We had a conversation with the Vice Chair of the MIPI A-PHY Working Group, Ariel Lasry. Ariel spoke about his presentation for AutoSens Brussels on 13th September 2022, “An Integrated Approach to Creating Functionally Safe Automotive Sensor Systems.”
In addition to his role as director, technical standards, at Qualcomm, Ariel Lasry serves as vice chair of the MIPI A-PHY Working Group. As such, Ariel has a unique vantage point on the history and future of A-PHY—the industry’s first standardized asymmetric SerDes specification —as well as the entire, end-to-end MIPI Automotive SerDes Solutions (MASS) framework for connecting cameras, sensors and displays with built-in functional safety, security and data protection.
We were so glad you joined us at AutoSens in Brussels last year! Was this your first AutoSens, and were you looking forward to the conference?
Yes, it was my first time at AutoSens, and I definitely looked forward to interacting with and learning from the ADAS and autonomous-vehicle specialists from around the world who are shaping the future of vehicle perception.
You’re the vice chair of the MIPI A-PHY Working Group and a winner of a MIPI Lifetime Achievement Award in 2019. Can you tell us a little bit about your experience, and what you do in your role with the MIPI Alliance?
Over the years, I’ve been involved in strategic marketing, product management and business development for not only the automotive sector but also other verticals such as mobile and industrial automation, and I’ve been active in standardization activities in several organizations for more than a decade. My engagement with the MIPI Alliance began nearly 15 years ago, and during that time I’ve contributed to multiple working groups and also represented my previous company on the MIPI Alliance Board of Directors for 10 years. Since becoming vice chair of the MIPI A-PHY Working Group, I’ve been busy as an author and speaker at industry events like this one because of the intense global interest in what the MIPI Alliance is doing with and for the automotive industry.
Can you give us some insight to your presentation at AutoSens, “An Integrated Approach to Creating Functionally Safe Automotive Sensor Systems”?
As more sensor-based safety critical systems are added to vehicles, the benefits of a standards-based approach for sensor connectivity also increase—and those benefits are even more profound if functional safety requirements are already integrated into that solution. I spoke about how MIPI’s MASS framework provides just such a standardized sensor-to-ECU solution for autonomous driving and ADAS systems.
My presentation resonated with others in the AutoSens community who are working in this space and it demonstrated how MIPI is addressing some of the most important challenges in this area of automotive design. The presentation described the MASS framework, specifically focusing on the functional safety features that have been embedded throughout the stack—from the MIPI CSE (Camera Service Extensions) layer, to the baseline MIPI CSI-2 (Camera Serial Interface 2) image sensor protocol, and to MIPI A-PHY, which serves as the foundation of the framework.
How does the MASS framework simplify integration?
Prior to MASS, automotive designers had to rely on proprietary SerDes solutions for long-reach links between peripheral components and processors, and create their own proprietary functional safety solutions, too. This lack of standardization and supporting connectivity framework limits supplier choice, prohibits economies of scale, and increases design complexity and cost.
MASS, however, leverages existing MIPI protocols that have already been widely implemented in automotive, together with new service extensions and adaptation layers to create an end-to-end connectivity solution. In these ways, MASS creates economies of scale around a set of industry-supported standards, reduces integration costs, and enables OEMs and Tier 1 suppliers to amortize engineering costs over larger volumes of components. It also promotes the development of enhanced support services such as test and software resources from a wide ecosystem of industry contributors.
And while MASS is designed for immediate implementation, it’s important to understand that MASS implementations will become even more streamlined over time as A-PHY and other MASS protocols are directly integrated into sensors, ECUs and displays, thereby eliminating the need for separate bridge components.
Functional Safety (FuSa) is imperative. Can you tell us more about these features in your technology?
MIPI evaluated the requirements defined in ISO standard 26262 and embedded corresponding FuSa-enabling features into both A-PHY and the upper layer camera and display protocols, to help designers build solutions that meet requirements from ASIL B through ASIL D.
For example, to overcome some of the challenges presented by the need to integrate more and more advanced cameras and other sensors into vehicles, the MIPI Camera Service Extensions specification enhances the MIPI CSI-2 image sensor interface with end-to-end FuSa, as well as other features for automotive applications. Image data transferred from sensors to ECUs, and the data that control those sensors, can be monitored at the protocol level to ensure any errors and failures are detected at the application layer. To achieve FuSa for this image data transmission, the CSE protocol utilizes three main features:
- a cyclic redundancy check (CRC-32) with a Hamming distance ≥ 3 to detect data transmission errors, ensuring the image data captured by the sensor is accurate when received at the ECU
- a frame counter to detect frame loss or duplication, with its accuracy verified by the CRC, ensuring the continuity of video streams from cameras to ECU
- timeout monitoring to detect any loss of data caused by a stopped or stuck transmission between an image sensor and ECU
In my presentation at AutoSens, I covered these features and others in more detail and how the different MASS components work together to contribute to the automotive industry’s FuSa goals.
What does the future look like for MIPI A-PHY?
As we did with the first version of A-PHY, we are moving forward with the process to have A-PHY v1.1 adopted as an IEEE standard. Meanwhile, the A-PHY Working Group has been proactive about understanding the key automotive industry requirements for the next major version of the interface. Our goal is to deliver a specification update that is focused on emerging architecture and use cases, and is poised to provide OEMs and Tier 1s with a clear path forward for product development based on A-PHY.
One key area is making sure A-PHY is set to properly serve the zonal architectures and SDV (software-defined vehicle) trends that are emerging in the industry and to better support the modern automotive cockpit environment. At the same time, MIPI has a heritage of new PHYs supporting backward compatibility and allowing for easy migration, and A-PHY is no different. Version 2.0 will continue to be compatible with the previous versions of the PHY. This is crucial to efficient system design, especially for evolving products projected for long life. For example, we are also committed to not making any changes in the upper layers of A-PHY to ensure easy migration and minimal impact at the system level. And, of course, we remain committed to maintaining A-PHY’s high immunity to EMI effects and low packet error rate.
New features are in development for the next version of A-PHY, following the same kind of intense technical discussions and feasibility studies that precipitated the rollout of A-PHY v1.0 and v1.1. For A-PHY v2.0, we are doubling the downlink bandwidth to up to 32 Gbps per lane, boosting the uplink bandwidth to 1.666 Gbps and expanding A-PHY secure control.