Hewlett Packard Enterprise Development LP

Aruba Networks, Inc.

Bernd Bandemer

Jose Tellado

Sachin Ganu

Arogyaswami J Paulraj

Charlie Hogg

Alexander Mendez

Hilmi Gunes Kayacik

Sai Pradeep Venkatraman

Gaurav Patwardhan

Isaac Theogaraj

Gopal Gupta

Abhinesh Mishra

Yusuke Sakamoto

Abtin Ansari

Sargam Jain

Charles Hogg

David Fehling, Jr

Susmita Ghose

Gunes Kayacik

Pradeep Nemavat

Systems and methods are provided for optimizing resource consumption by bringing intelligence to the key allocation process for fast roaming. Specifically, embodiments of the disclosed technology use machine learning to predict which AP a wireless client device will migrate to next. In some embodiments, machine learning may also be used to select a subset of top neighbors from a neighborhood list. Thus, instead of allocating keys for each of the APs on the neighborhood list, key allocation may be limited to the predicted next AP, and the subset of top neighbors. In some embodiments, a reinforcement learning model may be used to dynamically adjust the size of the subset in order to optimize resources while satisfying variable client demand.

Optimizing key allocation during roaming using machine learning

Examples of the presently disclosed technology provide automated firmware recommendation systems that inject the intelligence of machine learning into the firmware recommendation process. To accomplish this, examples train a machine learning model on troves of historical customer firmware update data on a dynamic basis (e.g., examples may train the machine learning model on weekly basis to predict accepted firmware updates made by a vendor's customers across the most recent 6 months). From this dynamic training, the machine learning model can learn to predict/recommend an optimal firmware version for a customer/network device cluster based on firmware-related features, recent customer preferences, and other customer-specific factors. Once trained, examples can deploy the machine learning model to make highly tailored firmware recommendations for individual network device clusters of individual customers taking the above described factors into account.

Machine learning based firmware version recommender

A method for identifying a client device in a network, and a first radio in the network that is coupled with the client device is provided. The method includes determining one or more sequences of roaming events for multiple client devices in the network, evaluating a performance metric for a roaming event and evaluating an interaction between the client device and one or more radios involved in the roaming events for the plurality of client devices. The method also includes selecting a second radio in the network based at least in part on (1) the one or more sequences of roaming events, (2) the performance metric, and (3) the interaction between the client device and the one or more radios, and recommending switching the client device from the first radio to the second radio. A system and a predictive tool to perform the above method are also provided.

Assisted network roaming with predictive network tool

Systems and methods are provided for receiving a set of feature vectors. Each feature vector in the set may comprise feature values for a plurality of features associated with network communications. A first score for a first subset of the feature vectors that have at least one common feature value for a first feature of the plurality of features may be determined. A second score for a second subset of the feature vectors may be determined. The second subset may comprise the first subset and other feature vectors that have a different feature value for the first feature. Based on a change between the first score and the second score, whether to group the common feature value and the different feature value together may be determined.

Systems and methods for identifying correlations of certain scenarios to performance of network communications

An example network information aggregator is disclosed. The network information aggregator includes a network interface, a memory, and processing circuitry. The processing circuitry is to receive a machine learning model trained using initialization data from a model creation device. The processing circuitry is also to generate a device steering rule to steer a client device from a first radio to a second radio using the machine learning model. The processing circuitry is also to send a steer command to the first radio.

Adaptive wireless client steering

Example implementations relate to determining when to adjust a power state of access points. A non-transitory computer readable medium may store instructions executable by a processing resource to: determine a subset of a group of access points (APs) that is to provide, to a client device, at least a performance threshold during a particular time interval, wherein an AP of the subset of the group of APs serves the client device; and determine when to adjust a power state of a remaining AP of the group of APs based on: a first degree of performance being provided by a first set of radios of the subset of the group of APs; and a second degree of performance to be provided by the remaining AP of the group of APs if the group of client devices is provided a network connectivity via a second radio of the remaining AP of the group of APs.

Determining when to adjust a power state of access points

Some of examples disclosed herein relate to classifying access points (APs). Some of the example may enable automatically classifying a set of APs based on at least one AP attribute of each AP in the set of APs, determining, based on the automatic classification, that a subset of APs in the set of APs are classified into a same class; and automatically determining a recommended configuration setting for the subset of APs.

Access point configuration parameter recommendation based on access point classification

Example implementations relate to adjusting power states of access points based on a power model. A non-transitory computer readable medium may store instructions executable by a processing resource to: in response to a client device being associated with an access point (AP) of a group of APs, determine: a first degree of performance being provided to the client device via a first radio of the AP of a group of APs; and a second degree of performance to be provided to the client device via a second radio of the group of APs, if the client device is provided a network connectivity via the second radio; determine, based on the first degree of performance and the second degree of performance, a subset of the group of APs whose power state is adjustable to a different power state; and adjust a power state of the subset of the group of APs.

Adjusting power states of access points

Example implementations relate to predicting power states of access points (APs). A non-transitory computer readable medium may store instructions executable by a processing resource to: in response to a first client device being associated with a first access point (AP) of a group of APs, determine a first degree of performance a second AP of the group is to provide if the first client device is associated with the second AP; and predict, based on the determined degree of performance, a second degree of performance the group of AP is to provide to a second client device if the second client device is associated with the group of APs during a first time interval at a location.

Growing community of inventors

Santa Clara, CA, United States of America

Bernd Bandemer