The process of automation in telecom may have begun many decades ago with, for instance, manual switchboards replaced by automated ones almost a century ago. Since then, automation has been a natural way to improve performance and take cost out of the operation. However, the scale of automation currently underway is drastically more significant, with dramatically higher stakes.
Automated operation today is not just about building a better version of the traditional way to run business. Instead, automation needs to turn the business logic around and excel in responding with what’s needed when it is needed and to do this for each and every customer individually.
Key automation ingredients are intelligence and programmability. Intelligence needs to be smart enough to understand and anticipate the needs of customers as well as the state of the network and the resources available. Programmability needs to be powerful enough to effectively co-ordinate responses and adaptions in highly efficient ways, so that customer demands are met in real time.
In telecom, just as in any other industry with an accelerating digital business agenda, automation is at the center of this development. The ability to realize a digital operation that effectively runs a “real-time” business will be critical not only for individual operators but also for the industry as a whole. Only with a telecom industry that is fit to participate in the new market realities of the rising digital economy will current opportunities be turned into a much-needed business-growth agenda.
Automation is a force that we see in many industries and situations. Automation is no doubt driven by technology progress, and the lesson so far is that we need to see beyond what's currently possible to envision the full impact of automation. Just think of the self-driving car as an example. Ten or fifteen years ago, most scientists agreed that self-driving cars would not be possible because the environment was simply too complex and unpredictable.
Now, the self-driving car is in all major carmakers' roadmaps, and we have several more or less fully automated vehicles out driving on the streets in numerous trials. For another complex example, we might look at telecom networks; Telecom network automation will reach very far and do this faster than we naturally envision.
How can you build trust in self-operating networks?
The car industry is interesting because in addition to developing the self-driving car, the industry also has also succeeded in establishing a straightforward industry model of the different steps in automation and a vision for how these steps can develop independently as well as how they align to realize the vision of the self-driving car.
The model takes us from the initial step of assisted driving—such as with a parking assistant or an automatic gear box—where the driver is in full control, all the way to the point where, with self-driving cars, the driver is not in control and human interaction is limited to specifying the destination. A key aspect in the model is the matter of confidence and trust. Of course, riding in a self-driving car requires trust. Trust is built over time with a gradual introduction of more and more sophisticated technology and increasingly complex use cases.
What we can learn from this framework and get inspired from is that we need to develop an equally simple and robust automation framework for the telecom industry. With self-operating networks, we aim for full automation but acknowledge that mission-critical automation, just as in the case of self-driving cars, comes in phases where each step adds direct benefits as well as takes us closer to the end goal with confidence.
To succeed, we need to see telecom automation as a multifaceted development with parallel progress across all fields of operation and not as a strict implementation where different levels are completed one by one. Still, to reach a high level of automation, it will be critical to have a framework where automation capabilities can be combined to create a reinforcing effect when control mechanisms driven by different purposes are applied and are increasingly combined. For instance, a radio network can be self-tuning on the basis of domain-specific data and configured to achieve optimal use of a spectrum or other resources.
The same mechanisms can also be applied for an optimization objective defined at a higher level of abstraction. For instance, when the anticipated satisfaction from a customer in a given situation defines the tuning perspective, it can lead to a quite different outcome using the same mechanisms. To effectively capture these complementary and reinforcing dimensions, it makes sense to structure automation progression with different levels that capture a gradual increase in the level of abstraction.
Five levels of automation
- The first and most straightforward level of automation is task-centric. In a sense, it represents the ever-ongoing ambition to do less with more and to streamline workflows. As such, the concept of task-centric automation has been on the agenda for a long time. It is a priority across all operational activities in business, and the prime objective is often cost reduction but also shorter execution times with an expected result. An example of task-centric automation is the migration configuration data from thousands of radio nodes in a short time and with a high degree of automation.
- The second level would be node- or domain-centric. Here, automation has been in focus for a long time. More recently, advanced closed-loop automation has matured to a level where complex activities are fully handled in automatically and with gradual learning execution. This translates to a continuous collection of performance data as well as fast and relevant analysis to determine the required actions and control mechanisms that are required to tune and update parameters and settings.
Automation ranges across deployment, configuration, and optimization as well as fault and healing situations. More specifically, it requires automation of neighbor planning and optimization such as addition/deletion/reordering, automatic coverage/capacity optimization, and fully automatic and continuous analysis and change proposals for network tuning. These are solutions that have proven to bring great improvements both in reduced labour intensity and asset utilization.
- The third level would be service-centric automation with services increasingly configured to individual customers' unique requirements and expectations of instant realization. In the future, service lifecycle management needs to be fully automated. Furthermore, automation needs to interact with the underlaying virtual and physical infrastructure in such a way that the resources are managed in an efficient way and customer agreements are effectively realized.
The automation needs to span design, orchestration, and assurance of services. An example here is how enterprise customers are provided with powerful platforms for rapidly configuring and managing services according to their preferences, and they are provided with a fully automated orchestration of on-boarding and validation of the required resources—all done in a fraction of time traditionally required. In addition, closed-loop policies drive assurance and provide the required healing to realize defined service-level performance.
- The fourth level is centered around user or customer automation. This is grounded in a deep understanding of what individual customers demand, how they perceive what they get, and what they will need in the future. In a market where no customer is like another and where a standard mass-market offering is no option, customers will expect well-thought-through recommendations to guide their consumption in real time. The automation journey starts with the ability to understand and translate this understanding into actions has a deep impact on operation priorities as well as on customer targeting and marketing.
- The fifth and final level is business automation where business value is at the center for automation. This is the equivalent to a self-driving car that is controlled only by means of giving instructions about where to go and then the “system” takes care of the rest. For telecom operations, this step relates to having business KPIs to control and guide the automated operations. Maximizing return on assets or customer acquisitions, optimizing customer satisfaction, or minimizing infrastructure usage could be some of the parameters by which future business operations are controlled and fine-tuned.
The vision of the self-operating network will provide improvements across a number of current business pain-points and it clearly holds a promise to realize better ways to run business, and these are important drivers. However, the real opportunity lies in using automation to tap into the digital economy and develop the capabilities that future customers will demand. This will require an automation framework that ultimately ties business objectives with customer needs and establishes a "real-time" operations model across domains, services, and customers.
Want to learn more about Ericsson’s key OSS architecture principles and vision?