3 Challenges That Virtualization and Machine Learning Are Called To Solve

The notion that the telecom industry is at a crucial time is not new. It is a time where the network owners are watching revenues falling. Traditional services, like voice and SMS, are in decline and no new Telco products are on the horizon. Costs are rising due to legacy infrastructure combined with a high demand for connectivity and bandwidth, thanks to the exponential growth of digital services.

At the same time, software companies such as Amazon, Netflix, Facebook, Google, among others, are jeopardizing Telcos. Agile by nature, these software giants are overlapping their sandbox with the telecom sandbox and blurring the lines that divide them. To stay alive, Communication Service Providers have to become Digital Service Providers. In other words, they need to embrace the Digital Transformation. This means adopting transformational changes in their core, the way they operate, the way do business… They need to become agile software-like companies.

For their digital transformation potential, Virtualization and Artificial Intelligence (specifically Machine Learning and Deep Learning) are hot topics today. Here are three obstacles that these two technologies are called to resolve:

Three Challenges Virtualization and Machine Learning are Called to Solve

1 – Adaptive Dynamic Bandwidth

Along with the boom of internet traffic and the amount connected devices, one must talk about how to scale the network to support colossal volume. Legacy networks were not designed to handle this. Right now, escalating resources needed to deliver services is quite linear and too expensive. Imagine, for a second, the current static bandwidth assignment by device multiplied by 1000 with an IoT explosion! The marginal revenue compared with the cost incurred drop business margins to nearly zero – or even worse – to negative.

The need to assign resources automatically based on what the service needs in real time is key. Virtualized technologies such as NFV and SDN allows to manage the network resource as needed, enabling – in a matter of seconds – new circuits, added capacity… you name it.

Thanks to Deep Learning, the network will constantly be tested and the behavior measured over the resources assigned. Once the network has learned the exact service level a device needs, it can automatically assign it, freeing unused ones and maintaining the resource utilization as efficient as possible.

2- Self-Optimized Network

Let’s picture a VNF/SDN scenario fully implemented. Control is decoupled from the physical infrastructure and running in white boxes. Network functions are virtualized, and the IT world and Network will converge to a single cloud-based virtualized infrastructure. Management and assignment of all the virtual resources elements – ranging from average vCPU and vRAM, vDISK, vNIC, to vEPC (Virtual Evolved Packet Core) – will be assigned and reassigned based on the needs of the network.

Machine Learning techniques will learn the current state of the network and IT, and make the adjustments necessary to add/remove/deploy resources to support the ongoing demand. For example, adding the resources needed to specific vEPC elements to support the high mobile requirements of a sports event.

3 – Self-healing Network

In a virtualized network, in which software makes most actions and decisions – not specialized hardware – pro-active bots (running artificial intelligence techniques) can scour the network, services, and devices to recognize fault signatures and proactively apply fixes to overcome issues before they impact customers.

What’s more, these Smart Data techniques will enable the network itself to correlate and connect seemly unrelated data collected from all sources available. Operators will be able to:

• Identify potential issues
• Fully automate operational decisions
• And, proactively take steps to fix these problems, improving the resolution time dramatically

All this is just the tip of the iceberg. There are great challenges ahead, but we can look forward to a future where connected devices and services will be well beyond human intervention and machines will have to step up. It will be a great time to be around and watch. It will be amazing.