2D Live Sports Tracking using Azure IoT and BigData Technologies by Andreas Fitzi

This session shows an example architecture and implementation of a real-time 2D Football Position Tracking solution in Azure. The solution uses Azure IoT and BigData technologies for the processing and storing, as well as the D3 JavaScript Framework for the visualization of position data. The base data for the solution is tracking data of soccer players from three different soccer games. Each dataset contains 20 position records per second, for each of the 11 players of the home team. The position data is sent to an Azure IoT Event Hub in chronological order, streamed and processed with a HDInsight Storm Cluster and live-visualized on a 2D soccer field, implemented with the D3 JavaScript visualization library. Additionally, the data is stored in parallel in an HDInsight HBase Cluster or an Azure SQL database in order to allow the creation of analytical reports on top of the collected position data.

Senior Consultant Business Intelligence at Trivadis AG

Transcript

1. 2D Live Sport Tracking in Azure Andreas Fitzi Business Intelligence Consultant Trivadis
2. 2D Live Sport Tracking in Azure 3 24.04.2017
3. BASLE BERN BRUGG DÜSSELDORF FRANKFURT A.M. FREIBURG I.BR. GENEVA HAMBURG COPENHAGEN LAUSANNE MUNICH STUTTGART VIENNA ZURICH Proof of Concept 2D Live Sport Tracking in Azure
4. Agenda 2D Live Sport Tracking in Azure 5 24.04.2017 Challenge Initial Situation / Objective Base Data Solution Architecture Staging Layer PMAS Layer Analytics Layer Result Live Report Speed Analysis Defending Space Analysis Summary
5. 2D Live Sport Tracking in Azure 6 24.04.2017 Challenge
6. Initial situation 2D Live Sport Tracking in Azure 7 24.04.2017 Sport tracking System „ZXY Sport tracking System“ of a Norwegian Start-up Sensor data of Football games – Movement data 11 fixed point Radio Receivers Sensor-Belt for players full data sets of 3 games free available for research
7. Objective 2D Live Sport Tracking in Azure 8 24.04.2017 Player Movement Analytics Infrastructure Big Data architecture in Cloud Storing, Processing & Visualization of movement data (Live-Stream) 2D Analysis of the recorded data (90min) Analysis in Performance Portal (Speed, Covered distance etc.) Speed Analysis Defensive Space Analysis
8. Base data 2D Live Sport Tracking in Azure 9 24.04.2017 Base data (ZXY Sport tracking System) Player movement data for 3 Football games 20 records for a player in a second Ca. 1.2 million records for 11 players in 90min Attribute Value Description Timestamp 19:03:51:03 Timestamp Tag 1 Player Number x_pos 62.68070798 Position X on Football pitch y_pos 13.71071585 Position Y on Football pitch heading -2.058309619 Targeted position in radian measure direction 1.593030775 Current running direction in radian measure energy 1151.73682 Energy (undefined measurement) speed 0.930764441 Speed in m/s total_distance 871.0633421 Total covered distance in meters
9. 2D Live Sport Tracking in Azure 10 24.04.2017 Solution
10. Architecture 2D Live Sport Tracking in Azure 11 24.04.2017 Movement- & Performance-Data Local Staging Area Analytics Player Movement Analytics System (PMAS) Game 1 Game 2 Game 3 Masterdata Masterdata Staging Database Staging Masterdata EventQueue (C# Console App) Event Hub Spout Bolt Bolt HBase Store SQL Database 2D Live Stream 2D Player Movement Analysis Player Analytics & Statistics Portal
11. Staging Layer 2D Live Sport Tracking in Azure 12 24.04.2017 Movement- & Performance-Data Local Staging Area Analytics Player Movement Analytics System (PMAS) Game 1 Game 2 Game 3 Masterdata Masterdata Staging Database Staging Masterdata EventQueue (C# Console App) Event Hub Spout Bolt Bolt HBase Store SQL Database 2D Live Stream 2D Player Movement Analysis Player Analytics & Statistics Portal
12. Staging Layer 2D Live Sport Tracking in Azure 13 24.04.2017 Staging Area (local environment) Text files with movement data SQL Server (Staging Area) SSIS for the transfer of the movement data into SQL Server Event Queue (local environment) C# Console application Reads data from SQL Server Sends data to the Event Hub (Azure)
13. PMAS Layer 2D Live Sport Tracking in Azure 14 24.04.2017 Movement- & Performance-Data Local Staging Area Analytics Player Movement Analytics System (PMAS) Game 1 Game 2 Game 3 Masterdata Masterdata Staging Database Staging Masterdata EventQueue (C# Console App) Event Hub Spout Bolt Bolt HBase Store SQL Database 2D Live Stream 2D Player Movement Analysis Player Analytics & Statistics Portal
14. PMAS Layer 2D Live Sport Tracking in Azure 15 24.04.2017 PMAS (Azure) IoT Event Hub (receives data from devices) HDInsight Storm Cluster (Storm Topology with Spouts and Bolts) HDInsight HBase Cluster (NoSQL Column Family Database) SQL Azure Database
15. Analytics Layer 2D Live Sport Tracking in Azure 16 24.04.2017 Movement- & Performance-Data Local Staging Area Analytics Player Movement Analytics System (PMAS) Game 1 Game 2 Game 3 Masterdata Masterdata Staging Database Staging Masterdata EventQueue (C# Console App) Event Hub Spout Bolt Bolt HBase Store SQL Database 2D Live Stream 2D Player Movement Analysis Player Analytics & Statistics Portal
16. Analytics Layer 2D Live Sport Tracking in Azure 17 24.04.2017 2D-Live Stream & 2D Player Movement Analysis (Azure) ASP.Net MVC Web application SignalR with Web Sockets for Real-Time communication D3 Javascript library for visualization Player Analytics & Statistics Portal (Azure) ASP.Net MVC Web application D3 Javascript library for visualization
17. 2D Live Sport Tracking in Azure 18 24.04.2017 Result
18. Result 19 24.04.2017 2D Live Sport Tracking in Azure 2D Live Report 2D Live stream of a game
19. Result 20 24.04.2017 2D Live Sport Tracking in Azure 2D Analysis 2D Replay of recorded football games Coloring of fast running players
20. Result 21 24.04.2017 2D Live Sport Tracking in Azure Speed Analysis Thesis: Speed of a player declines over 90 minutes Analysis of the maximum and average Speed by player No decrease in the maximum speeds of the players Decrease of average speed for some players (Substitution?)
21. Result 22 24.04.2017 2D Live Sport Tracking in Azure Defensive Space Analysis Thesis: Big Gaps in defending line can lead to goal opportunities for opponent Analysis of situations with where defensive line is not compact
22. 2D Live Sport Tracking in Azure 23 24.04.2017 Summary
23. Summary 2D Live Sport Tracking in Azure 24 24.04.2017 Easy to leverage Big Data as a Service within Azure Challenges with the compatibility of the different Technologies and Services Can be easily adapted for other scenarios Visualization of special events directly within the stream Interesting Scenarios for Coaches and Fans Data will win games
24. Q&A Andreas Fitzi Consultant Business Intelligence andreas.fitzi@trivadis.com