Firmware/software over the air (FOTA) scheduling for connected vehicles

1. A method comprising:

obtaining, by a service, wireless network connectivity data for a plurality of vehicles;

clustering, by the service, the plurality of vehicles into clusters using a machine learning-based model, wherein the machine learning-based model is trained using a first subset of the network connectivity data;

validating, by the service, the clustering of the plurality of vehicles based on an assessment of the first subset of the network connectivity data for a particular vehicle of the plurality of vehicles;

determining that a result of the validation of the clustering fails to satisfy a threshold;

re-training, by the service, the machine learning-based model using a second subset of the network connectivity data different from the first subset of the network connectivity data;

clustering, by the service, the plurality of vehicles into clusters using the re-trained machine learning-based model;

determining, by the service, an update schedule for the plurality of vehicles by assigning the clusters to timeslots of the schedule; and

pushing, by the service, a firmware or software update to the vehicles according to the update schedule.

2. The method as in claim 1 , wherein the machine learning-based model uses a record of previously accepted or rejected firmware or software updates as input.

3. The method as in claim 1 , wherein the network connectivity data comprises charging data records (CDRs) indicative of activity of the vehicles relative to a cellular network.

4. The method as in claim 1 , further comprising:

dividing the network connectivity data into a training dataset and a validation dataset;

using the training dataset to train the machine learning-based model that models network connectivity of the vehicles; and

using the validation dataset to validate the trained machine learning-based model.

5. The method as in claim 4 , further comprising:

updating, by the service, the machine learning-based model using new network connectivity data.

6. The method as in claim 1 , wherein determining the update schedule for the plurality of vehicles comprises:

assigning the vehicles to timeslots of the schedule, to maximize an estimated number of vehicles that will receive the update within a predefined period of time.

7. An apparatus, comprising:

one or more network interfaces to communicate with a network;

a processor coupled to the network interfaces and configured to execute one or more processes; and

a memory storing a process executable by the processor, the process when executed configured to:

obtain wireless network connectivity data for a plurality of vehicles;

cluster the plurality of vehicles into clusters using a machine learning-based model, wherein the machine learning-based model is trained using a first subset of the network connectivity data;

validate the clustering of the plurality of vehicles based on an assessment of the first subset of the network connectivity data for a particular vehicle of the plurality of vehicles;

determine that a result of the validation of the clustering fails to satisfy a threshold;

re-train the machine learning-based model using a second subset of the network connectivity data different from the first subset of the network connectivity data;

cluster the plurality of vehicles into clusters using the re-trained machine learning-based model;

determine an update schedule for the plurality of vehicles by assigning the clusters to timeslots of the schedule; and

push a firmware or software update to the vehicles according to the update schedule.

8. The apparatus as in claim 7 , wherein the network connectivity data comprises charging data records (CDRs) indicative of activity of the vehicles relative to a cellular network.

9. The apparatus as in claim 7 , wherein the apparatus determines the update schedule for the plurality of vehicles by:

assigning the vehicles to timeslots of the schedule, to maximize an estimated number of vehicles that will receive the update within a predefined period of time.

10. The apparatus as in claim 7 , wherein the process when executed is further configured to:

divide the network connectivity data into a training dataset and a validation dataset;

use the training dataset to train the machine learning-based model that models network connectivity of the vehicles; and

use the validation dataset to validate the trained machine learning-based model.

11. The apparatus as in claim 7 , wherein the process when executed is further configured to:

update the machine learning-based model using new network connectivity data.

12. The apparatus as in claim 7 , wherein the vehicles are autonomous or semi-autonomous vehicles.

13. A tangible, non-transitory, computer-readable medium storing program instructions that cause a device to execute a process comprising:

obtaining, by the device, wireless network connectivity data for a plurality of vehicles;

clustering, by the device, the plurality of vehicles into clusters using a machine learning-based model, wherein the machine learning-based model is trained using a first subset of the network connectivity data;

validating, by the service, the clustering of the plurality of vehicles based on an assessment of the first subset of the network connectivity data for a particular vehicle of the plurality of vehicles;

determining that a result of the validation of the clustering fails to satisfy a threshold;

re-training, by the service, the machine learning-based model using a second subset of the network connectivity data different from the first subset of the network connectivity data;

clustering, by the service, the plurality of vehicles into clusters using the re-trained machine learning-based model;

determining, by the device, an update schedule for the plurality of vehicles by assigning the clusters to timeslots of the schedule; and

pushing, by the device, a firmware or software update to the vehicles according to the update schedule.

14. The computer-readable medium as in claim 13 , wherein the network connectivity data comprises charging data records (CDRs) indicative of connectivity of the vehicles to a cellular network.

15. The computer-readable medium as in claim 13 , wherein determining the update schedule for the plurality of vehicles comprises:

assigning the vehicles to timeslots of the schedule, to maximize an estimated number of vehicles that will receive the update within a predefined period of time.

16. The computer-readable medium as in claim 13 , wherein the process further comprises:

dividing the network connectivity data into a training dataset and a validation dataset;

using the training dataset to train the machine learning-based model that models network connectivity of the vehicles; and

using the validation dataset to validate the trained machine learning-based model.

17. The computer-readable medium as in claim 16 , wherein the process further comprises:

updating, by the device, the machine learning-based model using new network connectivity data.