A system of user experience modeling based on machine learning for video applications

Description

[0001] A SYSTEM OF USER EXPERIENCE MODELING BASED ON MACHINE LEARNING FOR VIDEO APPLICATIONS

[0002] Technical Field

[0003] The present invention relates to a system which enables to create a pattern by measuring the data usage in video applications of users on a network and to compare the pattern with a data transfer model requested to be obtained.

[0004] Background of the Invention

[0005] Today, with the introduction of 5G mobile communication technologies into our lives, network architectures have increasingly become even more complex and this leads to the need for managing customer experience in a smarter and more dynamic way as well. Particularly due to the influence of video streaming applications and social media contents, mobile video traffic has increased significantly in recent years and become one of the most important applications which determine the user experience. Most of the metrics monitored by means of radio and core network units express quality of service (QoS) in network performance management, however this expression remain incapable to reflect quality of experience (QoE). In the current technique, field measurement is the most common and traditional method to model the user experience over an application. However, this solution is This solution is quite costly in terms of both time and money and it also has time limitation due to the fact that it can only be carried out on certain routes.

[0006] Considering the studies included in the current technique, it is understood that there is need for a system which enables to predict video quality over KPIs (key performance indicator) obtained only in the network interface independently of location and time. The United States patent document no. US20190379589A1, an application in the state of the art, discloses a system for examining network traffic and creating a pattern by detecting time-dependent flow differences. In the current invention, the performance monitoring data comprises data such as packet data, service and traffic data. Packet data comprises data such as bandwidth, latency jitter, error rate. Video quality metrics may be buffering events, changes of video packet rate or video conference feedback of user. Methods disclosed in the said invention comprises detection of network anomalies and taking preventive actions by means of multiple performance monitoring data time-series and single approach.

[0007] Summary of the Invention

[0008] An objective of the present invention is to realize a system which enables to convert network service quality into user experience in the light of field measurement data gathered by network and application.

[0009] Another objective of the present invention is to realize a system which enables to make a prediction of video quality (vMOS-video mean opinion square) from basic- level radio network KPIs by utilizing machine learning techniques.

[0010] Another objective of the present invention is to realize a system which enables to train a mode by means of field measurement data received at different locations, routes and time zones and to improve the results obtained, i.e. enhance the prediction success of an algorithm.

[0011] Detailed Description of the Invention

[0012] The “system of user experience modeling based on machine learning for video applications” realized to fulfil the objectives of the present invention is shown in the figures attached, in which: Figure l is a schematic view of the inventive system.

[0013] The components illustrated in the figure are individually numbered, where the numbers refer to the following:

[0014] 1. System

[0015] 2. Electronic device

[0016] 3. Pattern recognition server

[0017] 4. Prediction server

[0018] 5. Anomaly server

[0019] The inventive system (1) which enables to create a pattern by measuring the data usage in video applications of users on a network and to compare the pattern with a data transfer model requested to be obtained comprises: at least one electronic device (2) which is configured to run at least one application on itself and to enable users to access video contents; at least one pattern recognition server (3) which is configured to determine at least one pattern related to the quality of a video that is watched on the electronic device (2); at least one prediction server (4) which is configured to match performance indicators by using user-based tracks related to the quality of a video that is watched on the electronic device (2); at least one anomaly server (5) which is configured to establish communication with the pattern recognition server (3) and the prediction server (4), and to determine anomalies by analysing outputs.

[0020] The electronic device (2) included in the inventive system (1) is configured to run at least one application on itself and to enable users to access video contents. The electronic device (2) is a device such as mobile phone, tablet, computer. The patern recognition server (3) included in the inventive system (1) is configured to receive and then analyse parameters of video quality related to the video contents that are watched over the electronic device (2), and to create video-based pattern models.

[0021] The prediction server (4) included in the inventive system (1) is configured to match the radio performance indicators gathered by the network by using the measurements related to the user-based network uses gathered in different geographic locations and at various times of the day in video quality measurements related to video contents that are watched over the electronic device (2).

[0022] The anomaly server (5) included in the inventive system (1) is configured to receive the pattern models obtained from the pattern recognition server (3) and the matching results obtained from the prediction server (4) and then to carry out an analysis on the received data. The anomaly server (5) is configured to compare a predicted video quality with a predetermined typical patern and to detect the basic parameters affecting the anomaly as a result of the comparison.

[0023] Industrial applicability of the invention

[0024] In the inventive system (1), the pattern recognition server (3) determines a typical behaviour pattern in order to determine what to expect on the end user part in accordance with the content accessed on the electronic device (2) at first. Then, video MOS is predicted by the prediction server (4) by utilizing the critical radio network KPIs selected on the basis of correlation analysis. As a final step, the anomaly server (5) carries out an anomaly detection by comparing video MOS values with typical MOS values.

[0025] With the present invention, a system which enables to train a mode by means of field measurement data received at different locations, routes and time zones and to improve the results obtained, i.e. enhance the prediction success of an algorithm, is obtained.

[0026] The electronic device (2), the pattern recognition server (3), the prediction server (4) and the anomaly server included in the inventive system (1) submit information and approval to the user in accordance with the principles of data privacy and they operate within the scope of the Law on Protection of Personal Data (KVKK).

[0027] Within these basic concepts; it is possible to develop a wide variety of embodiments of the inventive system (1); the invention cannot be limited to examples disclosed herein and it is essentially according to claims.

Claims

CLAIMS1. A system (1) which enables to create a pattern by measuring the data usage in video applications of users on a network and to compare the pattern with a data transfer model requested to be obtained; comprising at least one electronic device (2) which is configured to run at least one application on itself and to enable users to access video contents; and characterized by at least one pattern recognition server (3) which is configured to determine at least one pattern related to the quality of a video that is watched on the electronic device (2); at least one prediction server (4) which is configured to match performance indicators by using user-based tracks related to the quality of a video that is watched on the electronic device (2); at least one anomaly server (5) which is configured to establish communication with the pattern recognition server (3) and the prediction server (4), and to determine anomalies by analysing outputs.

2. A system (1) according to Claim 1; characterized by the electronic device (2) which is configured to run at least one application on itself and to enable users to access video contents such as mobile phone, tablet, computer.

3. A system (1) according to Claim 1 or 2; characterized by the pattern recognition server (3) which is configured to receive and then analyse parameters of video quality related to the video contents that are watched over the electronic device (2), and to create video-based pattern models.

4. A system (1) according to any of the preceding claims; characterized by the prediction server (4) which is configured to match the radio performance6indicators gathered by the network by using the measurements related to the user-based network uses gathered in different geographic locations and at various times of the day in video quality measurements related to video contents that are watched over the electronic device (2). A system (1) according to any of the preceding claims; characterized by the anomaly server (5) which is configured to receive the pattern models obtained from the pattern recognition server (3) and the matching results obtained from the prediction server (4) and then to carry out an analysis on the received data. A system (1) according to any of the preceding claims; characterized by the anomaly server (5) which is configured to compare a predicted video quality with a predetermined typical pattern and to detect the basic parameters affecting the anomaly as a result of the comparison.7

Images