optimising network costs
- Improving customer experience
Thorough and real time data clarity, lets operators see the end-to-end traffic flow through their networks. With such data visibility, operators can discover if untapped upsell opportunities exist. For example, current customers might be using off-net services where the operators could promote their own in-house solutions, or outreaching to potential customers with focused data. The visibility will also help operators in their pricing strategies as they will be able to see individual customers impact on costs and profit margin. The data and analytics can even become a value-added service in its own right. With an easy-to-use interface, the best AIOps tools can provide customers with self-service opportunities for interactive views.
The data and analytics generated by the AIOPs solutions should also be made available and approachable to the operator’s internal sales and marketing teams which can also help to improve the return on investment (ROI). The extracted data and generated analytics from AIOps can provide operators with the right tools to optimize their network planning and thus, reduce network costs.
Thanks to AIOPs solutions’ capability to identify customers throughout the traffic routes, operators equipped with such data analytics cannot only optimize the traffic from OTTs and CDN’s but also negotiate better terms and contracts from a stronger, data-supported position. It can also enable operators to implement fair-use policies and to create service packages that are founded on their own usage patterns.
AIOps’ role in improving customer experience is the most obvious in the domains of service assurance, performance management, and network defense. For example, with the thorough data visibility and analytics, operators that are equipped with AIOps, can vastly reduce the time it takes to identify and resolve issues when failures in the network occur. Furthermore, robust machine learning capabilities also help the network operators get better at predicting and preventing failures from recurring.