Skip to content

Automotive storage: powering smart vehicles

Advances associated with increasingly intelligent vehicles are also driving innovation in the storage space, writes Christoph Mutz

Today’s software-defined smart vehicles are transforming the automobile from a hardware-based product to a software-centric device on wheels. Premium vehicles today have up to 150 million lines of software code, distributed among hundreds of electronic control units (ECUs), sensors, cameras, LiDAR and more. This is an industry gaining traction, with the sector expected to generate worldwide revenue of US$4bn in 2024, according to Statista.

In a world already embracing AI, smart vehicles are also tapping into the benefits of this technology. They will be able to process and compute vast volumes of data to perform enhanced functions such as controlling engine power and safety features. At the heart of this transition is embedded AI, which will make it possible to analyse data at the device level (the vehicle) in real-time. However, implementing AI in the vehicle means a fundamental change in vehicle design, as small but powerful, high-capacity memory is required to effectively store and process the data for these AI analyses.

Hardware responses to software demands

The latest technological advances associated with increasingly intelligent vehicles are also driving innovation in the storage space. Western Digital estimates that simple smart applications such as lane keeping assist, cameras, voice recognition and Wi-Fi will reach 2TB of data storage by 2025. With cutting-edge sensors, 5G, AI and machine learning (ML) essential for advancement of connected and increasingly autonomous vehicles, these data storage requirements will only increase. For car manufacturers, this means equipping their vehicles with well-designed systems that meet the specific performance, reliability and capacity requirements imposed by the software-centric future of smart vehicles. With the move towards more data-intensive systems, powerful in-vehicle networks will need to process large amounts of data as efficiently as possible with scalability built in to ensure that they’re futureproof.

Vehicles that increasingly rely on AI for their intelligent decision making will need huge amounts of data to train AI models

Normally, most storage memory is taken up by the operating system (OS), the application software and the log data. With the new centralised high-performance computing system in smart vehicles, the size of the operating system and application software will only continue to increase. In addition, over the air (OTA) updates will require additional storage capacity for maintenance and futureproofing after the vehicle has been delivered. This means that future storage capacity must be considered. While a 128 GB device may be sufficient today, 256 GB or more may be required a few years down the line.

With AI computing moving from the core to the vehicles (or edge), these new, write- and read-intensive workloads will not only determine capacity needs, but also new considerations regarding the performance. There are several interface types from eMMC to UFS (universal flash storage) to NVMe PCIe. The choice of which to use is often dictated by the system-on-chip (SoC) that is selected. Some SoCs only support one or two of these. There are also advantages and disadvantages to each. For example, although NVMe PCIe is fast as an interface, it is slower as a boot device, thus UFS or eMMC would be a better choice if that is the purpose of the device.

In many smart vehicle applications, UFS has emerged as the first choice for high data throughput, with shorter write and read sequences, in conjunction with high-capacity storage. Originally developed for mobile devices, UFS offers next-generation vehicles the benefits derived from its development and optimisation in the mobile market, a field that is subject to high technical demands, and the processing of large amounts of data. Alongside the benefits of high-level data throughput and storage capacity, UFS provides the opportunity for an industry standard platform.

Beyond performance and capacity, data reliability is important as well. Drives need to be extremely rugged in order to operate in harsh conditions, handling vibration and potential impact events. Predictive maintenance will allow automakers to analyse storage performance and health in real time to head off and fix any potential issues.

Quality is the other key measure of a good storage device. Automotive SPICE (ASPICE)—an industry-standard guideline for evaluating the automotive embedded software development process—was introduced to ensure that best practices are used in developing automotive software. Managed NAND Flash products also have software (often referred to as firmware) in their devices. Automotive suppliers and manufacturers are now seeking increasingly for level 3 certified storage products that exceed functional safety standards and deliver high-quality, highly reliable products to help redefine the future of vehicles.

Data intensive applications

For all these new and exciting features offered by innovative smart vehicles, data is critical. Vehicles that increasingly rely on AI for their intelligent decision making will need huge amounts of data to train AI models. As software defined cars may not always have reliable internet connection for cloud computing, embedded AI will remove the need for network connectivity. This means, however, that huge amounts of data must be stored in the vehicle itself.

As cars and other vehicles become smarter, they will act as platforms within the Internet of Things (IoT), with the ability to enhance connectivity, enable autonomous driving, improve safety and security, and adjust settings to best suit the needs of drivers and passengers. This, like other smart applications, is very data intensive. For instance, smart vehicles can adjust temperature settings, navigation, seat recline, or entertainment settings automatically. Some vehicles may even change driving settings from normal to sport and eco based on driver preferences. In order for these vehicles to provide personalised experiences, accessible data on users and their preferences is key.

Moreover, software-defined vehicles will benefit from enhanced maintenance features, which may reduce reliance on mechanics, garages and other experts. Today’s cars flag to drivers when there is an issue that needs to be addressed. Smart vehicles will be able to go a step further and proactively flag glitches before they become problems, which may improve the running and longevity of the vehicle. In some situations, OTA software updates can be used to resolve problems proactively, without the need for human intervention.

One of the biggest benefits of smart vehicles is safety. Vehicles equipped with sensors and embedded AI will be able to flag risks in real-time, such as congestion or even obstacles in the road. This will help improve the safety in the vehicle. In some vehicles, these systems will even be able to proactively steer the vehicle away from an obstacle or bring the vehicle to a safe stop.

While a 128 GB device may be sufficient today, 256 GB or more may be required a few years down the line

If a connected vehicle senses that a driver may have been taken ill at the wheel, the vehicle may get the driver to safety and alert emergency services. These are steps to full self-driving cars, an innovation which removes the need for drivers to take full control. Although this technology is still in testing stages in most locations, these vehicles will likely be approved for road in the future. But before this happens, AI associated with safety functions must be trained to an extremely vigorous standard on high quality data to eliminate any risks.

Despite the enhanced functions of the vehicle, design must still be sleek and appealing to the modern driver. Therefore, storage and computing hardware within the vehicles must be extremely powerful but small to not disrupt the design of the car.

The implications of smart vehicles

The proliferation of onboard software will have a profound implication for the entire automotive industry. For drivers, vehicles will be more responsive and customisable. Soon, software will likely take on more of the driving process in the journey towards self-driving cars.

As cars become smarter, we may even one day have better road safety through AI analytics. Roads and cities may be less likely to suffer severe congestion as real-time data analytics enables better traffic management. Overall, this will improve the driving experience. However, sitting at the heart of this rapid development is the need for vast volumes of data. Therefore, onboard embedded storage will be critical.


About the author: Christoph Mutz is Product Marketing Manager, AME, at Western Digital

Welcome back , to continue browsing the site, please click here