Cloud-based control of a Wi-Fi network
Systems and methods include obtaining measurements from a Wi-Fi network that includes i) a plurality of access points interconnected to one another and at least one connected to a gateway, and ii) a plurality of Wi-Fi client devices that each connect to an access point of the plurality of access points; learning usage patterns in the Wi-Fi network through analysis with one or more machine learning algorithms; and providing a configuration to the Wi-Fi network based on the learned usage pattern where the configuration includes a topology of the plurality of access points.
1 . A non-transitory computer-readable storage medium having computer readable code stored thereon for programming a device to perform steps of:
obtaining measurements from a Wi-Fi network, the measurements corresponding to one or more access points, a gateway, and a plurality of Wi-Fi client devices that each connect to an access point of the one or more access points;
analyzing, via execution of at least one machine learning algorithm, the measurements for the one or more access points, gateway, and plurality of Wi-Fi client devices;
learning usage patterns in the Wi-Fi network based on the analysis of the measurements for the one or more access points, gateway, and plurality of Wi-Fi client devices, the usage patterns comprising information related to usage of steering commands for the plurality of Wi-Fi client devices, the steering commands corresponding to disassociation of at least one of the plurality of Wi-Fi client devices to the Wi-Fi network, such that a respective disassociation of a Wi-Fi client device is based on a response to a probe request associated with a respective disassociation request, the probe request corresponding to presence of the Wi-Fi client device within an associated client list (ACL list) hosted by a cloud controller, the steering commands further comprising a communicated authentication failure to the Wi-Fi client device; and
providing a configuration to the Wi-Fi network based on the learned usage pattern where the configuration includes at least a frequency selection and a dissociation of a Wi-Fi client device among the plurality of Wi-Fi client devices in accordance with the steering commands, the providing of the configuration causing a modification to a topology of the Wi-Fi network that corresponds to which nodes and channels within the Wi-Fi network are available, which impacts how the nodes within the Wi-Fi network interact via routes among such nodes as based on the information related to the steering commands.
2 . The non-transitory computer-readable storage medium of claim 1 , wherein the routes correspond to at least one of a client link and backhaul link.
3 . The non-transitory computer-readable storage medium of claim 1 , wherein the steps further comprise:
obtaining measurements from a plurality of networks with statistics regarding behavior of a number of Wi-Fi client devices; and
training the one or more machine learning algorithms with the statistics regarding the behavior of the number of Wi-Fi client devices.
4 . The non-transitory computer-readable storage medium of claim 1 , wherein the steps further comprise:
classifying the plurality of Wi-Fi client devices based on their mobility.
5 . The non-transitory computer-readable storage medium of claim 4 , wherein the steps further comprise:
predicting associations of mobile Wi-Fi client devices with the plurality of access points such that the configuration is based in part thereon.
6 . The non-transitory computer-readable storage medium of claim 4 , wherein the classifying is utilized for a choice of frequency band.
7 . The non-transitory computer-readable storage medium of claim 1 , wherein the measurements include time-series data collected both on transmit and receive.
8 . The non-transitory computer-readable storage medium of claim 1 , wherein the steps further comprise:
learning what types of Wi-Fi client devices connect to the Wi-Fi network and what types of steering approaches the types of Wi-Fi client devices utilize.
9 . The non-transitory computer-readable storage medium of claim 1 , wherein the configuration includes a plurality of channel and bandwidth (BW) selection, routes, Request to Send/Clear to Send (RTS/CTS) settings, Transmitter (TX) power, clear channel assessment thresholds, client association steering, and band steering.
10 . The non-transitory computer-readable storage medium of claim 1 , wherein the steps further comprise:
periodically repeating the obtaining, learning, and providing.
11 . A method comprising steps of:
obtaining measurements from a Wi-Fi network, the measurements corresponding to one or more access points, a gateway, and a plurality of Wi-Fi client devices that each connect to an access point of the one or more access points;
analyzing, via execution of at least one machine learning algorithm, the measurements for the one or more access points, gateway, and plurality of Wi-Fi client devices;
learning usage patterns in the Wi-Fi network based on the analysis of the measurements for the one or more access points, gateway, and plurality of Wi-Fi client devices, the usage patterns comprising information related to usage of steering commands for the plurality of Wi-Fi client devices, the steering commands corresponding to disassociation of at least one of the plurality of Wi-Fi client devices to the Wi-Fi network, such that a respective disassociation of a Wi-Fi client device is based on a response to a probe request associated with a respective disassociation request, the probe request corresponding to presence of the Wi-Fi client device within an associated client list (ACL list) hosted by a cloud controller, the steering commands further comprising a communicated authentication failure to the Wi-Fi client device; and
providing a configuration to the Wi-Fi network based on the learned usage pattern where the configuration includes at least a frequency selection and a dissociation of a Wi-Fi client device among the plurality of Wi-Fi client devices in accordance with the steering commands, the providing of the configuration causing a modification to a topology of the Wi-Fi network that corresponds to which nodes and channels within the Wi-Fi network are available, which impacts how the nodes within the Wi-Fi network interact via routes among such nodes as based on the information related to the steering commands.
12 . The method of claim 11 , wherein the routes correspond to at least one of a client link and backhaul link.
13 . The method of claim 11 , wherein the steps further comprise:
obtaining measurements from a plurality of networks with statistics regarding behavior of a number of Wi-Fi client devices; and
training the one or more machine learning algorithms with the statistics regarding the behavior of the number of Wi-Fi client devices.
14 . The method of claim 11 , wherein the steps further include classifying the plurality of Wi-Fi client devices based on their mobility.
15 . The method of claim 14 , wherein the steps further comprise:
predicting associations of mobile Wi-Fi client devices with the plurality of access points such that the configuration is based in part thereon.
16 . The method of claim 14 , wherein the classifying is utilized for a choice of frequency band.
17 . The method of claim 11 , wherein the measurements include time-series data collected both on transmit and receive.
18 . The method of claim 11 , wherein the steps further comprise:
learning what types of Wi-Fi client devices connect to the Wi-Fi network and what types of steering approaches the types of Wi-Fi client devices utilize.
19 . The method of claim 11 , wherein the configuration includes a plurality of channel and bandwidth (BW) selection, routes, Request to Send/Clear to Send (RTS/CTS) settings, Transmitter (TX) power, clear channel assessment thresholds, client association steering, and band steering.
20 . The method of claim 11 , wherein the steps further comprise:
periodically repeating the obtaining, learning, and providing.