Telecom industry is facing an overarching demand for wireless bandwidth in presence of a burgeoning Internet of Things (IoT) industry that promises to accelerate the “experience economy” by bringing people, places and businesses together at scale.
A recent report derived from Open Signal revealed how Airtel surged ahead of the market leader in subcontinent, Reliance Jio, in five out of six experience categories. This serves as a strong testimonial for laying an early foundation for 5G experience. With the onset of COVID-19 epidemic, telecom companies are further challenged to meet the surge and sudden shift in user experience for collaboration tools and high volume of video calls.
A next-generation ground truth mining solution providing KPIs for 5G experience planning
I led a small team to envision, design and implement a 5G experience planning project with applications of Deep Learning over modern data architectures like Snowflake or a cloud-native Data Lake powered by User Defined Functions (UDFs) that would allow for data curation and labeling. We aspired to build a Neural Net that could predict 5G hotspots in presence of fuzzy, uncertain and non-digitized information in a short period of time. The idea was to build a continuous intelligence rendering solution that would provide the necessary KPIs for movable placement and planning of expensive 5G infrastructure.
With the objective of achieving 5G experience consistency in shifting markets, the solution needed to have two key aspects. It should provide a way to save one or more location or consumer behavior-based calculations (e.g. distance to nearest infrastructure, income levels etc.) as ‘queryable’ tasks, aka UDFs. And it should be able to apply Deep Learning to suggest positional, economic and demographic attributes that would recommend next generation of locations hotspots; thereby providing the similarity in quality of 5G experience regardless of the subsequent metamorphic phase of market and consumer behavior.
Results were surprising and real.
This approach helped identify 6% additional hotspots where 5G experience gaps prevailed within just one county of the state of Texas that included 30,000 odd location points and more than 30 UDF-extracted or labeled attributes attached against each of them. The architecture, as described below, also provided room for consumption and labeling of new generation of information emerging from the rollout.
A modern data architecture is the hallmark for digital innovation and user experience planning where operational and business decision-making can be automated by artificial intelligence (AI) based extraction of inferences from raw information. This form of next generation ground truth mining in presence of unpredictable market conditions, will shape the future of “experience economy” in this new decade.