The software-defined vehicle (SDV) is a significant evolution from traditional vehicles to ones controlled primarily by software. This transformation enables advanced digital driver assistance systems, increases safety and security, and connects vehicles with other road users and the road infrastructure. SDVs also stay up to date throughout their lifecycle through over-the-air (OTA) updates. However, the shift to SDVs also brings the challenge of managing enormous amounts of data.

Solutions for current and future data management challenges

The amount of data is increasing, both inside the vehicle due to increasingly complex onboard functions and outside the vehicle due to increased connectivity. Particularly in the development of vehicles, these data volumes present engineers with ever greater challenges when it comes to error analysis and rectification.

"Modern vehicles are complex due to software-driven vehicle architecture and the wide range of driver assistance systems. This makes it necessary to efficiently manage and analyze data, even in the petabyte range. To meet these challenges, we have developed IAV Merida", says Dr. Remo Lachmann, data expert at IAV. The SaaS solution makes it easier to handle the constantly growing mountain of data in the development of the SDV. It helps to store, manage and analyze vehicle data efficiently.

Data as the basis for efficient testing

IAV Merida receives data from a variety of sources, from data loggers or telemetry units installed in vehicles. These record the relevant vehicle data in a targeted manner and transmit it to IAV Merida. This data is crucial for identifying sources of error. "Based on the data, we can see where new connectivity functions fail or where the vehicle as a whole shows unexpected behavior", explains Lachmann.

Before development vehicles come onto the market, they undergo extensive analyses and tests to verify their quality. Tests can be carried out both on the road and on the test stand. It doesn't matter whether they are cold-country or hot-country tests, or mileage burning endurance tests. "During these tests, we observe certain behaviors and new functions or features. Each test produces a measurement that we efficiently store and analyze", explains Lachmann. This makes it possible to capture data streams in the car, for example from the engine to the central control unit or to the display. This communication is permanently recorded and stored in IAV Merida.

Storing gigantic amounts of data

These extensive tests mean that IAV Merida stores between 200 and 300 terabytes of new vehicle data every month. “In total, Merida processes more than 5 petabytes of data per month, because each measurement is subjected to various analyses. These gigantic amounts of data are currently stored on IAV's own premises. However, IAV Merida can also be operated in a cloud infrastructure such as Azure or Tencent, according to customer requirements and needs.”, Lachmann says.

Worldwide data management and compliance

“We are able to host data worldwide in compliance with the law and ensure that all local regulations are adhered to”, Lachmann says. In Germany, IAV currently relies on Microsoft Azure to operate Merida, but other providers are possible.

IAV Merida is also used by customers in China. This is a challenge because China has very strict data protection regulations. IAV is therefore working with the local cloud provider Tencent. “In China, exporting geodata is prohibited. That is why we work with local providers to ensure compliance. This means that the full range of IAV Merida functions can be used.”

Automated analyses and interactive dashboards

“IAV Merida can extract valuable insights from complex measurement data. Our more than 300,000 analyses per month are highly automated and generate reports in the form of dashboards that contain, for example, diagrams, tables, statistics and geographical maps”, says Lachmann.

Customers can choose from a range of analysis options both before and during testing. IAV Merida then performs these automatically for each new measurement. The reports are interactive and can also be analyzed and displayed live if required. This is particularly useful for observing the vehicle's behavior while it is in motion, allowing you to react to any abnormalities as the tests are being carried out. The analyzes and reports can be applied to both individual vehicles and entire fleets.

Integration of artificial intelligence and AI-supported anomaly detection

In addition to statistical and rule-based analyses, IAV Merida also offers the option of analyzing complex systems with AI-based algorithms. The AI algorithms are used to detect faults or patterns that cannot be detected at all or only with a great deal of effort using conventional methods. Lachmann explains: “Using AI algorithms in Merida enables us to detect unknown faults or patterns, which is essential for developing modern vehicles.”

One special feature of Merida is the AI-supported anomaly detection. This calculates an anomaly score that detects failures and defects in advance. This means that customers can be informed about possible problems in advance, before they actually occur.