16 June 2023

Justin Head – Founder, Executive Vice Chairman – PowerX Download whitepaper here

Digital transformation is progressing at speed across South Asia. The region is seeing the biggest increases in mobile internet adoption which accounted for, according to GSMA, more than 50% of new mobile subscribers globally in 2021.

From an infrastructure perspective, we see some of the largest telecom loads reflecting the increasing demand for mobile connectivity and data consumption with ongoing reviews of the infrastructure’s resilience, fit-for-purpose status, efficiency and sustainability. The same questions keep coming up: ‘are mobile sites right-sized for their load?’, ‘are the right assets in the right place?’, ‘with high CAPEX and low returns, can existing assets be better deployed?’ ‘what other energy efficiencies can be achieved to support more sustainable growth?’. These questions are very hard to tackle at the scale of large tower networks with scarce resources. Unlocking the value of the data that still sits underutilised in the passive infrastructure is key to managing the challenges above.

PowerX applies advanced data science, machine learning (ML) and artificial intelligence (AI) in its breadth of solutions for passive tower infrastructure. One of the most frequent questions I’m asked is “what exactly is AI?” and “how specifically will it help better manage mobile towers infrastructure?”

What is Data Science?

The data science space is phenomenally broad. Defining it has become something of a Rorschach Test depending on who’s asking – and who’s answering – the question. What are you looking for? Deep learning? Neural networks? Big data analysis? AGI? ML? AI?

AI is the most well-known, and I find it more productive to first chip away at what AI isn’t, since there are many misconceptions associated with the term that simply get in the way of understanding and utilizing AI’s practical power.

AI is still mistrusted, frequently and recently portrayed as technology that reaches a magical threshold of brute computing power that gains self-awareness, usually with malevolent intentions, or as a tool that will steal your job. Functional AI has nothing to do with the pursuit of machine consciousness or self-awareness. Whilst it will make you more efficient, it won’t steal your job. Instead, leading edge advances in AI are better characterized by their application to diagnose health diseases, predict patient volumes in hospitals or customer demand in call centres. Applications closer to home include predicting power usage in an electrical distribution grid or balancing the load on electricity grids. In our industry, its potential in driving real-time predictive maintenance, power and site efficiency management remains largely untapped.

Data science in the telecom sector

As of today, the telecommunications sector has a somewhat unbalanced adoption of real-time AI technology. Data science and AI is used at scale in call centres, to determine consumers’ propensity to buy and even on 5G active equipment management. On the passive infrastructure side though, the take-up of AI is far behind and yet to reach critical mass. It is incorrectly seen as a tech buzzword, rather than the highly effective tool for delivering tangible business and operational benefits that it is. Aside from early investments in maintenance prediction software, the complex operation of managing thousands of sites spread across large terrains with three or four collocating tenants and huge data demands has, to date, been left behind by this technological wave.

This is why understanding ‘what AI actually is’ becomes critical: it is a set of data science tools which help existing teams drive new efficiencies. It is an augmenter and enricher of people and processes, not a replacement.

So, returning to our earlier question: “What is data science and AI?”, it might be more helpful to frame the question as “What can data science do?” Tower networks produce unimaginably huge troves of data, terabytes of potential insight into day-to-day operations that typically remain unutilised in archived spreadsheets and dormant data banks. With the application of data science, the more data, the higher quality the model, the more evolved, impactful and meaningful the predictions and the resolutions. The benefits are many: improved asset efficiency; reduced diesel, energy and other operational costs; optimised predictive maintenance; increased revenue assurance; improved resilience; and reductions in carbon emissions.

Data science, AI/ML algorithms sift through immense quantities of sterile data and uncover hidden patterns in the micro-operation of a single tower or across a sprawling network. A single generator in Pakistan, for example, may regularly run for 30 minutes after the site starts receiving power from the grid, while a site in Vietnam may consume excess energy due to a faulty rectifier. These inefficiencies, added up over a network, result in thousands of dollars of operational losses that could otherwise be invested in growth. Data science tools embedded in an organization’s processes highlight the most inefficient towers and put a spotlight on the exact issues to be fixed, allowing NOC and operations teams to do what they do best with significantly greater scale, reach and insights.

Real world examples

PowerX was recently installed onto a network of 10,000 towers, and very quickly allowed the team to find over 70,000 previously unseen anomalies on the network. No human – or team of humans – can discover anomalies at that scale and prioritise them fast to ensure the biggest impact on improving operations. PowerX’s AI and data science automation, combines phenomenal data crunching capabilities with pattern recognition algorithms to excavate issues buried deep in the noise. Whether it is detecting rectifier step change problems, reducing avoidable generator run times or spotting inefficient generator loads, data science works for the humans, not as potential replacements. In fact, the insights and alerts generated by AI can increase the efficiency of tower operation teams by a factor of 30 – 50x.

For a sector that has a well-deserved reputation as a technology innovator, the question: ‘should I digitize tower infrastructure using AI?’ cannot be shelved anymore. Embedding data science and AI into operational processes is now essential in order to support the increase in demand for connectivity and mobile data, alongside the need for greater operational efficiency, high resilience and sustainable growth.