Electronic Design For Reliable Autonomous Driving

Current design processes are not keeping up with the ambitious demands of the automotive industry.


In the area of advanced driver assistance systems, most car makers and their suppliers have laid out exciting road maps all the way to highly automated and fully automated driving in 5 to 10 years. But are the electronics keeping up with these ambitious plans? At least for the automotive industry as a mass market, the current design processes for microchips and systems are not yet ready.

An autonomous vehicle is a high-performance system that also requires correspondingly high-performance electronics. To evaluate the extensive data on the vehicle’s surroundings and execute the computational steps required by a self-driving car, today’s prototype vehicles are chock-full of sophisticated computer technology. To turn these prototypes into mass products, this technology must take up significantly less volume while still satisfying the high demands of the automotive industry on lifespan, reliability and functional safety. Today’s most advanced, ultra-scaled semiconductor technologies are ideal for solving the volume problem. Their miniaturized designs perfectly support the integration of diverse components in minimal space.

In terms of reliability in the vehicle, however, they fall short. The reason is that they have not been developed for the harsh environmental conditions and loads encountered by cars. They are at home in the relative comfort of consumer electronics, and have not yet been adapted for the extreme demands awaiting them on the roads of the world. For example, electronic components in vehicles must withstand wide temperature ranges and guarantee correct operation even at extremes of up to 160 degrees Celsius – and over a lifespan of at least 15 years. The same is true of factors such as moisture, vibration, electromagnetic interference and more. Additional methodologies to support the design, qualification, manufacturing and testing of semiconductors are therefore required.

Aging of the integrated components themselves as a result of intrinsic degradation mechanisms also plays a major role under automotive conditions in the technologies in question. The increased susceptibility of the components arises from the use of innovative materials, such as high-k dielectrics for the gate oxide and ultra-low-k materials for the insulation of the metal layers, as well as from new process steps in manufacturing. In this way, known aging effects are accelerated, new effects are produced, and dependencies arise between the various influencing factors.

For the safe and reliable operation of electronic systems for many years in a vehicle, it is therefore important to identify potential failure mechanisms at the physical level already during the design process, and to understand their impact on individual circuits and the overall system. Corresponding guidelines must be continually improved in consideration of the complex degradation mechanisms in order to qualify technologies to meet automotive requirements, because the degradation models included in today’s industry standards are not always sufficient for new technologies. At the same time, the aging effects relevant to advanced circuits must be described such that they can be taken into account during the design process. New approaches and tools are absolutely required to better support the designers of vehicle electronics in estimating the system reliability of a given technology.

Moreover, it is very important to include real-world conditions in the reliability assessment. For example, even though temperatures of up to 160 degrees Celsius are specified for automotive applications, such loads over thousands of hours of operation are exceptional cases. More typical are temperature profiles specified by car makers that define a distribution of temperature ranges for various installation locations. Such profiles are also useful with regard to electrical and mechanical stress because designing circuits for sustained maximum stress would lead to significant overdesign. If suitable lifetime models are available, the impact of various operating ranges on the overall reliability can be calculated. This also makes it possible to investigate the criticality of brief violations of the specified limits.

Designers of vehicle electronics are facing new requirements in the context of functional safety, as well. As the level of autonomy increases, so does the amount of safety-critical electronics. ISO 26262:2018, the new version of the standard, will contain a separate section on semiconductors in recognition of their importance for reliable and safe operation. Additional design support will also be required in this area. Integrated tools for analysis, verification and thorough documentation of functional safety are under development. Alongside tools for verification and fault simulation of mixed-signal circuits, there is a need for reliable aging models for the estimation of failure rates. Before such tools are employed, however, a consistent methodology must be developed and interfaces to partners must be defined.

The requirements on the electronics for use in an autonomous vehicle, for example, are therefore quite diverse. Lifetime predictions supported by simulations will also become indispensable in future development processes. This requires the development of state-of-the-art models based on current insights into key degradation effects and their parameter dependencies. If designers get access to these models as part of the process design kit (PDK), the knowledge obtained about such a technology can be integrated into the design, and electronic components can be optimally designed to meet even extremely demanding requirements. Only by combining better information and data exchange with new and powerful predictive models will it be possible to more effectively design electronics within the context of their real-world applications and to produce complete systems that remain reliable in the face of harsh operating requirements – even in the case of fully automated vehicles.