Defining autonomous networks and the business case for them

Image by ipopba

‘Autonomous network’ means different things to different people, and in this article, the phrase refers to the ability of the network to operate on its own, without human intervention. Other phrases like self-organizing or self-optimizing network, will typically mean a subset of autonomous networks.  Any reference to autonomous networks imply key characteristics – scale, zero-touch, visibility, data-driven, intelligence, and above all, customer experience. Each of these words relate to some aspect of the network being autonomous.

Scale refers to the ability to dynamically scale-up and -down the capacity of the network, on-demand, automatically. This is primarily enabled by virtualization of network functions and use of commodity infrastructure. Network virtualization provides the flexibility of choice, in terms of vendor, technology and service provider.

Zero-touch – Configuring services on the network without human intervention by automating provisioning and configuration activities. It is also important to automate orchestration across multiple layers and network domains. Errors and fallouts are handled by a zero-touch system enabled by close-loop automation. Zero-touch also demands usage of agile methodologies in network operations.

Visibility – what you see in the inventory data is actually present and configured in the network, providing accurate data to the user, and available on the fly. The user here could be any of the network operations functions or simply, a software or a BOT.

Data-driven – Any decision taken as part of the network operations is driven by data derived from various parts of the network. A good analytics engine, which can analyze both structured data from various databases and unstructured data received from the network elements, shall provide all the data required to drive the decision.

Intelligence – Intelligent decisions come from applying the collected data onto the AI, and enabling machine learning on the network information. Most of the actions performed by rule-based engines are now performed by a cognitive AI engine, leading to more informed and deeper decision making. In a fault situation, it enables proactive identification and cross-correlation of events faster and enables automatic resolution. Additionally, intelligence is a crucial part of predicting and preventing unwanted events.

Experience – All of the above tenets lead to one key, yet common, goal – provide services to customers in real-time. Customers expect that their requests be delivered as requested, without delay or further clarifications. Being able to deliver the expected experience in real-time means that a great volume of demands can be handled and met soon, if not immediately after they are communicated.

The road to autonomy

Achieving these tenets in the network would mean uplifting or transforming the existing network management software. The required changes cut across multiple systems covering order management, provisioning, activation, inventory management, traditional FCAPS and so on.

This means that the operator needs to spend capital expenditure (CapEx) on replacing many of the existing systems with automated intelligent orchestration, inventory federation, network analytics and network management systems. Such systems would be over and above what is necessary for a good amount of virtualization in the network, but devising and implementing them could prove time consuming and expensive.

Any rewards to reap?

The key question is whether there is a return for the investments in network virtualization, intelligence and automation. While the operator is moving towards 5G roll-out, autonomous networks become a basic necessity owing to the volume and complexity of operations. However, outside of 5G deployment, let us examine the all-round impact and the returns thereof. The areas impacted can be classified under one of the following:

  • Cost of service delivery, network operations, downtime;
  • Opportunity cost due to delay in capacity provisioning and obsolescence;
  • Revenue due to faster service delivery and enhanced experience.

In most networks, activities such as planning, provisioning, service delivery and network operations is done manually. A typical network team has people with multiple skills in varied levels of expertise. The fact that autonomous networks can handle service provisioning, scale-up & -down, and resolve network faults without human intervention, eliminates the people costs associated, including aspects such as training, cost of errors, working facilities etc.

Apart from the direct human cost, every fault in the network that could be an alarm, performance issue, capacity overrun, adds to the cost of downtime, service penalty and cost of reputation. Intelligent networks can predict, and many times prevent a fault from occurring. Hence, there are direct and indirect savings that can be accrued due to autonomous networks.

There are costs that can be avoided by implementing autonomous networks; loss of revenue or service downtime due to temporary shortage of network or compute capacity being examples. With next-gen applications utilizing video streaming, AR/VR/MR, real-time interactions etc., there are demands for a higher capacity, lower latency or other type of service demands, provisioned dynamically for use in a window of time. With virtualization, it is possible to dynamically scale the network and compute resources based on the demand at any point in time and that can solve the issue of temporary or last-minute capacity need. Similarly, the virtual network function and associated infrastructure can be upgraded continuously and hence, the cost of obsolescence is avoided.

The fact that demand-capacity cycle is automated end-to-end, and the service delivery is orchestrated means that the services can be provisioned earlier than before. This advances the revenue inflow to the operator. As services are automatically provisioned and managed, this could fetch a premium for additional resources, or special type of demands, or last-minute changes or better QoS. The ability to integrate CI-CD tools and agile methodology into network domain has created automated deployment and thus, reduce time to market. This again positively impacts the revenue of the operator.

While there seems to be definite returns for autonomous networks, one needs to look at the overall cost of implementing the solutions and match it against the cost / revenue across various heads to arrive at the actual business case. 5G deployment related mandates surely add to the returns.

Written by Viswanathan Ramaswamy, Vice President & 5G Business Head at Wipro, one of the guest speakers at TM Forum’s Digital Transformation Asia in Kuala Lumpur this week. This article was originally published on TM Forum’s Inform site in the run up to Digital Transformation Asia.

Be the first to comment

What do you think?

This site uses Akismet to reduce spam. Learn how your comment data is processed.