Several years ago, we started to develop artificial intelligence (AI) algorithms that discovered signal fading events such as rain, multipath or line-of-sight blockage in microwave links. These algorithms were then applied to live data from large microwave networks, making it possible to identify the root cause of fading events causing degradations in the network in real time. The algorithms also made it possible to take preventive action to avoid capacity degradations in the future. AI was doing the work for us.
The second stage of our work involved training the algorithms to detect interference, as well as the impact of unstable antennas and wet snow, and connect this to the decline in network performance.
Testing algorithms on live networks
To further enhance this development program, we have involved some of our customers in algorithm trials on their networks. For example, in Stockholm we found that there were sporadic capacity drops between 07.00 and 16.00 on weekdays caused by construction work being carried out in between the microwave radios. Without the algorithms doing their job, these sporadic capacity drops might well have been reported as intermittent radio problems. Service engineers would then have visited the sites and replaced the radios with new ones, without seeing any improvement. Later on, at the repair center, a fault analysis would have been performed on the radios that had been exchanged, but no problems would have been identified.
Time and money can now be saved through the remote confirmation of the true cause of the capacity drops: the ongoing construction work rather than faulty equipment.
Mitigating the impact of a storm
During stormy weather, we could see in our on-screen map view of the network how different links were affected. Most of the links were up and running as normal after the storm, but the signal quality of one of them had been reduced (even though it still managed to deliver the capacity expected). This meant there was a risk of losing this link altogether during the next storm.
In this case, the alignment of the antenna had been impacted, and we were able to proactively plan maintenance and realignment to avoid problems in harsh weather conditions in the future.
Complex networks require speedier issue detection
With denser networks, high spectrum usage coupled with the upcoming introduction of Wi-Fi 6E in some geographical regions, there will be an even more pressing need to detect interference fast in fixed service networks. We have found that we can both identify the interferer and indicate the area of its location using our algorithms.
Introducing Advanced Microwave Insights
After five years of research and development work, we have now presented the first generation of our cloud-based AI-powered network analysis product, Advanced Microwave Insights . This product uses AI and machine learning to help service providers with near real-time root-cause analysis of any transmission issues affecting their MINI-LINK network. Advanced Microwave Insights saves service providers time that they would otherwise spend analyzing transmission issues, and it also reduces unnecessary site visits.
The use of AI and machine learning to improve the performance of microwave links has just started. While it already makes the management of MINI-LINK networks more efficient, it will continue to enhance our ways of working as we progress towards recommendations and self-healing systems.
Other related content
"network" - Google News
September 18, 2020 at 02:05PM
https://ift.tt/3hIaT5e
How AI can secure optimal network performance - Ericsson
"network" - Google News
https://ift.tt/2v9ojEM
https://ift.tt/2KVQLik
Bagikan Berita Ini
0 Response to "How AI can secure optimal network performance - Ericsson"
Post a Comment