Sinelec Delivers Smart Roads Vision with Neosperience Using AWS

Sinelec is the technology arm of ASTM Group, one of the largest toll motorway operators, by road miles, in the world. The company is building a platform for future vehicle communication. The solution is already in operation, but the company is also preparing for massive growth as the technology is eventually included in every new car. With the potential for Sinelec’s systems to scale from zero to all V2X-capable vehicles, it partner Neosperience chose AWS to build the scalable solution Sinelec needed.

Sinelec is the technology arm of ASTM Group, one of the largest toll motorway operators, by road miles, in the world. Sinelec runs nearly 900 toll collection lanes and processes about €3 billion in fees every year. Its operation and traffic control platform manages 1,400 km of Italian roads and processes 1.5 million journeys a day.  

The company is using its expertise in smart mobility and road management to build a next-generation vehicle communication network, working side by side with car manufacturers. This then links to Amazon Virtual Private Cloud (Amazon VPC) via an AWS Site-to-Site VPN. The system, which complies with European Telecommunications Standards Institute (ETSI) specifications, allows two-way communication between cars and road safety management systems. Drivers will benefit from a mixture of traffic information sourced directly from other connected cars, as well as additional information from other networks and third parties like the police or road maintenance services. 

All new cars sold in the European Union since 2018 already include eCall technology, which automatically sends precise location data to emergency services if the car crashes, and activates a microphone in the vehicle to allow communication with emergency dispatchers. 

Sinelec is already working with road operators and city authorities to help reduce congestion and improve air quality in cities. In future, vehicle-to-infrastructure (V2I) communication will allow these organizations to control traffic systems in real time based on real data. This will let them use vehicle data to build machine-learning models to predict traffic problems before they arise, and take action to reduce them. Such systems will also be able to warn about the presence of accident black spots and make driving safer.

Sinelec currently uses its control rooms and intelligent camera systems and sensors to run its toll booth systems effectively and to aid traffic management. Its staff respond to events and help the emergency services react quickly to events on the roads.

The main components of a V2I system are the roadside antennas that collect data from vehicles and a central C-ITS application hosted on a datacenter that ingests and analyzes all the data collected from the antennas.

Sinelec chose AWS Partner Neosperience to help build its C-ITS platform, networking, and analytics capabilities. The two companies use AWS to provide a rapidly scalable infrastructure as they move from pilot projects involving a few hundred vehicles to widespread adoption with thousands of vehicles over the next few years.

Neosperience works with both Sinelec and the roadside antenna providers that collect the data from vehicles. Neosperience is also able to provide extensive know-how on Kubernetes, which was chosen as the best platform to use for the central application.

The main challenge of the project is scalability. In the future, each vehicle will be able to exchange, through its on-board-unit (OBU), a large amount of data with the roadside infrastructure. That data must be received and pre-processed by the roadside antenna before transmission to central systems. Even if the communication protocol is a global standard provided by ETSI, the antennas are developed by different manufacturers that usually add some customization and use different transmission technologies, adding another layer of complexity. Using AWS, Sinelec can manage and balance the communication across these various channels.

Simone Conso, head of connected mobility at Sinelec explains: “With today’s IT tools, it’s quite easy to develop cloud-based software, but when high reliability, high scalability, and security come into play, you have to deal with a major technology challenge that requires the support of the right technology partners.”

Each vehicle sends its location in real time, together with data from its onboard sensors. With such data-intensive protocols, there are huge problems around managing the amount of data, bandwidth, and real-time analysis.

Every packet from every vehicle must be collected, analyzed, and stored for further statistical analysis and modelling. Sinelec’s on-premises system is already able to manage the current amount of data coming from road-side sensors and camera streams. However, the number of connected vehicles is growing rapidly and a typical stretch of motorway may have thousands of vehicles simultaneously sending their OBU data to the roadside infrastructure. Making on-premises systems scale while the traffic of vehicle-to-everything (V2X)-capable vehicles grows is costly and very challenging to predict and manage.

In order to be compliant to the European safety rules, systems that process vehicle data need to work in near real-time when collecting and sending data back to the vehicles. The near real-time requirement is useful, for example, to warn of a crash risk or a danger along the road.

Scalability is a challenge at every stage of the data journey: each roadside antenna needs to receive simultaneous messages from hundreds of cars, and the central systems need to receive simultaneous messages from hundreds of antennas. The data throughput will also increase exponentially in case of traffic slowdowns and jams.

Machine learning will play an important role for the analysis of the huge amount of data provided by the vehicles and will help to provide a variety of actionable insights for different stakeholders. Increased vehicle vibration, for instance, could be evidence of a tire problem for an individual vehicle. But if the vibration data comes from different vehicles in a particular section of road, it might be an early warning for the road operator that the road surface needs inspection or repair.

By truly digitizing road transport, pollution could be dramatically reduced. The suggested speed limit is automatically adjusted according to the scenario in order to optimize the traffic flow and increase road safety. Emergency vehicles and ambulances could have special status so the system could speed them through city environments.

Working in V2X technologies comes with different challenges. Data formats and standards are evolving fast.

Roadside antennas and car manufacturers use a variety of software standards, libraries, and software development kits that are still under development and constantly changing. Network bandwidth is still a huge bottleneck for data transmission and many algorithms and methods for data cleaning and data compression must be used during the various steps of the data journey.

Sinelec is currently installing its network of roadside antenna on motorways and is testing and studying real case scenarios with the V2I system end-to-end, starting from vehicle OBU and ending in its proprietary motorway operational control systems. Already up and running in Italy, the company is about to start trials in other countries.

Sinelec can rapidly scale-up its systems using Amazon Elastic Compute Cloud (Amazon EC2) to provide on-demand compute power, and AWS Auto Scaling groups to dynamically resize the Amazon EC2 fleet based on traffic and load. These processes are automated using Amazon Elastic Kubernetes Service (EKS) to manage the Kubernetes cluster that automatically orchestrates and scales the application containers. This capability is vital for V2I systems because a constant flow of V2X-ready vehicles will be put on the road by car manufacturers in the next few years. Data transmission flow and computational spikes will be unpredictable until it’s possible to accurately provide precise predictive models. By relying on a cloud infrastructure that can be scaled in a few moments, Sinelec’s C-ITS platform is already able to cope with future computational loads.