Claims:We claim:
1. A method of generating synchronized labelled training dataset (213) for building a learning model, the method comprising:
determining, by a training data generation system (107), timing parameters related to a User Equipment (UE) (101) capable of receiving a multimedia content and timing parameters related to one or more network nodes (103) used to facilitate streaming of the multimedia content to the UE (101);
determining, by the training data generation system (107), a timing advance factor based on the timing parameters related to the UE (101) and the timing parameters related to the one or more network nodes (103) to achieve time synchronization between the UE (101) and the one or more network nodes (103);
signalling, by the training data generation system (107), the UE (101) to initiate playback of the multimedia content based on the timing advance factor;
receiving, by the training data generation system (107), network Key Performance Indicator [KPI] data (209) from the one or more network nodes (103) and a user experience data (211) from the UE (101), concurrently, for the streamed multimedia content; and
correlating, by the training data generation system (107), the user experience data (211) with the corresponding network KPI data (209), based on timestamp corresponding to the user experience data (211) and the network KPI data (209), to generate a synchronized labelled training dataset (213) for building a learning model.
2. The method as claimed in claim 1, wherein the user experience data (211) comprises at least one of UE (101) Identifier (ID), multimedia content type, predefined sample intervals, Mean Opinion Score (MOS), timestamp of the MOS and a label type.

3. The method as claimed in claim 1, wherein the network KPI data (209) comprises at least one of type of network KPIs and corresponding data types, network node layer Identifier (ID), network KPI record count per sample, network KPI values, predefined sample intervals and timestamp of network KPI records.

4. The method as claimed in claim 1, wherein the timing parameters related to the UE (101) comprises of Round Trip Time (RTT) for receiving the user experience data (211).

5. The method as claimed in claim 1, wherein the timing parameters related to the one or more nodes, comprises at least one of, RTT for receiving the network KPI data (209), predefined sample intervals, and timestamp of network KPI records per sample.

6. The method as claimed in claim 1, wherein the user experience data (211) and the network KPI data (209) are received by the training data generation system (107) at predefined sample intervals.

7. The method as claimed in claim 1, wherein the learning model is built using one or more predefined model building techniques.

8. The method as claimed in claim 1 further comprises deploying, by the training data generation system (107), the learning model in an external analytics system (115) for enabling the external analytics system (115) to perform real-time predictions of user experience when the multimedia content is streamed.

9. The method as claimed in claim 8 further comprises validating, by the training data generation system (107), accuracy of the real-time predictions of the user experience by comparing the user experience data (211) received from the UE (101) with the real-time predictions of the user experience.

10. The method as claimed in claim 9 further comprises rebuilding, by the training data generation system (107), the learning model, at regular intervals, based on current user experience data (211) and current network KPI data (209), when result of the validation is negative.

11. The method as claimed in claim 1, wherein the user experience data (211) is determined by at least one of predefined deep learning techniques configured in the UE (101), or by a Subject Matter Expert (SME) associated with the UE (101).

12. The method as claimed in claim 1, wherein the UE (101) and the one or more network nodes (103) are preconfigured with parameters related to the network KPI data (209), the time synchronization and a rebuilding criteria for the learning model, by a Central Configuration Manager (CCM) (105) associated with the training data generation system (107).

13. The method as claimed in claim 1 further comprises:

determining, by the training data generation system (107), a transmission rating factor based on one or more network parameters related to streaming of the multimedia content to the UE (101); and

evaluating, by the training data generation system (107), quality of the user experience data (211) received from the UE (101), based on the transmission rating factor, to determine usability of the user experience data (211) for building the learning model, prior to correlating the user experience data (211) with the network KPI data (209).

14. A training data generation system (107) for generating synchronized labelled training dataset (213) for building a learning model, the training data generation system (107) comprising:
a processor (109); and
a memory (113) communicatively coupled to the processor (109), wherein the memory (113) stores the processor (109)-executable instructions, which, on execution, causes the processor (109) to:
determine timing parameters related to a User Equipment (UE) (101) capable of receiving a multimedia content and timing parameters related to one or more network nodes (103) used to facilitate streaming of the multimedia content to the UE (101);
determine a timing advance factor based on the timing parameters related to the UE (101) and the timing parameters related to the one or more network nodes (103) to achieve time synchronization between the UE (101) and the one or more network nodes (103);
signal the UE (101) to initiate playback of the multimedia content based on the timing advance factor;
receive network Key Performance Indicator [KPI] data (209) from the one or more network nodes (103) and a user experience data (211) from the UE (101), concurrently, for the streamed multimedia content; and
correlate the user experience data (211) with the corresponding network KPI data (209), based on timestamp corresponding to the user experience data (211) and the network KPI data (209), to generate a synchronized labelled training dataset (213) for building a learning model.
15. The training data generation system (107) as claimed in claim 14, wherein the user experience data (211) comprises at least one of UE (101) Identifier (ID), multimedia content type, predefined sample intervals, Mean Opinion Score (MOS), timestamp of the MOS and a label type.

16. The training data generation system (107) as claimed in claim 14, wherein the network KPI data (209) comprises at least one of type of network KPIs and corresponding data types, network node layer Identifier (ID), network KPI record count per sample, network KPI values, predefined sample intervals and timestamp of network KPI records.

17. The training data generation system (107) as claimed in claim 14, wherein the timing parameters related to the UE (101) comprises of Round Trip Time (RTT) for receiving the user experience data (211).

18. The training data generation system (107) as claimed in claim 14, wherein the timing parameters related to the one or more nodes, comprises at least one of, RTT for receiving the network KPI data (209), predefined sample intervals, and timestamp of network KPI records per sample.

19. The training data generation system (107) as claimed in claim 14, wherein the processor (109) receives the user experience data (211) and the network KPI data (209) at predefined sample intervals.

20. The training data generation system (107) as claimed in claim 14, wherein the learning model is built using one or more predefined model building techniques.

21. The training data generation system (107) as claimed in claim 14, wherein the processor (109) is further configured to deploy the learning model in an external analytics system (115), for enabling the external analytics system (115) to perform real-time predictions of user experience when the multimedia content is streamed.

22. The training data generation system (107) as claimed in claim 21, wherein the processor (109) is further configured to validate accuracy of the real-time predictions of the user experience by comparing the user experience data (211) received from the UE (101) with the real-time predictions of the user experience.

23. The training data generation system (107) as claimed in claim 22, wherein the processor (109) is further configured to rebuild the learning model, at regular intervals, based on current user experience data (211) and current network KPI data (209), when result of the validation is negative.

24. The training data generation system (107) as claimed in claim 14, wherein the user experience data (211) is determined by at least one of predefined deep learning techniques configured in the UE (101), or by a Subject Matter Expert (SME) associated with the UE (101).

25. The training data generation system (107) as claimed in claim 14, wherein the UE (101) and the one or more network nodes (103) are preconfigured with parameters related to the network KPI data (209), the time synchronization and a rebuilding criteria for the learning model, by a Central Configuration Manager (CCM) (105) associated with the training data generation system (107).

26. The training data generation system (107) as claimed in claim 14, wherein the processor (109) is further configured to:

determine a transmission rating factor based on one or more network parameters related to the streaming of the multimedia content to the UE (101); and

evaluate quality of the user experience data (211) received from the UE (101), based on the transmission rating factor, to determine usability of the user experience data (211) for building the learning model, prior to correlating the user experience data (211) with the network KPI data (209).
Dated this 17th day of March, 2018

SWETHA S N
IN/PA-2123
OF K & S PARTNERS
AGENT FOR THE APPLICANT
, Description:TECHNICAL FIELD
The present subject matter relates generally to machine learning model, and more particularly, but not exclusively to a method and a system for generating synchronized labelled training dataset for building a learning model.