AI-Driven RRM Overview
AI-Driven Radio Resource Management (RRM) is a centralized algorithm that runs in the RUCKUS AI cloud and guarantees zero interfering links for the access points (APs) managed by SmartZone controllers, whenever theoretically achievable, thus minimizing co-channel interference to the lowest level possible. This is accomplished by continuously gathering RF data from all the access points, holistically analyzing the RF environment and usage patterns, and jointly making optimal choices for channel re-use, channel bandwidth, and AP transmit power selection to maximize the network throughput. This new technology relies on sophisticated techniques of machine learning, artificial intelligence, graph algorithms, and cloud scale computation to jointly optimize channel, channel width, and AP transmit power. This method goes beyond adjusting the channel width because it optimizes across all combinations of channel, channel width, and AP transmit power to search for the optimal values across the network.
Benefits of AI-Driven RRM for an End User
With the introduction of AI-Driven RRM, customer Wi-Fi client devices can expect to operate in interference-free RF conditions, when possible, and experience higher throughput and higher reliability with fewer retries and errors. This results in higher user satisfaction when connecting wirelessly to the network. Maximizing the channel bandwidth allocation in the interference-free environment also ensures optimal performance to support high-throughput and low-latency applications.
Advantages of AI-Driven RRM for a Network Administrator
While professional wireless engineers routinely optimize their network performance by selecting channel and power settings in addition to tuning other available configuration settings, this task is getting more difficult with the advent of the 6 GHz spectrum. A glance at the newly introduced spectrum and the available channel and channel width options make it tedious to manually optimize channel and channel width parameters required for a properly tuned Wi-Fi network. Not all enterprises have the wireless RF professionals available to tune these settings across the network. For a busy network administrator, sub-optimal conditions often go undiscovered until an end-user escalation.
With AI-Driven RRM, network conditions are continuously monitored. When there is an opportunity to improve upon a sub-optimal configuration, the network administrator is presented with an optimized choice of channel and channel width in the form of an AI recommendation. With a single click, the network administrator can make changes to the network and apply the most optimal parameters to all the APs in a zone.
Prerequisites for AI-Driven RRM
- AI-Driven RRM requires SmartZone controller release 5.2.1 and later. All access points supported on this release are compatible.
- Zone must be fully licensed.
- Zone must have at least two APs
- Zone and APs should generate sufficient data including channel plan and CCIR for RUCKUS AI to generate recommendation.
How AI-Driven RRM Works
AI-Driven RRM analyzes interference information received from every AP in the network, the user configuration hierarchy, access point capability, historical data about access point radio activity, rogue access points, and traffic patterns to jointly optimize channel, channel width, and AP transmit power.
- What is the recommendation?
- Why is this recommendation being made?
- What are some of the potential trade-offs?
- What is the history of actions taken on this recommendation?
When a network administrator applies a recommendation, a scheduling calender is displayed to select a date and time for the channel and channel width changes to be applied. This gives control to the network administrator to pick off-peak hours when the network is less busy to make the change. Changes are applied at the chosen time using SmartZone API calls to the controller from RUCKUS AI, and the scheduled time is stored in the system and used for future changes to channel and channel width if required.
After an AI-Driven RRM recommendation is scheduled and applied successfully for the first time, the system continuously monitors the network environment for any changes that may cause co-channel interference. Upon detection of such changes, the AI-Driven RRM algorithm automatically applies subsequent recommendations persistently. This automatic process continues indefinitely, unless the user chooses to revert the recommendation manually by selecting the Revert option. If the user reverts the recommendation, the original configuration is restored.
After the AI-Driven RRM recommendation is applied, it is pinned in the list of recommendations. Once the recommendation is applied, the KPI panel moves into the monitoring state for 24 hours where AI-Driven RRM starts gathering feedback about the change that was recently applied. The AI-Driven RRM is always active and any subsequent changes to the RF network, if required, are made at the selected time.
Full Optimization and Partial Optimization with AI-Driven RRM
AI-Driven RRM provides the option to choose between full optimization and partial optimization of network zones depending on your configuration requirements. By default, AI-Driven RRM is enabled with full optimization criteria that consider a combination of channel plan, channel bandwidth, and AP Tx power while generating an RRM recommendation to optimize the zone. However, you can choose partial optimization option in which only the channel plan is optimized. The RRM recommendation with partial optimization will not consider channel bandwidth and AP Tx power. When you enable partial optimization, a new recommendation is generated within the next 24 hours only if a better channel plan for the zone is available. For more information on how to enable full optimization and partial optimization, refer to AI-Driven RRM.
AI-Driven RRM Considerations
- When an AI-Driven RRM recommendation is applied, changes are made to RF parameters that overwrite previous configurations. If a ChannelFly configuration was previously selected for channel selection at the zone, the configuration moves under the control of AI-Driven RRM, and ChannelFly is disabled.
- The background scanning configuration and scanning interval is not changed, but continues to operate collecting data to discover the RF neighborhood that is used for seamless roaming, rogue AP detection, and AI-Driven RRM algorithms.
- AI-Driven RRM will overwrite existing static, manual configurations except ApRadioDeploy and ChannelRange. If a static, manual configuration is detected, RUCKUS AI flags this in the recommendation itself with corresponding warnings. This allows the network administrator to decide whether to override the static, manual configuration by activating AI-Driven RRM.
- AI-Driven RRM does not co-exist with manual override or intervention after the AI-Driven RRM recommendation is applied. The network administrator must disable AI-Driven RRM completely to have manual override for a subset of APs in a zone.
- AI-Driven RRM enables rogue detection at the zone level. This is done to gather a complete RF picture of the environment before optimization decisions are made.
- AI-Driven RRM recommendations are triggered only for zones with 100 percent licensed APs. Any unlicensed APs added to the zone after AI-Driven RRM is applied are not considered, which may result in sub-optimal channel planning in the zone.
- AI-Driven RRM does not operate when zones have mesh configuration enabled.
- AI-Driven RRM requires SmartZone controller release 5.2.1 and later. All access points supported on this release are compatible.
- SmartZone controllers that are already onboarded to RUCKUS AI can immediately begin using AI-Driven RRM. For systems that are newly onboarded, AI-Driven RRM can be used immediately; however, the system improves in performance as historical information accumulates.
- AI-Driven RRM recommends channel and
channel width configuration items at the zone level. The network administrator is
required to pick a date and time to apply the configuration. This is the local time for
the zone for which recommendation is made. It is a best practice to include access
points in the same time zone in a SmartZone-managed zone node because off-peak hours
might differ across time zones.Note: If the RUCKUS AI license is removed or expired, the last run AI-Driven RRM configuration will be used, and the channel plan will be set to static and not ChannelFly or Background Scanning.
AI-Driven RRM Behavior in 2.4 GHz
It is generally accepted and understood that 2.4 GHz is a crowded spectrum with only three non-overlapping channels. However, it is important because several clients still support only the 2.4 GHz band. RF propagation characteristics unique to 2.4 GHz make it a useful choice due to its increased range.
Since the 2.4 GHz band has only three non-overlapping channels (1, 6, and 11) it is likely that APs in this band will hear other APs and the zero interfering links solution does not exist in dense AP deployments in 2.4 GHz. In this scenario, AI-Driven RRM will still guarantee the lowest possible co-channel interference. It does not take action to turn off AP radios in the 2.4 GHz band.
AI-Driven RRM with Dynamic Frequency Selection (DFS) Channels
AI-Driven RRM is aware of the constraints that DFS channels pose in 5 GHz spectrum usage. While the actual decision to operate in a DFS channel is still done at an AP radio level after radar detection measures have been applied, AI-Driven RRM keeps track of radar activity on different DFS channels and intelligently crowdsource this information across multiple APs within the same physical proximity. Based on this crowdsourced information, AI-Driven RRM may restrict the use of some of these DFS channels to avoid disruptions to end users. Of course, optimality in terms of zero interfering links and channel bandwidth selection will still be maintained.
AI-Driven RRM Operation at Zone and AP Level
AI-Driven RRM Interoperability in a Mixed AP Deployment
The AI-Driven RRM algorithm works on the information it receives from RUCKUS access points. Any third-party access point is treated as a rogue AP. These data points are fed into the computation to search for the best option for channel and channel width. These changes are recommended to the network administrator using the AI recommendation mechanism. There is no deauthentication action taken against rogue APs because the algorithms have built-in rogue AP avoidance. Even in the presence of rogue access points or third-party access points, AI-Driven RRM delivers the most optimum solution for interfering AP links and co-channel interference possible.
AI-Driven RRM AP Transmit Power
In high-density deployment scenarios, an ideal channel plan with no interfering links may not be achievable. To address this, AI-Driven RRM automatically detects the residual co-channel interference and provides recommendations for further reducing the interference by lowering the AP transmit power, while at the same time, ensuring that no coverage holes will be created. This minimizes the possibility of neighboring APs interfering with each other's signals, resulting in an enhanced Wi-Fi end-user experience.
The image below displays an example of an AI-Driven RRM recommendation to reduce the AP transmit power in three APs to reduce the interfering links from five to two.
There are two types of recommendations that offer Tx power guidance to the user. The Wi-Fi client experience category of recommendations guides a user to adjust the Tx power setting for 2.4 and 5 GHz radios, mainly to encourage clients to move to a 5 GHz radio instead of a 2.4 GHz radio. This recommendation is also useful when APs have mesh networking enabled. While both recommendations may offer guidance about Tx Power, AI-Driven RRM recommends channel, channel width, and transmit power of radios to holistically reduce co-channel interference and maximize network throughput.
AI-Driven RRM with Automated Frequency Coordination
AI-Driven RRM is enhanced with configuration from Automated Frequency Coordination (AFC) for 6 GHz channel allocation in the US region.
AFC is a system designed to manage spectrum use in the 6 GHz band, maximizing spectrum access and minimizing interference between unlicensed Wi-Fi 6e/7 devices and licensed devices. AFC involves a registered database that contains information about all licensed users currently operating in the 6 GHz band in a specific area. When a new Wi-Fi device, such as an AP, wants to operate in the 6 GHz band, it must register with the AFC system, and thereafter must check in with the system every 24 hours, to obtain a current list of available channels on which to operate. These periodic checks ensure its operation will not interfere with registered devices already using that band. Standard power APs, especially when used outdoors, have a higher potential to interfere with existing 6 GHz users. Therefore, these APs must use the AFC system to protect incumbent operations from RF interference. The AFC system is crucial for maintaining harmony in the spectrum usage, allowing new and existing technologies to coexist without disrupting each other’s services. Without AFC registration, indoor APs operate in low power mode and outdoor APs cannot operate in the 6 Ghz spectrum at all.
AI-Driven RRM solves this by integrating AFC (based on AFC’s channel list) to provide better optimized network configurations. When the channel list is selected for RRM, instead of using only the channels defined in the SmartZone controller, RRM will consider the AFC response data from the APs and compare the AFC channel list with the controller-configured channels. The overlapping or common channels identified through this comparison will be utilized to provide an optimized configuration recommendation.
Prerequisites for AI-Driven RRM with AFC
- AI-Driven RRM recommendation must be in the Applied state.
- AI-Driven RRM based on AFC is applicable only to the US region.
- AFC is applicable only for outdoor APs in 6 GHz band.