Cloud-based AIOps take Wi-Fi from a best-effort art to a highly reliable network science, which is essential for meeting the stringent service-level requirements of digital-age applications.
Wireless networks have evolved into an indispensable part of daily life. Sophisticated enterprise use cases are driving up both service-level expectations and wireless complexity, as IT teams struggle to keep pace with double-digit network growth, the latest RF technology, and the strict uptime and latency demands of modern applications.
Easing network complexity while improving performance is one aspect of the digital transformation initiatives many businesses have implemented to derive greater value from their IT infrastructures. For its part, the wireless industry has begun applying artificial intelligence (AI) and machine learning (ML) to Wi-Fi environments to resolve age-old RF configuration challenges, streamline operations, and deliver predictable, high-availability service levels.
AI Redefines RF Operations
AI operations (AIOps) platforms now available for use with the latest generation of Wi-Fi networks achieve these outcomes by aggregating billions of network- and client-side telemetric data points about user and device experience levels. The systems crunch that data in the cloud, applying ML algorithms to it and learning the dynamic network and traffic behaviors of every connected device. Using these insights, for example, AIOps platforms automatically and quickly identify anomalies, classify them in order of severity, determine their root causes, and provide remediation recommendations through a unified dashboard.
Without AI/ML capabilities, this level of proactive Wi-Fi troubleshooting and resolution is impossible, putting service levels at risk. Traditional wireless triage typically takes anywhere from hours to days or even weeks, leaving precious little time for wireless personnel to accomplish more than keeping the proverbial lights on. Wi-Fi AIOps, by contrast, offload the tedious task of identifying and fixing complex network issues from IT staff while accelerating resolution time.
Case in point: A large Wi-Fi managed service provider (MSP) using a RUCKUS Networks Wi-Fi AIOps platform reported a 67% reduction in mean time to resolution (MTTR), 60% savings in professional IT time, and a 50% decrease in the time required to train new IT hires through its use of automation. The MSP also said its AIOps platform allowed it to improve its Wi-Fi service levels, which it credits for reducing customer churn by an impressive 80% in the year following deployment.
AI and ML boost service uptime and performance beyond what has been possible in the past by continually auto-tuning access points, network-wide, for optimum channel use, aggregation, and transmit power. This automation is quickly becoming a necessity for supporting the real-time application requirements of IoT, robotics, augmented reality, and other modern applications.
Revolutionizing RRM Effectiveness
With AI/ML-driven operations, the wireless industry has finally cut the Gordian knot that has long challenged effective Wi-Fi radio resource management (RRM): how to balance channel plan, channel bandwidth, and transmit power settings in a way that minimizes co-channel interference while maximizing network coverage and capacity. Solving this problem is critical in Wi-Fi, because the technology operates in unlicensed, shared frequency bands. In unlicensed bands, no single network operator has exclusive control or use of the spectrum, so interference among devices run by different entities is an issue that must be managed.
Historically, wireless network planners have labored to strike just the right configuration balance for their unlicensed 2.4-GHz networks, which contain three non-overlapping 20-MHz channels, and their 5-GHz networks, which have 25. Channels in the 5-GHz band may be combined for greater bandwidth per channel by aggregating 12 channels at 40-MHz each, six 80-MHz channels, or two 160-MHz channels. The trade-off is that having fewer channels with more capacity increases the interference risk; opting for more channels, each with less capacity, better protects against interference but reduces available bandwidth.
Transmit power is another consideration. Wi-Fi access points can be adjusted downward to a lower transmission power, which reduces interference and increases channel reuse. Again, however, lower power comes at the expense of less data throughput.
These considerations became even more complicated when Wi-Fi expanded into the 6-GHz band with the Wi-Fi 6E standard, first implemented in 2021. Wi-Fi 6E made 59 more 6-GHz channels available for an additional 1200-MHz of channel capacity, adding five channel bandwidth options of up to 320-MHz to the pile of Wi-Fi planning considerations. Now another new variable with trade-offs to consider is coming, first in the U.S. An automated frequency coordination (AFC) system is projected to be approved for the 6-GHz band later this year by the FCC’s Office of Engineering and Technology (OET). In addition to helping coordinate spectrum usage in Wi-Fi 6E and forthcoming Wi-Fi 7 networks for improved interference avoidance, AFC will provide the option for 6-GHz Wi-Fi devices to transmit at a higher power but with about half the number of available channels.
Scaling Beyond Human Limitations
These many wireless permutations will only increase as technology evolves. Determining the optimum configuration, access point by access point, is surpassing the ability of mere mortals—even the best RF engineers in the business—in large and dynamic networks. AI-driven RRM capabilities are stepping in to enable operators to leverage network and client intelligence to optimally configure, continually tune, and proactively troubleshoot each network-wide access point, a level of wireless automation and accuracy that was unthinkable just two years ago.
Business and IT decision-makers will require AI-powered networks to keep up with Wi-Fi network growth, complexity, and the service-level requirements of dynamic wireless applications in the digital enterprise. As wireless network datasets get larger and continue to inform AI/ML algorithms, new capabilities will emerge to create and hone autonomous, purpose-driven AIOps. These platforms will continue to transform channel, bandwidth, and power configuration from what has been something of a best-effort art into a predictable, measurable, and automated network science.
________________________________________________________________________
Read how enterprises and operators are using RUCKUS AI-driven analytics to better meet their SLAs and deliver a better customer experience, and how they’re using the new RUCKUS One – an AI-driven cloud-native platform that provides network assurance, service delivery and business intelligence in a unified dashboard to simplify converged network management across multi-access public and private networks.
