Future wireless networks will be more complicated with multiple access networks, frequency bands and cells - all of which will have overlapping coverage areas.
Machine learning (ML) and artificial intelligence (AI) could be the answer for wireless operators looking for help with network planning and deployment challenges, said Omkar Dharmadhikari, wireless architect, CableLabs. ML and AI can address the various challenges by analyzing geographic information, engineering parameters and historic data. This will help operators forecast peak traffic, resource utilization and application types; figure out network parameters for expanding capacity; and use interference and inter-site distance information to fix holes in coverage.
Since 5G allows for simultaneous connections to multiple IoT devices, massive amounts of data are generated. 5G multi-access edge computing (MEC) combined with ML and AI will provide operators the opportunity to offer automation at the network edge, application-based traffic steering, dynamic network slicing, and ML/AI-as-a-service offering.
This works, in part, because 5G uses millimeter-wave, with beam-based cell coverage, Dharmadhikari wrote in a recent blog post. Using a machine-learned algorithm, the 5G cell site can compute a set of candidate beams; the user equipment reports Beam State Information, including parameters such as Beam Index and Beam Reference Signal Received Power (BRSRP). The goal is to find and connect to the best beam.
5G also utilizes a technology called massive MIMO, which utilizes 32 or more logical antenna ports in the base station antenna array. The result is an increase in throughput, network capacity and coverage, and a decrease in interference. Weights are "critical" for maximizing beamforming effect, Dharmadhikari said, and one of the roles ML and AI can play is to dynamically optimize the antenna weights by analyzing historical data.
Along another vein, network slicing will allow for multiple dedicated virtual networks in one common physical infrastructure. Each slice can be managed separately. In this case, ML algoritms and AI can be used to collect real-time data to construct a panoramic data map of each of these slices.
"With future heterogenous wireless networks implemented with varied technologies addressing different use cases providing connectivity to millions of users simultaneously requiring customization per slice and per service, involving large amounts of KPIs to maintain, ML and AI will be an essential and required methodology to be adopted by wireless operators in near future," Dharmadhikari said.
In wireless networks, AI and ML can be embedded in individual edge devices, deployed at the network edge, or built within the system orchestrator for centralized deployment, Dharmadhikari said.
While ML and AI will play a key role in 5G networks, application is still in its infancy, Dharmadhikari said. "The network topology, design and propagation models along with user's mobility and usage patterns in 5G will be complex …. ML and AI can be used to address several use cases to help wireless operators transition from a human management model to self-driven automatic management."