Managing task offloading to edge servers in a multi-access edge computing, MEC, system

1. A method of managing task offloading to edge servers in a multi-access edge computing (MEC) system, the method comprising:

receiving, by a task broker implemented at a host management level of the MEC system, a task offloading request from a UE application;

analyzing, by the task broker, resource requirements of a task of the task offloading request and running a forecasting mechanism that determines whether the task offloading request will eventually be dropped or not by a system management level of the MEC system; and

by the task broker, based on the determination of the forecasting mechanism, either rejecting or provisionally admitting the task offloading request.

2. The method according to claim 1 , further comprising:

relaying, by the task broker, the task offloading request to the system management level of the MEC system,

wherein an MEC orchestrator, in case the task offloading request had been provisionally admitted by the task broker, decides, based on the overall requirements of a plurality of task offloading requests received within a system-level time slot (T sl ), and the available resources of the respective edge server, either to drop or to finally accept the provisionally admitted request.

3. The method according to claim 2 , further comprising:

rejecting, by the task broker, all task offloading requests having a resource requirement higher than a resource requirement of the last provisionally accepted request that was dropped by the MEC orchestrator within the current system-level time slot (T sl ), and

applying, by the task broker, a learning algorithm to forecast based on a task offloading request having a first resource requirement lower than a second resource requirement of the last provisionally accepted request that was dropped by the MEC orchestrator within the current system-level time slot (T sl ), will eventually be dropped or not.

4. The method according to claim 3 , wherein an MEC Platform Manger runs an Offloading Management Service (OMS) that labels the active applications running on the respective edge server so as to discriminate between provisionally admitted and finally accepted task offloading requests, wherein the task broker interacts with the OMS in order to keep track of a load of a respective edge server and to collect labeled training samples generated by the OMS for training a learning algorithm of the task broker.

5. The method according to claim 1 , wherein the task broker is co-located with an edge server of an edge cloud.

6. The method according to claim 1 , wherein the task broker, upon receiving the task offloading request from the UE application, provides the UE application a token to be used by the UE application to interact with the task broker for future requests.

7. The method according to claim 1 , wherein the UE application, whenever it needs to offload a same task to an edge cloud and its previous resource assignment has expired, requests access to the task broker of a last edge server the UE application was assigned to.

8. The method according to claim 1 , further comprising:

receiving, by the task broker, a repetitive task offloading request with a known period from the UE application; and

issuing, by the task broker, proactive requests to the system management level of the MEC system on behalf of the UE application so as to reserve the required resources on an associated edge server by the time the UE application needs them.

9. The method according to claim 1 , wherein the task broker keeps record of a configurable number of latest system management level decisions regarding the allocation of resources of its associated edge server to task offloading requests of UE applications.

10. A multi-access edge computing (MEC) system for managing task offloading to edge servers, the system comprising one or more processors and a task broker implemented at a host management level of the MEC system, wherein the task broker is configured to:

receive a task offloading request from a UE application;

analyze resource requirements of a task of the task offloading request and run a forecasting mechanism that determines whether the task offloading request will eventually be dropped or not by a system management level of the MEC system; and

based on the determination of the forecasting mechanism, either reject or provisionally admit the task offloading request.

11. The system according to claim 10 , wherein the task broker is co-located with an edge server of an edge cloud.

12. The system according to claim 10 , wherein the task broker is further configured to:

receive a repetitive task offloading request with a known period from the UE application, and

issue proactive requests to the system management level of the MEC system on behalf of the UE application so as to reserve required resources on the associated edge server by the time the UE application needs them.

13. The system according to claim 10 , wherein the MEC Platform Manger runs an Offloading Management Service (OMS) that labels the active applications running on respective edge server so as to discriminate between provisionally admitted and finally accepted task offloading requests.

14. The system according to claim 13 , wherein the forecasting mechanism of the task broker includes a learning algorithm, and wherein the task broker is configured to interact with the OMS in order to keep track of a load of a respective edge server and to collect labeled training samples generated by the OMS for training the learning algorithm of the task broker.

15. A tangible, non-transitory computer-readable medium comprising instructions for a task broker for deployment in an MEC system, the instructions, upon being executed by one or more processors, providing for execution of the following steps by the task broker:

receive a task offloading request from a UE application;

analyze resource requirements of a task of the task offloading request and run a forecasting mechanism that determines whether the task offloading request will eventually be dropped or not by a system management level of the MEC system; and

based on the determination of the forecasting mechanism, either reject or provisionally admit the task offloading request.