Abstract: The present disclosure provides techniques for optimizing roaming by using client location and motion state data. An access point (AP) receives contextual data transmitted from a client device associated with the AP. The AP analyzes the contextual data to determine at least one of a current location of the client device or one or more movement characteristics of the client device. The AP identifies one or more target APs, from a plurality of neighboring APs, for the client device to roam to. The AP sends a roaming recommendation to the client device including the one or more target APs.

Abstract: Devices, systems, methods, and processes for tackling the problem of sticky clients in wireless networks using machine learning processes are described herein. Thresholds related to data rates and/or Received Signal Strength Indicators (RSSI), for Access Points (APs) in a network are identified. These thresholds are configured to dynamically adjust, either based on the time of day or real-time roaming patterns, with the aim of mitigating sticky clients and enhancing overall network performance. Advanced machine learning techniques such as Graph Neural Networks are implemented within the domain of deep learning to significantly alleviate sticky client concerns and promote seamless connectivity for all users.

Abstract: A system and method for using 802.11v action report messages to enhance roaming. In one embodiment, a method includes transmitting, to a station, a transition management message that identifies a recommended access point for roaming for the station, receiving, from the station, an action report in response to the transition management message, where the action report indicates feedback to the recommended access point identified in the transition management message, and determining a roaming recommendation based on the action report.

Abstract: Described herein are devices, systems, methods, and processes for managing roaming actions in a wireless network. The embodiments utilize a machine learning model to generate roaming recommendations based on a plurality of roaming-related metrics. The metrics include data about the current network conditions, the station's previous roaming experiences, and the capabilities of potential roaming target candidates. The roaming recommendations can be provided to a station by an access point (AP). The station can then attempt to perform a roaming action based on the recommendations. After the attempt, the station transmits a roaming feedback to the AP, which includes data about the success or failure of the roaming action and any additional relevant data. In case the station rejects the roaming recommendations, the station may also provide a feedback indicating the rejection. The feedback is utilized to update the machine learning model, thereby improving the accuracy of future roaming recommendations.

Abstract: A system and method are provided for tracking the quality of telemetry data in a wireless network, and to provide profiling of the telemetry to prevent erroneous radio resource management (RRM) computations. Statistical profiles are generated from the telemetry data that includes both computation data, which is used in RRM computations, and other network data. A data-quality score is generated based on the other data and statistical profiles of the computation data. The data-quality score represents whether the telemetry data is of sufficient quality to be used in RRM computations. The data-quality score can be based, at least in part, on detecting changes in the statistical profiles relative to a baseline statistical profile of the telemetry data and using the second network data to assess a likelihood that the detected changes arise from a degradation in a quality of the first network data.

Abstract: Cell edge prediction for optimized roaming may be provided. Cell edge prediction can include predicting cell edges for a plurality of APs including a connected AP and one or more additional APs. A cell edge prediction can be for a client connected to the connected AP. The cell edge prediction may comprise an indication of one or more candidate APs for the client to roam to of the one or more additional APs and an estimated time the client will reach the cell edge of the connected AP. After generating the cell edge prediction, the cell edge prediction can be transmitted to the client.

Abstract: Roaming optimization, and particularly transmitting directional neighbor reports to optimize roaming may be provided. For roaming optimization, it is first determined that a client is going to roam to a new Access Point (AP). In response to determining the client is going to roam to a new AP, an estimated path of the client is determined. One or more candidate APs are determined based at least in part on the estimated path, and probabilities of roaming to the one or more candidate APs is determined based at least in part on the estimated path. A directional neighbor report including a list of the one or more candidate APs and the probabilities is generated and transmitted to the client.

Abstract: Techniques for classification of wireless signals are disclosed. These techniques include receiving data describing radio frequency characteristics of a wireless network transmission environment and presenting the data for display on a graphical user interface (GUI) associated with a network device in the wireless network. The techniques further include receiving input from the GUI identifying one or more portions of the data, and training a machine learning (ML) model to classify an interferer in the wireless network transmission environment based on the identified one or more portions of the data.

Abstract: In one embodiment, a network assurance service that monitors a wireless network receives data regarding connection failures of a wireless client of the wireless network. The network assurance service forms a behavioral profile for the wireless client based on the received data regarding the connection failures of the wireless client. The network assurance service uses machine learning to determine whether the behavioral profile of the wireless client is an outlier in relation to behavioral profiles of other wireless clients of the wireless network. The network assurance service causes performance of a mitigation action with respect to the wireless client, when the wireless client is determined to be an outlier.

Abstract: In one embodiment, a network assurance service that monitors a wireless network receives data regarding connection failures of a wireless client of the wireless network. The network assurance service forms a behavioral profile for the wireless client based on the received data regarding the connection failures of the wireless client. The network assurance service uses machine learning to determine whether the behavioral profile of the wireless client is an outlier in relation to behavioral profiles of other wireless clients of the wireless network. The network assurance service causes performance of a mitigation action with respect to the wireless client, when the wireless client is determined to be an outlier.