Home » knowledge » Technical Article – Smart Storage Improves Reliability of Self-Driving Vehicles

Technical Article – Smart Storage Improves Reliability of Self-Driving Vehicles

The advent of self-driving cars will dramatically change our travel habits and bring about a dramatic change in the transportation industry. The digital transformation of the automotive industry will bring many societal benefits, such as fewer accidents, lower carbon emissions, better traffic flow, lower car ownership costs, lower insurance costs, and improved fuel efficiency and mobility.

However, with real-world trials on today’s roads, the capabilities that autonomous vehicles must support continue to expand and rapidly increase in sophistication. These automated systems will have ever-increasing demands on performance, power consumption, safety, security and reliability. For automotive OEMs, ensuring that autonomous vehicles comply with safety regulations requires designing hardware and software according to the ISO2626 functional safety standard. If developers are ill-prepared for this, additional capital and time will be required to demonstrate that their products meet safety standards, potentially significantly delaying time to market, compressing profitability and eroding market share.

The core goal of autonomous vehicle safety and reliability is to prevent personal injury and property damage. When did the accident happen and who should be held responsible for the accident are all legal issues that need to be considered. Under such traffic conditions, autonomous driving faces many legal issues, and how to determine the attribution of responsibility in the event of an accident remains unresolved. Therefore, failures must be avoided. This makes automotive OEMs and automotive market suppliers more focused on reliability. Therefore, it is crucial to prove that every component in a smart car is safe and reliable.

Smarter, more reliable storage

Vehicles with autonomous driving technology are equipped with high-level advanced driver assistance system (ADAS) functions. These vehicles have multiple sensors (cameras, lidars, etc.) and controls to enable autonomous driving and avoid collisions. These sensors and controls are mission critical and cannot fail. Figure 1 shows the schematic diagram of a fully automated driving system with levels 3, 4, and 5 automation that enables unsupervised driving.

Technical Article – Smart Storage Improves Reliability of Self-Driving Vehicles

Figure 1: Autonomous Driving System: Level 3/Level 4/Level 5 System Schematic

Nonvolatile memory devices play an important role in ADAS systems, providing boot code storage and data logging for important mission-critical events. As these systems become more intelligent, they need to process more data faster and with higher levels of reliability. In addition, even if an ADAS design is otherwise reliable, if the memory is not protected (that is, it is not verified that the memory bits have changed at startup or while the device is running), it is easy to create a weak link.

Because NOR flash provides non-volatile storage supported by high reliability and integrated diagnostics, it is an ideal memory technology for mission-critical applications. Integrated diagnostics ensure data integrity, detect possible failures, and even correct errors. In addition, benefits such as instant-on capability and high-performance fast system startup times facilitate immediate access to code, configuration data and graphics images when the car is powered up.

Today, to meet automotive functional safety standards such as ISO26262, memory device families need to be designed from the ground up. These next-generation memories not only provide higher reliability, but also improve performance, dramatically reduce power consumption, and reduce total cost of ownership.


One of the most effective ways to simplify a system is integration. When a system consists of numerous components, each component and its interconnection with other components can be a potential point of failure. For example, integrating an MCU with memory provides faster data and code access, higher processing efficiency, greater reliability, and lower cost. In addition, development is simplified as components that previously had to be integrated into larger systems by developers can now be managed internally by the MCU.

The advantages of integration now extend to NOR flash as well. As memory manufacturers begin to integrate memory with processors such as the Arm Cortex-M0, complex processing is required to maintain the reliability of high-density, high-speed memory (see Figure 2). The advent of on-board processors enables smarter storage, revolutionizing the way engineers design with flash memory. For example, in the past, increasing the lifespan of flash memory required extensive development of wear-leveling software. Now, wear leveling is managed internally by the integrated MCU.

Figure 2: Integrated Arm Cortex M0 in Smart Flash Storage

The new generation of complex SoCs produced using 16nm FinFET technology cannot yet embed flash memory into the die. So they must rely on smarter and more reliable external NOR flash technology. The onboard processor can be used not only to manage all safety-critical areas of memory storage, but also to manage the network security area of ​​memory to guard against malicious attacks. By incorporating the integrated processor into the flash memory, these units are self-managed by the memory device and can be quickly configured to meet the specific requirements of the application.

changing requirements

Currently, the automotive industry is moving from driver assistance to full automation. These systems will require intelligence at every level to reduce latency and increase efficiency. At the same time, the internal architecture of the car is also evolving from discrete systems that operate primarily independently to interconnected systems. Connected systems can transfer data between systems in real-time and leverage artificial intelligence and machine learning. In addition, the data collected from the car will be used to implement predictive maintenance, so that the car can prompt the driver to maintain the vehicle before it breaks down. Data also needs to be sent to the cloud in order to perform more complex analysis and complete new software upgrades from the cloud to the car.

Smart flash storage is at the heart of these systems, as critical code and data stored in these non-volatile memories still need to be reliable and last over 20 years without failure in extreme environments. With the addition of an onboard processor, these memories can now provide a higher level of functionality and reliability, while offloading memory management tasks such as wear leveling, enhancing system security with cryptographic protection and performing safety-critical diagnostics.

Autonomous driving is a fast-moving industry, and new safety and security features will be developed and compliant at the same pace. OEMs need a flexible architecture to adapt to these standards in a timely manner and introduce advanced features that enhance long-term reliability. For example, when the memory can predict certain types of failures, it can begin to prioritize.

To help automotive OEMs build compliant systems, memory manufacturers need to provide ISO 26262-compliant safety documentation, including detailed safety analysis reports such as safety manuals, failure mode effects and diagnostic analysis (FMEDA), dependent failure analysis ( DFA) and Context Independent Security Elements (SEooC). In addition, memory manufacturers need to actively develop and comply with these standards to ensure that the components they produce continue to comply with regulatory requirements.

The Links:   G150XVN010 CM1400DU-24NF