A Race For Storage

September 27, 2019 By King

A Race For Storage

Every once and a while we like to cover an application that drives the use of digital storage and memory. We had a chance to talk recently with Ettienne Reinecke, CTO from NTT about their partnership with the Tour de France over the last five years, NTT is the official tech partner of Amaury Sports Organisation, the organizer of the Tour de France.  NTT provides technical resources for this international race and enabling new ways to experience this multi-day international sporting event.

Here are some details about the technology used in the modern Tour de France. Sensors devices weighing less than 100 grams are mounted beneath the saddle of every rider in the Tour.  These sensors provide real time data on speed and GPS location.  This data is transmitted using a moving mesh-network through gateways on the television motorbikes, helicopters and aircraft, where it is multiplexed with the broadcast video and transmitted to the finish line.

Today In: Innovation

The ‘NTT Big Data Truck’ is located in the NTT Zone at the race finish line and acts as communication and control center, streaming the live data to the NTT Cloud, and to the television graphics.

NTT’s cloud based real time analytics platformprocesses millions of data points a minute, organizing and distributing it to a global team of technologists, data scientists, and marketers to create the stories that have come to define the Tour de France viewing experience.  Their data produces real-time data insight into the race situation, team tactics and individual’s performances.

800 MB of raw data and 2.7 GB of enriched data for speed and GPS location (per stage) is captured by the sensors and augmented with 53 attributes (wind, altitude, etc.), collecting millions of records per stage, stored on this cloud platform. The stage predictions were updated live throughout the stage based on the events occurring within the raceand data gathered and stored from races over five years.

The NTTPredictor machine learning engine made live race predictions such as stage favorites, catch the break predictions, and estimated time of arrival for the peloton at any point on the course.

The 3rdgeneration Stage Favorites model was significantly enhanced to increase the factors it considers, including the performance of riders with similar characteristics, and the profile of the rider’s team.  It uses 29 input features as input to a model comprising of 150,000 decision trees with over 60,000 unique decision nodes!

Through data storytelling, NTT with the Tour de France provides an immersive viewing experience across broadcast, social and digital channels attracting a new demographic of fans to the sport. By the end of 2018’s event, over 135 million data points had been captured, analyzed and transformed into compelling stories for fans to enjoy across mobile, social, digital and broadcast.

In the 2019 Tour de France, which ran over most of July, the official mobile app featured an improved “Estimated Time of Arrival” feature which uses Machine Learning to provide fans at the race with a predication of when the race will arrive at their location, based on the live tracking data.   This improved user experience to make it easier for them to access and view live tracking data, particularly on mobile devices.

Automation was used to simplify race setup, and provide API integrations with key data partners.  This data is streamlining into the data pipelines and providing more regular and more reliable updates.  3D projection views of the stages were available for fans to follow the event.

At the end of the 2019 Tour de France, NTT predictions were on par with a veteran cycling journalist with 15 years of experience.  The 3rd generation Stage Favorites model was significantly enhanced to increase the factors it considers including the performance of riders with similar characteristics, and the profile of a rider’s team.

NTT’s technology enabled a more immersive Tour de France fan experience using IoT technology in the field and artificial intelligence to track and make predictions on the race outcome.