AI Roaming

What Is AI Roaming?

AI roaming is an enhanced roaming feature that builds upon smart roaming to enhance the quality of connections during Station (STA) roaming. It achieves this by enhancing roaming sensitivity, ensuring that STA signals remain strong during the roaming process. Moreover, AI roaming gathers pertinent information like STA roaming thresholds and roaming steering policies to create specific roaming profiles for different types of STAs. This approach aims to improve the success rate of roaming and provide an enhanced roaming experience for STAs, ultimately leading to improved overall performance.

Traditional WLAN and Smart Roaming Issues

In a wireless local area network (WLAN), users need to maintain connectivity while being mobile. However, the signal range of a single access point (AP) is limited. Consequently, users frequently transition from one AP's coverage area to another. To ensure uninterrupted network connectivity during these movements, the concept of WLAN roaming is implemented.

Traditional WLAN Roaming

WLAN roaming is a procedure where a station (STA) seamlessly switches its association from one access point (AP) to another as it moves towards the coverage boundary between the two APs. This transition ensures uninterrupted connectivity without any disruptions. In essence, WLAN roaming can be compared to the handover process in cellular networks. Just like a mobile phone smoothly transitions from one base station's coverage area to another, maintaining uninterrupted and seamless call experiences, WLAN roaming enables STAs to maintain continuous connectivity while moving between APs.

Smart Roaming

Traditional WLAN roaming, also known as proactive STA roaming, is initiated by the station (STA) itself. However, some STAs exhibit low roaming aggressiveness, leading them to remain connected to their initially associated access points (APs) despite weak signals, long distances, or low data rates. These STAs, referred to as sticky STAs, do not roam to neighboring APs with stronger signals.

Proactive STA roaming alone cannot effectively address the issue of sticky STAs. This is where smart roaming comes into play. Smart roaming introduces intelligence to guide STAs in making roaming decisions. It utilizes advanced algorithms and techniques to steer STAs towards a more optimal roaming behavior. Compared to proactive STA roaming, smart roaming demonstrates increased intelligence and effectiveness in managing STAs' roaming behavior.

STA Roaming Process

1. Triggering: When the received signal strength indicator (RSSI) of the downlink falls below a certain threshold, a station (STA) initiates the channel scanning process.

2. Scanning: The STA performs active or passive scanning to identify available access points (APs) in its vicinity and collects network information for network selection.

• Active scanning: The STA periodically sends Probe Request frames, and APs respond with Probe Response frames, allowing the STA to detect AP presence.

• Passive scanning: The STA listens for Beacon frames periodically broadcasted by APs to detect their presence.

The choice between active and passive scanning modes is determined by the STA, and wireless network adapters in devices like mobile phones or computers generally support both modes. Passive scanning is commonly used by Voice over Internet Protocol (VoIP) terminals.

3. Network selection: Based on the information gathered during scanning, the STA selects an AP as the target for roaming.

4. Handover: The STA selects an appropriate roaming mode based on its own capabilities and the capabilities of the network.

Problems with smart roaming

• Triggering phase: The roaming initiation in a station (STA) is determined by a fixed received signal strength indicator (RSSI) threshold set on the network side, without considering variations in RSSI thresholds between different STAs. This can result in severe degradation of service quality during roaming when the RSSI threshold for some STAs falls below the fixed threshold. Furthermore, when a STA passes through areas with poor signal coverage, such as a corridor corner, the downlink RSSI from the associated access point (AP) sharply decreases. In such cases, the STA triggers roaming based on an RSSI significantly lower than its expected roaming threshold. This leads to a prolonged period of poor user experience as the STA goes through the scanning, network selection, and handover processes.

• Scanning phase: STAs with low RSSI values scan all channels, causing significant disruptions to normal services. During channel switching, the STA briefly remains on each channel (duration varies based on STA type) to actively or passively scan for APs operating on the same channel. In passive scanning mode, the STA may fail to detect certain APs.

• Network selection phase: The network side selects an optimal roaming target based on the uplink RSSI, which can differ from the downlink RSSI. As a result, the AP chosen by the network may not be the most suitable one for the STA. Additionally, the network side does not consider variations in handover conditions between different STAs, leading to a decrease in the success rate of roaming steering. The uplink RSSI refers to the signal strength received by a STA from an AP, while the downlink RSSI is the signal strength received by an AP from a STA.

• Handover phase:The handover process takes a considerable amount of time, resulting in degraded user experience.

The Significance of AI Roaming

AI roaming offers several advantages over smart roaming:

• Proactive roaming triggering: AI roaming triggers roaming in advance, preventing prolonged poor user experience caused by sudden drops in received signal strength indicator (RSSI). This enhances STA roaming sensitivity and ensures timely handover.

• Coordinated scanning of roaming neighbors: AI roaming utilizes the downlink RSSI measurement capability of STAs and coordinated scanning using a separate scanning radio in APs. This enables the prediction of STA motion patterns and potential roaming paths in advance, reducing unnecessary roaming events.

• Differentiated roaming behavior identification: AI roaming creates roaming profiles for different types of STAs. It learns the conditions that lead to successful roaming steering for a particular STA type, improving the overall success rate of STA roaming steering in subsequent instances.

• Advanced identification of optimal roaming AP: AI roaming leverages STA location and roaming profile information to identify the optimal AP for successful roaming in advance. This reduces the time required for AP handover, enhancing the efficiency of the roaming process.

Key Technologies of AI Roaming

STA Identification and Roaming Profile

STAs of various models utilize different Wi-Fi chips and WLAN module drivers, resulting in support for different roaming steering modes. To address this, a STA identification function is employed. When a STA connects to an access point (AP), the AP identifies the STA type and reports this information, along with the STA's capability details, to the Access Controller (AC) or iMaster NCE-CampusInsight analyzer (CI analyzer). Using the STA identification result, the AC or CI analyzer analyzes the roaming behaviors of STAs. This includes determining the RSSI threshold for STA roaming, the types of measurement frames (LM/TPC) that the STAs respond to, and the 802.11v-based steering conditions supported by the STAs. STA profiles are created based on the STA models.

When a new STA comes online, if a matching STA profile is found, the AC or CI analyzer sends the corresponding STA profile information to the AP. The AP then steers the STA to roam according to the specific profile. If no matching STA profile is found, no STA profile information is delivered. In this case, the AP relies on the STA's default roaming steering policy and reports the roaming process as a sample to the AC or CI analyzer. This allows the generation of a new STA profile through online automatic learning. This approach ensures that STAs with different capabilities and roaming modes are accurately identified and steered accordingly, enhancing the efficiency and effectiveness of WLAN roaming.

Optimal AP Identification and Roaming Steering

Once a STA is connected to an access point (AP), continuous link measurements, primarily focusing on RSSI information, are performed between the STA and its associated AP. Based on the changes in these measurement results, the AP determines whether the STA is moving away from it.

If the STA is observed to be moving significantly away from its associated AP, the potential target AP, to which the STA may roam, initiates a coordinated scanning process. This involves simulating the behavior of the STA's associated AP and performing link measurements with the STA. The measurement results are then forwarded by the STA to its currently associated AP.

Using the received measurement results, the associated AP generates an AP coverage snapshot of the STA's current location. By leveraging this snapshot along with the STA's roaming profile information, the AP identifies the optimal target AP for the STA to roam to. Finally, the AP steers the STA to perform the necessary handover and switch its association to the selected target AP.

Application scenarios of AI Roaming

Office Scenario with Obstacles

When a station (STA) moves along a path with obstacles or corners in an office setting, the received signal strength indicator (RSSI) of the STA experiences a sharp decline until the STA roams to a new access point (AP). This period is characterized by a poor quality link between the STA and AP, leading to limited capability to counteract interference, reduced STA speed, and prolonged transmission times. In such situations, AI roaming proves beneficial by accurately predicting the STA's roaming trajectory and initiating the roaming process proactively. This proactive approach enhances the sensitivity of roaming, preventing service disruptions caused by delayed roaming.

Office Scenario Without Obstacles

In office environments where there are no obstacles, access points (APs) are commonly deployed in either straight-line or W-shaped configurations. These deployment modes guarantee uninterrupted AP signal coverage and consistent signal transitions as stations (STAs) roam between APs. Consequently, AI roaming can be utilized to actively guide STA roaming from the network side, ensuring a proactive and timely roaming process. Additionally, the received signal strength indicator (RSSI) of the STAs remains consistently within an optimal range throughout the roaming process.