By: Manoj Kumar, Product Manager in GBS

As communication service providers (CSPs) leap towards 5G rollouts, there will be huge CAPEX investments in network infrastructure and spectrum licensing. It will therefore be imperative to control ongoing OPEX spend by utilizing automation-led operations. Existing legacy networks such as 3G/4G have a few such capabilities that use self-organizing network (SON) modules that cater to self-configuration, optimizations, and healing. More so, 5G networks will unleash room for automation due to “cloudification” and containerization of network functions.

This will be feasible, given that there will be several new APIs for consumption, thus leading to no deterrence to AI/ML use cases, unlike in legacy networks. However, there are standardizations underway, and several new Cloud-native Network Functions (CNF hereon) or Virtual Network Functions (VNF hereon) entrants have bloated the market. Many new CSPs have forayed into open ecosystems while existing players are looking for a seamless upgrade from 4G to 5G or the co-existence of both technologies until 5G is heavily rolled out.

To tap the full potential of automation within the 5G ecosystem and reduce OPEX, some of the immediate focus areas for CSPs will be – prescriptive networks with advanced SON, cross-vendor automated testing and deployment, and hybrid cloud management and operations with tools/frameworks.

1. Prescriptive networks with advanced SON

5G networks will be complex in terms of spectrum management, resource allocation, network slicing, and interworking with legacy technologies. Considering the various stages from planning to maintenance, it is essential to have closed-loop automation weaved at every stage. For instance, during the planning phase, Radio Access Network (RAN hereon) features can be automatically enabled or disabled depending on the scenario. Once deployed, key feature parameters are adjusted based on the performance threshold. From an Operations & Maintenance (O&M hereon) standpoint, traffic and capacity-based auto infra and software scaling are feasible and limited to not only pre-defined rules and configurations, i.e., OSS should be able to run scenario-based tuning and optimizations using advanced AI/ML models. With ubiquitous coverage of 5G, CSPs may choose to shut down 4G sites. In such an event, 5G sites should be able to quickly augment capacity needs without any manual interventions. As networks run into a prescriptive mode, there will be huge OPEX savings (~30-40%) and an increase in service revenue (~5-10%) due to decreased network congestion during special events.

2. Cross-vendor automated testing and deployment

With many vendors providing 5G VNF/CNFs solutions, it will be important for any CSP to have a partner to automate the testing and scalable deployment of such functions with minimal disruption. During testing, key considerations will be analytics-driven conformance, performance, functional, and integration testing – such that any non-compliance can be quickly resolved. Further, the important element will be a centralized orchestrator with AI capabilities that can help scale and heal across the vendor ecosystems. This will require AI/ML models to predict service usage based on historical data and discover patterns for better management of VNFs/CNFs across multiple partners. An example could be – A CSP has several 5G enterprise customers spanning multiple industry segments, and they are consuming services with a dedicated network slice. Based on the load and changing business triggers, the orchestrator should be able to scale up/down the capacity across multiple vendors VNF/CNFs (RAN, Transport, and Core) without any degradation in the quality of service.

3. Hybrid cloud management and operations

Due to the nature of data processed in the Telco ecosystem, some applications will still reside in regional data centers as per regulatory requirements. This will need efficient and scalable management of the hybrid cloud environment, and it is not wrong to anticipate multiple cloud infra and PaaS/SaaS service providers in a single CSPs network. Utilizing state-of-the-art tools with automated workflows for migration, maintenance, and transformation will provide room for cost savings and faster time-to-market. For example, the possibility of network slicing with 5G will pave the way for industry-specific private 5G networks and network as a service (NaaS) use cases, which will require holistic instantiation of resources across multiple cloud providers. In such a scenario, AI/ML-based capacity management will provide zero-touch service activation in a responsive and automated manner. Further, operations in such scenarios will be complicated. Hence, a tool-based automated approach to scale workloads, manage costs across cloud providers, and automatically update patches will significantly optimize the running costs.

Conclusion

At LTIMindtree, we believe that automation enablers and frameworks will be essential levers for any CSP while looking for system integrators or a professional services partner for 5G implementation. With two decades of experience in the telecom domain, we have helped several network equipment manufacturers (NEM) and CSPs with a differentiated approach during product development, deployment, cloud transformation, application development, and customer services operations. In a recent engagement for a Europe-based large CSP, Mindtree helped automate the SON deployment in 4.5G RAN sharing mode. We built a custom filter and mediation layer that helped collect neighboring cells’ info from other CSPs OSS, thus reducing deployment efforts significantly.

With the proliferation of 5G services, it is evident that there will be new revenue streams for CSPs, with costs also rising proportionally. It will be critical to focus on cost optimization from multiple levers to reap faster ROI. In fact, many CSPs are building in-house 5G capabilities while others are harnessing ecosystem partnerships with NEMs, VNF/CNF providers, cloud players, and application developers to roll out a cost-effective 5G solution.