Optimizing device-to-device communication protocol selection in an edge computing environment

1. A method, comprising:

receiving, by a computing system, a request for a service from a user device, wherein the computing system is one of plural edge computing devices in an edge computing environment;

determining, by the computing system, a computational task performed in providing the service;

selecting, by the computing system and using a machine learning model, a set of the edge computing devices to perform the computational task and communication protocols to be used between selected ones of the set of the edge computing devices while performing the computational task, wherein the machine learning model is configured to select the set of the edge computing devices and the communication protocols based on minimizing a time to perform the computational task; and

sending, by the computing system and to the set of the edge computing devices, instructions to perform the computational task, thereby causing the set of the edge computing devices to perform the service in response to the request from the user device.

2. The method of claim 1 , further comprising training the machine learning model using historic performance data of the plural edge computing devices in the edge computing environment.

3. The method of claim 2 , wherein:

the historic performance data includes actual bandwidth usage and actual power consumption by respective ones of the edge computing devices; and

the machine learning model is configured to select the set of the edge computing devices and the communication protocols based on minimizing bandwidth usage and power consumption in performing the computational task.

4. The method of claim 2 , wherein the training the machine learning model comprises training the machine learning model using a Markov's decision process with reinforcement learning.

5. The method of claim 1 , wherein:

one or more of the edge computing devices stores a replicated copy of data stored at a cloud server; and

the sending instructions comprises sending data obtained from the replicated copy of data.

6. The method of claim 5 , wherein a distance between the edge computing devices and the user device is shorter than a distance between the cloud server and the user device.

7. The method of claim 5 , wherein the computational task comprises plural respective computational tasks, and further comprising:

determining one of the plural computational tasks cannot be performed by the set of edge computing devices within a time required by an SLA; and

in response to the determining, offloading at least one lower priority process from one or more of the set of edge computing devices to the cloud server.

8. A computer program product comprising one or more computer readable storage media having program instructions collectively stored on the one or more computer readable storage media, the program instructions executable to:

receive a request for a service from a user device, wherein the request is received by a computing device that is one of plural edge computing devices in an edge computing environment;

determine a computational task performed in providing the service;

select, using a machine learning model, a set of the edge computing devices to perform the computational task and communication protocols to be used between selected ones of the set of the edge computing devices while performing the computational task, wherein the machine learning model is configured to select the set of the edge computing devices and the communication protocols based on minimizing a time to perform the computational task; and

send, to the set of the edge computing devices, instructions to perform the computational task, thereby causing the set of the edge computing devices to perform the service in response to the request from the user device.

9. The computer program product of claim 8 , wherein the program instructions executable to train the machine learning model using historic performance data of the plural edge computing devices in the edge computing environment.

10. The computer program product of claim 9 , wherein:

the historic performance data includes actual bandwidth usage and actual power consumption by respective ones of the edge computing devices; and

the machine learning model is configured to select the set of the edge computing devices and the communication protocols based on minimizing bandwidth usage and power consumption in performing the computational task.

11. The computer program product of claim 9 , wherein the training the machine learning model comprises training the machine learning model using a Markov's decision process with reinforcement learning.

12. The computer program product of claim 8 , wherein:

one or more of the edge computing devices stores a replicated copy of data stored at a cloud server; and

the sending instructions comprises sending data obtained from the replicated copy of data.

13. The computer program product of claim 12 , wherein a distance between the edge computing devices and the user device is shorter than a distance between the cloud server and the user device.

14. The computer program product of claim 12 , wherein the computational task comprises plural computational tasks, and the program instructions executable to:

determine one of the plural computational tasks cannot be performed by the set of edge computing devices within a time required by an SLA; and

in response to the determining, offload at least one lower priority process from one or more of the set of edge computing devices to the cloud server.

15. A system comprising:

a processor, a computer readable memory, one or more computer readable storage media, and program instructions collectively stored on the one or more computer readable storage media, the program instructions executable to:

receive a request for a service from a user device, wherein the request is received by a computing device that is one of plural edge computing devices in an edge computing environment;

determine a computational task performed in providing the service;

select, using a machine learning model, a set of the edge computing devices to perform the computational task and communication protocols to be used between selected ones of the set of the edge computing devices while performing the computational task,

wherein the machine learning model is configured to select the set of the edge computing devices and the communication protocols based on minimizing a time to perform the computational task; and

send, to the set of the edge computing devices, instructions to perform the computational task, thereby causing the set of the edge computing devices to perform the service in response to the request from the user device.

16. The system of claim 15 , wherein the program instructions executable to train the machine learning model using historic performance data of the plural edge computing devices in the edge computing environment.

17. The system of claim 16 , wherein:

the historic performance data includes actual bandwidth usage and actual power consumption by respective ones of the edge computing devices; and

the machine learning model is configured to select the set of the edge computing devices and the communication protocols based on minimizing bandwidth usage and power consumption in performing the computational task.

18. The system of claim 15 , wherein:

one or more of the edge computing devices stores a replicated copy of data stored at a cloud server; and

the sending instructions comprises sending data obtained from the replicated copy of data.

19. The system of claim 18 , wherein a distance between the edge computing devices and the user device is shorter than a distance between the cloud server and the user device.

20. The system of claim 18 , wherein the computational task comprises plural computational tasks, and the program instructions executable to:

determine one of the plural computational tasks cannot be performed by the set of edge computing devices within a time required by an SLA; and

in response to the determining, offload at least one lower priority process from one or more of the set of edge computing devices to the cloud server.