The reality of autonomous vehicles will dramatically change our travel habits and a cause a rapid shift in the transportation industry. Digital transformation of the automotive industry will deliver many social benefits such as reducing the number of accidents, lowering carbon emissions, improving traffic flow, reducing the cost of car ownership, lower insurance policy premiums, increasing fuel efficiency and improving mobility.
However, the complexity of an autonomous vehicle is increasingly rapidly as the capabilities they must support continues to expand with real-life trials on today’s roads. These autonomous systems will require increasing performance, power, safety, security and reliability. For automotive OEMs, the task of ensuring safety compliance for autonomous vehicles will require both hardware and software designed to ISO2626 functional safety standards. If developers are not prepared, the added investment and time required to prove out safety standards could significantly delay time to market, reduce profitability, and erode market share.
At the core of safety and reliability for autonomous vehicles is the goal of preventing injury to people and property. There are also legal issues to consider when an incident occurs and who will assume liability. With the legal issues surrounding autonomous driving in such a state of flux, it is still unclear where liability will fall in the case of accidents. Thus, failure is not an option, making it even more critical for automotive OEMs and suppliers to the automotive market to focus on reliability. As a result, every component in a smart car must be proven to be safe, secure and reliable.
Smarter, More Reliable Storage
Vehicles with automated driving technology are equipped with a high-level of Advanced Driver Assistance Systems (ADAS) capability. These vehicles have multiple sensors (camera, LIDAR, etc.) and control devices that can perform automated driving with zero collisions, which are mission-critical and cannot fail. Figure-1 shows the block diagram for a fully autonomous driving system capable of levels 3, 4, and 5 autonomy necessary for non-monitored driving.
click for larger image
Figure-1: Autonomous Driving System: Level 3 / Level 4 / Level 5 System Block Diagram. (Source: Cypress)
Non-volatile memory devices play a key role in ADAS systems to provide boot code storage and data logging of key mission critical events. As these systems become smarter, they need to process more data faster with a higher level of reliability. In addition, unprotected memory – in terms of verifying that memory bits have not changed at boot time or during device operation – can easily become a weak link in an otherwise robust ADAS design.
NOR Flash is an ideal memory technology for mission-critical applications as it provides non-volatile storage backed by high reliability and integrated diagnostics that can ensure data integrity, detect possible failures, and even correct errors. Additionally, the benefits of instant-on functionality and fast system boot time with high performance enables immediate access to code, configuration data, and graphic images when the car is powering up.
Today, memory device families are architected from the ground up to meet automotive functional safety standards such as ISO26262. These next-generation memories not only provide greater reliability, they increase performance, substantially reduce power consumption, and lower total cost of ownership.
One of the most effective ways to reduce system complexity is through integration. When a system comprises many components, each component – as well as each interconnect it has to other components – represents a potential point of failure. When MCUs integrated memory, for example, this resulted in faster access to data and code, more efficient processing, greater reliability, and lower cost. It also simplified development since components that developers previously had to integrate into the greater system are now managed internally by the MCU itself.
These same benefits of integration are now being brought to NOR Flash as memory manufacturers begin to integrate memory with a processor like the Arm Cortex-M0 to handle many of the complexities required to maintain reliability with dense, high-speed memories (see Figure 2). The availability of an onboard processor enables smarter memory and completely changes how engineers design with Flash. For example, in the past, significant development was required to implement wear-leveling software to extend the endurance of Flash. Wear-leveling is now handled internally by the integrated MCU.
click for larger image
Figure-2: Integrated Arm Cortex M0 inside a Smart Flash Storage (Source: Cypress)
Next-generation complex SoCs fabricated with 16nm FinFET technology don’t yet have the option of embedding Flash memory inside the die. Thus, they have to rely on external NOR Flash technology that are smarter and more reliable. The onboard processor can be used to manage all the safety-critical aspects of memory storage. It can also be used to manage the security aspects of memory storage to provide protection against malicious attacks. With an integrated processor inside the flash memory, these units are managed by the memory device itself and can be quickly configured to meet the specific requirements of the application.
The automotive industry is currently advancing from driver assistance to full automation, and these systems will require intelligence implemented at all levels to reduce latency and improve efficiency. At the same time, the internal architecture of the car is moving away from discrete systems that act primarily independently to connected systems that stream real-time data to each other, utilizing artificial intelligence and machine learning. The data collected inside a vehicle will also be used to implement predictive maintenance so that cars can alert drivers to service vehicles before they break down. Data will also need to be sent to the cloud for more complex analytics and new software upgrades from the cloud to vehicle.
Smart Flash storage will be at the heart of these systems because the critical code and data stored in these non-volatile memories needs to be reliable in extreme environments and last for over 20 years without failing. With the addition of an onboard processor, these memories can now provide an even higher level of functionality and reliability while offloading memory management tasks like wear-leveling, enhancing system security with cryptographic protection, and performing safety-critical diagnostics.
Autonomous driving is a fast-changing industry, and new safety, security features will be developed – and mandated – just as quickly. OEMs need a flexible architecture that enables them to adapt to these standards in a timely fashion as well as adopt advanced functionality that improves long-term reliability. For example, when the memory can predict a particular type of failure, it can begin to preempt it.
To aid automotive OEMs in building compliant systems, memory manufacturers need to provide safety documentation that are compliant to ISO 26262 specification, including detailed safety analysis reports such as Safety Manual, Failure Mode Effects and Diagnostic Analysis (FMEDA), Dependent Failure Analysis (DFA) and Safety Element out of Context (SEooC). Memory manufacturers also need to play an active role in defining and maintaining these standards to ensure their components continue to be compliant.
Memory devices like Cypress’s Semper NOR Flash are purpose-built to address the challenges for next-generation automotive and industrial system that meet the quality, reliability and safety standards.
Sandeep Krishnegowda is marketing and applications director of the Flash Business Unit at Cypress Semiconductor Corp. He has worked in Cypress’ memory products division for more than ten years in a variety of engineering, management and marketing roles. He earned an MS in Electronics and Communication from Rensselaer Polytechnic Institute and a BE in Electronics and Communication from Visvesvaraya Technological University