How AI and ML will help in reducing the number of 5G sites

A telecom operator in Manhattan, New York, has used AI and algorithmic ML automated design processes to provide better 5G coverage and dominance while reducing the number of sites required from 185 to 111.
Huawei 5G base station in China“This reduction provided significant savings while additionally creating optimized coverage,” said Small Cell Forum (SCF) and 5G Americas — in a white paper called Precision Planning for 5G Era Networks with Small Cells.

5G Americas, the industry trade association, earlier announced that mobile connections will be reaching 10 billion by 2023. The global 5G connections are expected to reach 1.3 billion in 2023.

Precision planning process of small cell siting and deployment of Machine Learning (ML) and Artificial Intelligence (AI) in network design can reduce the cost of deployments while optimizing coverage.

The demand for mobile data is driving network densification with the deployment of small cells. Though lower cost than macro towers, the compact, low-power nature of small cells means they also serve a smaller area. This in turn means they need to be located closer to demand hotspots in order to effectively cover the mobile data demands of customers.

Telecom engineers must focus on measurements of network quality, signal strength and quality, traffic patterns, and other topographical considerations for maximizing a network operators’ return on capital investment.

AI and ML models in small cell design and siting efforts can provide optimal coverage and throughput with the most efficient capital investment.

“By using big data analytics, including machine learning, to digitally model specific use cases, will deliver better returns on investment (RoI) for network evolution plans and hence a better business outcome,” Peter Love, 5G Principal Architect at Nokia, said.

Artificial Intelligence and Machine Learning technologies can enable remarkable capital and operational efficiencies, where the design software learns and adapts to draw on many inputs, each providing an immense amount of granular data to inform decisions.

“It is accepted wisdom that HetNets and densification will be the new normal for 5G network rollouts. Automatic design processes specifically for dense urban environments are required to reduce planning time and cost,” Iris Barcia, chief operating officer at Keima, said.

For maximum return on investment, small cells should be placed as close as possible to demand peaks; best practice is within 20-40m.

Network operators would like equipment that estimates location of usage and quality reports to adopt smarter algorithms such as the machine learning approach demonstrated. Median locate errors less than 20m are expected for small cell planning purposes.

Machine learning models should be part of any small cell design effort. Different inputs and assumptions will be factors in the resulting models that are generated.

The aggregation of very large data sets is important to provide algorithms with sufficient test data to inform results. These data sets provide algorithms with information on factors such as power and backhaul availability, signal-to-interference ratio, spectral efficiency, line of sight, traffic estimates, overlapping cell coverage, agreement with site owners, and other considerations.