With all of the concern over AI and the loss of jobs it could entail, no one is talking about the imminent demise of TV color commentators. And yet, improbable as it may seem, the likes of Howard Cosell and Bob Costas might just be replaced someday by an algorithm. An offshoot of data and video analytics, the field of automated sports analytics is providing performance intelligence for many professional sports teams and broadcasters today — and the field is growing quickly. Like Moneyball, but in real time, automated sports analytics now provide live intelligence to backroom and field-side coaches, as well as enriching the experience for fans in the stadium and making life easier for television commentators.
This revolution in sports analysis and performance management is rooted in sophisticated machine learning and AI software that is capable of not only crunching data, but also analyzing on-field video footage as well as providing real-time feedback and intelligence on player performance, among other things. But all of this software intelligence would go nowhere without a high-performance network to link HDTV cameras, coaching staff, databases and the cloud. Software-defined wide area networking, or SD-WAN, is a key part of delivering this service, as well as making it manageable and cost-effective.
The typical sports analytics system uses multiple high-resolution cameras, advanced deep learning and computer vision technologies to generate accurate on-pitch data. It tracks the movement of all players, the ball and referees, and plots them in three-dimensional space. It sends multiple high-definition live video streams, recorded during the game, to an analytics engine that is typically hosted in the cloud. Feedback is produced very quickly, with analytical results after only 10 seconds of reviewing video streams, allowing a coach to dynamically see the results of her tactical moves on the bench, or empowering fans with next-generation statistics and visuals displayed on a TV or the web.
Analyzing HDTV footage is at the heart of this kind of application. The typical system requires visibility, control and delivery of high-definition video streams from 10 to 20 different cameras around the stadium. The very high broadband and near-real-time nature of the application puts enormous demands on the network to deliver guaranteed performance without risking the security or integrity of the data.
To accommodate these requirements, a single full mesh SD-WAN VPN overlay is ideal. It can be deployed using virtual customer premises equipment (vCPE) endpoints from each stadium to the analytics engine in the cloud. Head offices and partner agencies that use the data for non-real-time processing can also be connected with universal CPE devices.
More than one WAN transport network underlay should be used to ensure availability, as this is a mission-critical application. For instance, you might have a fiber-based internet broadband connection — e.g., 1 Gbps — and, for resiliency purposes, each stadium might be connected to two separate underlay WAN transport networks such as IP-MPLS or even 4G LTE. All active links to the SD-WAN are connected to these redundant links.
This application also puts a premium on security due to the competitive environment and the need to keep player data and coaching moves secret. Every SD-WAN link should be encrypted via IPSec to secure the video content. In addition, each video flow should be securely isolated from the rest of the network using microsegmentation, thus providing a further level of protection to the other network resources.
Ideally, the application can use policies to automate both performance and security assurance. For assurance sake, SD-WAN link performance must be measured constantly for packet loss, latency and jitter. If these cannot be assured, or if one of the two independent SD-WAN vCPE endpoints fails, the network would switch to the redundant WAN link for full automated resiliency. Equally, the SD-WAN would be able to take automatic remedial security actions based on suspicious network behavior that is detected in real time and without user action.
With all the management issues that most sports organizations face, having a demanding and sophisticated network might seem like one prima donna too many. But the beauty of SD-WAN is its manageability. It is highly flexible, being able to adapt to almost any kind of underlying WAN link or links. Many of the key network management issues can be automated by policies and, to the extent that hands-on management is required, routine tasks can usually be handled from a single pane of glass. Integration with the cloud also means that processing resources can be switched on and off as needed, without having to manage any kind of data center operations.
If this all seems too good to be true, remember that the power of machine learning and AI, which is driving new applications like sports analytics, is also creating a revolution in networks. Software-driven networking and the cloud lie at the heart of SD-WAN, and SD-WAN makes the management of highly sophisticated networks relatively simple and far less costly than only a few years ago. If SD-WAN is the power hitter that is making sports analytics possible, it is also providing a platform for the use of analytics, AI and machine learning in a wide variety of fields and industries.
All IoT Agenda network contributors are responsible for the content and accuracy of their posts. Opinions are of the writers and do not necessarily convey the thoughts of IoT Agenda.