Rolling out stronger with the AI power

Telcos must utilise Artificial Intelligence to launch 5G-enabled services that can meet the evolving needs of enterprise customers

5G promises speeds of up to 1 GBPS in India. It’s a game changer in communications. As an equally exciting technology, Artificial Intelligence (AI) will play an enabling role for 5G telecom companies and their enterprise customers.

Telecom companies will benefit from the insights that AI brings in during deployment and monitoring, while enterprise customers will leverage 5G Multi-access Edge Compute (MEC) and AI capabilities to deploy revolutionary applications.

5G planning and monitoring

Nationwide 5G deployment is an expensive and time-consuming process. For telecom companies, it is important to deploy the coverage in a manner that creates the greatest impact on demand and customer sentiment. Traditionally, the rollout was prioritised based on historical customer density. However, this approach ignored the impact on sentiment, churn, and potential customer expansion. Telcos are now deploying AI models to maximise their competitive advantage. Including external data such as socioeconomic variables, market distribution, sentiment, churn, and propensity to pay for rollout planning helps prioritise rollout on a granular regional level and increase ROI from the network upgrade.

Network deployment impact needs to be measured across telephony and data KPIs. Telecom providers leverage Operations Support Systems (OSS) to collect KPIs from every node of the deployed 5G network. Comparable data is available from the 4G LTE nodes. Rule-based AI is used to calculate scores for telephony and data components of the network, such as setup times, throughput, and latency. The weighted source of truth provides a single quantifiable metric that can be explored in more granularity as desired.

The comparison of performance before the 5G deployment and after is called pre-post analysis. With the analytics software, pre-post analysis can be conducted with statistical significance to calculate confidence in the computed differential. Additionally, AI can readily determine the cause of the differential at the base KPI levels. This analysis is called attribution. With attribution, the operations team significantly reduces diagnosis time.

The data stack as explained in ‘Hierarchy of performance and quality indicators’ helps compute the network quality score. However, customer sentiment is another metric to measure the rollout’s impact. NPS stands for Net Promoter Score. It’s a customer satisfaction benchmark that measures how likely your customers are to recommend the business. A correlation can be created between the network quality score and the promoter score to observe the ROI from 5G deployment in each region.

Rule-based AI can be used to calculate scores for telephony and data components of the network, such as setup times, throughput, and latency.

Visualisation is key in monitoring the network health across regions. Exploratory Data Analysis, a key step in the AI journey, is applied to the OSS dataset to build a dashboard. The dashboard displays visualisations, correlations, and comparisons between quality indicators across regions.

5G MEC utilisation and application support

With 5G implementation, an exciting add-on is MEC. Edge Compute is the collection of cloud instances on network nodes near the user that help avoid latencies caused by communication with the central cloud. Telecom companies have partnered with cloud providers such as AWS, Azure, GCP, and IBM to roll out nationwide MEC.

AI will help the MEC ecosystem in multiple dimensions. Third-party Developers will need to develop low-latency, high-throughput applications. Real-time compute-friendly SDKs will be provided for technologies such as Machine Learning, Natural Language Processing, computer vision, remote (3D) rendering, customer support (conversational AI), and content delivery. These will enable exciting applications such as Immersive Bots that will allow conversations via text, speech, and movement of 3D models.

Once an application is developed, it will need to be deployed on certain instances. MEC is more expensive than the regional cloud computing counterpart. AI will forecast the traffic from different regions to help select relevant locations. Once launched, the app will provide traffic and performance statistics analytics through a MEC management console with redeployment recommendations.

Similarly, deployment of the application on the MEC will help in real-time data collection enabling the monitoring of the network and compute (CPU/GPU) usage. When devices need to connect to a nearby instance, an AI-powered discovery service will evaluate the optimal instance based on proximity and availability. Optionally, additional parameters such as network plan, subscription level, and device may be considered at the application developers’ discretion via a config.

In conclusion, the rollout of 5G technology in India presents a significant opportunity for telcos to enhance their enterprise services offerings and stay ahead of the competition. AI has the potential to play a critical role in enabling telcos to make the most of the capabilities of 5G and meet the evolving needs of enterprise customers. By optimising their network infrastructure, enhancing customer experience management, offering customised services, ensuring network security, and automating network operations, telcos can provide more value to their enterprise customers and position themselves as leaders in the 5G space.

By Anand Mahurkar

Anand Mahurkar is the Founder and CEO of Findability Sciences

feedbackvnd@cybermedia.co.in