Next Generation Centralized SON Drives Customer Experience through AI Powered Network Automation

The advent of 5G is upon us and it is the cornerstone for digitalization of industries and building a connected society. In order to prepare for 5G, mobile operators are densifying their networks to meet the demand for massive and low-latency mobile connectivity. Carrier and site additions for already equipped, high traffic locations are the quickest way to drive network densification, yet not the most effective and sustainable ones.

An important success factor for 5G transformation is making the mobile network smarter and more customer centric. Associated investments bring a significant cost burden and operational workload. To address these challenges, P.I. Works proposes use of automation for the planning and deployment of new carriers and sites that is built on machine learning algorithms to analyze the network behavior and take predictive actions.

As an enabler of AI based mobile network management, P.I. Works contributed to the research paper, “Prediction of Active User Equipment Number with Bayesian Neural Networks for Self-Organizing LTE Networks”, which won the best paper award on Network of Future Conference in London. This research primarily offers blending of Bayesian Neural Networks method into our SON (Self Organizing Networks) powered Energy Saving solution. Conventional SON solutions take improvement actions in a reactive manner. They take automated actions based on predefined rule-sets and thresholds for each specific network problem. As outlined in the paper, we’ve taken a massive step-forward and proposed an artificial intelligence (AI) powered SON solution that takes unsupervised and predictive improvement actions based on the customer behavior and network traffic pattern.

The quality prediction of customer behavior relies on the strong analysis of the historical KPI (Key Performance Indicators) values. Formerly, operators used to deploy and configure new cell sites based on the network traffic during the busiest hours in any location. This approach inevitably leads to unnecessary power consumption during the rest of the day due to lower network traffic and capacity demand. To address this challenge, P.I. Works solution correlates data from multiple sources and builds relationships among the KPI values on a real-time basis so to take more accurate actions with respect to capacity utilization and Quality of Experience. AI powered KPI prediction enables OpEx savings by reducing power consumption through dynamic management of number of active carriers based on the traffic load prediction. At the same time, preemptive fault detection capability of the solution ensures subscriber experience. 

P.I. Works is highly committed to revolutionizing the way mobile networks are managed. We equip mobile operators with leading-edge, AI-driven solutions to support them during their 5G journey. For more information about the research paper and award, please refer to: http://ieeexplore.ieee.org/document/8251223/