Adaptive wireless client steering

1. A network information aggregator of a first network comprising a network interface, a memory, and processing circuitry to:

receive, from a model creation device, a machine learning model trained using initialization data;

generate, using the machine learning model, a device steering rule to steer a client device from a first radio to a second radio;

send a steer command to the first radio, wherein the steer command is sent in response to, in part, radio environment characteristics of the first radio matching radio environment criteria of the device steering rule;

receive a response to the steer command, including a result of applying the steer command;

generate a roaming record including the result of applying the steer command;

send the roaming record to the model creation device; and

receive, from the model creation device, an updated machine learning model trained using data including the roaming record, wherein the data including the roaming record also includes a roaming record generated by another network information aggregator of a second network similar to the first network.

2. The network information aggregator of claim 1 , wherein the roaming record includes radio environment characteristics of the first radio.

3. The network information aggregator of claim 1 , wherein the roaming record includes client state information of the client device.

4. The network information aggregator of claim 1 , wherein the network information aggregator is a component of an access point including the first radio.

5. A method, comprising:

receiving information from a first radio of a first network;

generating a roaming record based on the information from the first radio and radio environment characteristics of the first radio;

updating a machine learning model by training the machine learning model using data including the roaming record, wherein the data including the roaming record also includes a roaming record generated by a network information aggregator of a second network similar to the first network;

generating, using the updated machine learning model, a device steering rule to steer a client device from a second radio to a third radio; and

sending a steer command to the second radio in response to, in part, radio environment characteristics of the second radio matching radio environment criteria of the device steering rule.

6. The method of claim 5 , comprising generating the machine learning model by training the machine learning model using an initial set of data.

7. The method of claim 5 , wherein the first radio and the second radio are the same radio.

8. The method of claim 5 , comprising deleting, using the updated machine learning model, a second device steering rule.

9. The method of claim 5 , wherein the steer command is sent in response to, in part, client state information of the client device matching client state criteria of the device steering rule.

10. The method of claim 9 , wherein the client state information includes at least one of: a status of an application executing on the client device, a type of application executing on the client device, a pattern of data use by the client device, types of data being transceived by the client device, an activity pattern of the client device, and a power state of a client device.

11. The method of claim 5 , comprising generating a second roaming record based on a response to the steer command that includes a success determination.

12. The method of claim 11 , comprising transmitting the second roaming record to a training data repository.

13. A non-transitory, computer-readable medium comprising instructions that, when executed by processing circuitry, cause the processing circuitry to:

train a machine learning model using initialization data stored in a training data repository, the data including roaming records;

receive a roaming record from a network device of a first network;

update the machine learning model using data including the roaming record from the network device, wherein the data including the roaming record also includes a roaming record generated by a network information aggregator of a second network similar to the first network;

generate, using the machine learning model, a device steering rule including radio environment criteria; and

transmit the device steering rule to a network device.

14. The computer-readable medium of claim 13 , wherein the device steering rule is stored in a rule repository and wherein the device steering rule is transmitted in response to a request from the network device.

15. The computer-readable medium of claim 13 , wherein the device steering rule includes client state criteria.