With today’s high-tech world progressing so swiftly, closed-loop assurance is becoming more and more popular—and for good reason: it is the key to unlocking autonomous operations. Now, more than ever, companies are utilizing closed-loop systems.

But do they really? Are their systems closed-loop? Are they fully autonomous?

Let’s find out.

In general, a ‘closed-loop’ system is one that self-adjusts based on feedback. One classic example is your car’s anti-lock brakes. They require no human intervention—you simply push down the brake pedal as hard as possible, and then the closed-loop process for the anti-lock brake system will handle pulsing the brakes without locking them. This will allow the car to stop most efficiently.

Information is required to accomplish this task, with a wheel speed-sensor used to gather the necessary data. The small circuit processes the logic and has the ability to modulate the brake pressure based on the outcome. Applying business or operational logic to telemetry is how closed-loop systems decide to make adjustments. The logic is straightforward: If the wheel is not skidding, the brakes should be allowed to have their full effect. However, if the wheel stops and skids, the brake modulates its pressure just enough for it to start rotating normally again.

By repeating measurement, analysis, and intervention in a regular and rapid way, the system automatically discerns and maintains the most efficient level of braking regardless of road conditions. As a driver, you simply ‘stand on it’ and go along for the ride.

When looking to apply closed-loop operations to such complex systems, it is useful to examine a basic system to reinforce our understanding. Here, we have information (data in context), we have logic, and we have a technique to autonomously take action based on that logic. This is the minimum requirement of a ‘closed-loop.’

Note that there are no people involved in a closed-loop cycle. If someone is needed for the acquisition of information, the decision to act, or if they are required for action to occur, then it is not a closed-loop. No if’s, and’s, or but’s about it. A closed-loop system must be autonomous, with no people necessary. It is self-contained and self-sufficient.

This may be obvious, but with the industry applying a multitude of approaches to improving assurance, there is a fair amount of confusion around what each should and can do. In an effort to simplify things, we will elaborate on each of the today’s major assurance trends:

1. Analytics

1. Machine Learning

1. Closed-Loop Assurance

Particularly in the service provider space (and business in general), there is a ton of data—an overflow of data, in most cases. This overflow also lacks the context that Service Providers need to make sense of it. There are plenty of alarms and plenty of data points, but it is extremely difficult to know where they fall on the map of systems and how the individual alarms and data points relate to one another. This reinforces the reasons why ‘context’ is so critical—it is the bridge that turns data into something more valuable.  

The first two trends—Analytics and Machine Learning—represent ways to transform large amounts of data into more meaningful information, generally known as ‘insights.’ Insights can be useful in a variety of ways, such as troubleshooting, repair, or for optimization. Overall, these platforms help to re-apply context to the data and translate it into useful information.

It is important to note that Analytics and Machine Learning platforms are not closed-loop assurance platforms. While they can ultimately provide insights to closed-loop systems, today they mainly give insights to human operators, who then fit the insights into their context—and then take decisions and act. With humans making or breaking this process, these systems do not self-adjust. This is a major weakness for our industry; even though insights can be provided, Analytics and Machine Learning still do not have enough context to make decisions on their own, let alone the capability to act on those decisions. In short, they can help figure out what the problem may be, but they cannot come to a decision and cannot program the other machines.

Closed-Loop Assurance is different. Unlike Analytics and Machine Learning, a CLA system has clear operational context plus business or operational logic for making decisions, so no additional human context or decision making is required. As is the case in our anti-lock brake example, the CLA platform knows what systems it is working with, as well as what information is relevant to those systems. In addition, the platform has the business logic to make decisions, and can execute those decisions directly. This system self-adjusts.

On top of that, closed-loop platforms enhance efficiency. Assurance is a daily manual operation—think about how much your company can save by using automated closed-loop assurance. With manual actions largely eliminated, the answer is a lot, and this is savings that cannot be achieved by Analytics and ML systems alone. Significantly lower assurance costs is CLA’s domain. Analytics and ML systems are incapable of carrying out their own recommendations without human involvement.

The diagram above illustrates just how many manual actions can be removed using closed-loop platforms. This is why we are seeing so many customers that are looking to automate all Tier 1 and Tier 2 operations, or all ‘business as usual’ (BAU) functions so they can focus on innovation and growth.

An assurance revolution is underway and Closed-Loop Assurance is at the center of it. Analytics and Machine Learning are useful too, but if you want you services to truly care for themselves, Closed-Loop Assurance is the answer.