San Diego’s Coronado Island Marriott Resort and Spa hosted SEMI’s MEMS and Sensors Executive Congress (MSEC 2019) from Oct 23 -25, 2019. While it’s “Resort and Spa” setting helped to de-stress the many busy executives attending, the excellent facilities and a very relevant agenda encouraged everybody to work hard and make this congress a very interactive event, focused on exchanging information and ideas about how to utilize the many strengths of microelectromechanical systems (MEMS) and sensor to enhance the capabilities of electronic systems. Artificial intelligence (AI) can only provide its highly anticipated value if it gets vast amounts of data from these – until recently – totally underestimated “eyes and ears” that capture real word events and help systems – like trucks, cars, drones, and robots – to function properly and safely in their environment.
After Dave Anderson, President, SEMI Americas and John Chong, Chairman of the MEMS & Sensors Industry Group (MSIG) gave brief introductory remarks, the keynotes session started.
The first keynoter was Stefan Finkbeiner, CEO, and GM of Bosch Sensortec. He gave a perfect example of the synergy between AI and MEMS/Sensors. Finkbeiner outlined how his broad product portfolio has evolved from simply reading data in 2005 to using customizable software for evaluating it by 2012, to applying machine learning to it today. Bosch offers MEMS, sensors and customizable software as system-level building blocks, packed in single-chip solutions. By moving AI to the edge, Bosch simplifies personalization, enhances privacy, minimizes system reaction times and power needed for data transmission as well as off-loads network and cloud.
Tingfang Ji, Senior Engineering Director at Qualcomm’s Wireless R&D Center, described Qualcomm’s focus on 5G technology chips for handsets and base stations. He stated that over thirty 5G networks are deployed today, a few of them already use mm-wave technology. They serve a wide range of industrial applications and make transportation, medical, agriculture, smart cities as well as other segments more efficient. Ji highlighted that 5G offers bandwidth and latency improvements, as well as better security. Knowing that RF MEMS are very important as high-quality filters for 5G phones, I was hoping that Ji would give them the credit due, but either I didn’t notice when he talked about them or RF filters are not part of Ji’s responsibilities.
The third keynoter, Rajendra Rao, GM, IBM Food Trust, started his message with very concerning data points about today’s food supply chain (Figure 2). His team is working on replacing today’s error-prone and vulnerable, hand-written documentation methods with blockchain technology to track food shipments from the source to their final destination automatically and monitor with integrated sensors if/when the goods were exposed to detrimental conditions, e.g. disruption of the cooling chain.
IBM engages all parties involved in the production, storage and distribution of food, shows them how to utilize modern technologies – including sensors – to not only fulfill the purpose of a safer supply chain but also to increase profits.
Google’s Nick Kreeger, Senior Software Engineer, Google Brain, rounded out the keynote session on Wednesday morning. He proudly conveyed that two billion mobile devices use Google’s TensorFlow technology today. Android phones even utilize Machine Learning (ML) to optimize battery management and voice recognition.
Many questions from the very engaged audience in the following panel session demonstrated that these four keynotes did not only convey where and how MEMS and sensors impact system capabilities and our lives, but also gave the audience, comprising mostly MEMS and sensors designers and manufacturers, lots of ideas how to enhance the market value of their products, how to best position them as essential system building blocks in the eyes of system architects and which additional markets to target.
MSEC 2019 Technology Showcase
Stephen Whalley, Managing Director, Microtech Ventures, introduced the Technology Showcase, which was sponsored by Jessica Gomez, Founder, and CEO, Rogue Valley Microdevices. Like in previous years, this session gave several entrepreneurs the opportunity to very briefly introduce their companies and their ideas to serve specific applications with MEMS/sensor-based solutions.
Last year’s winner, Marc Rippen, Founder and CEO of Alertgy, experienced another year of running his medical device startup and had a few words of wisdom for this year’s contestants and the audience: 1) VCs don’t appreciate hardware companies. They prefer software, because these companies need less capital and pay back the investors faster. 2) Never give up – failure is not an option.
The first contestant, Jeremy Holleman, CTO of Syntiant, presented Syntiant’s Neural Decision Processor and how it can move AI to IoT edge nodes, lowering their power dissipation and dependence on networks and the cloud.
Marc Dupaquier, Co-Founder of Cartesiam, explained how the company’s “NanoEdge AI” solution minimizes data transfer to the cloud and allows customers to modify Cartesiam’s software for their specific applications.
Ross Bundy, CoFounder of Cardea, described how graphene field-effect biosensors can be used to monitor biochemical functions and bring the bio world closer to the digital world.
Joshua Windmiller, CTO at Biolinq, showed how a sensor front-end for medical applications was built using MEMS-based skin patches, e.g. for needle-free diabetes testing,
Benjamin Hansen, VP of Biomechanics & Innovation at Motus Global, explained how the company’s MEMS-based inertial measurement units (IMUs) are being used to measure workloads for limbs, e.g. on the throwing arm of a baseball pitcher.
Paul Banks, CEO of TetraVue emphasized the importance of 3D cameras for autonomous driving, augmented reality glasses, cell phones, and many other applications. TetraVue’s optical time-of-flight technology enables the capture of high -resolution 3D pictures with 2D image sensors.
The audience enjoyed these very creative and diverse examples for utilizing MEMS and sensors and chose Marc Dupaquier as this year’s Technology Showcase winner.
MSEC Highlights Through the EDA Lens
Many other presenters showed their companies’ commitment to improving MEMS and sensors by adding AI to provide higher-value solutions. They explained how to save power and overcome security concerns as well as network and cloud capacity limits. Allow me to only highlight one of these presentations, somewhat subjectively.
Having spent most of the 1970s as an analog/mixed-signal board and system design engineer, I really liked the presentation from Tom Doyle, CEO, and Founder of Aspinity. Like other presenters, he recognizes the need to move intelligence to the edge nodes, but unlike all others, he developed a very low power, analog-based solution for always-on listening devices – like Alexa from Amazon. The core of this solution is a trainable, reconfigurable analog modular processor (RAMP). It suppresses noise, listens for the “wake-up word” and compresses data by magnitudes (Figure 3).
In the closing remarks, Carmelo Sansone, Director SEMI-MSIG, invited every attendee to join the MEMS and sensors industry group (MSIG), a SEMI strategic association partner, to contribute to and benefit from joint efforts in this high-potential industry segment.
Conclusion: Where is Design Support?
As mentioned earlier, MSEC 2019 clearly conveyed that making MEMS and sensors ubiquitous is essential for fully utilizing the power of AI and making our lives easier. The advantages of moving AI to the edge nodes, right next to the “eyes and ears” of systems, makes sense for most applications and is in full swing. Companies either use single-die (monolithic) ICs or multiple ICs to accomplish these integrations – no interposer-based 2.5D-ICs solution has been presented at MSEC – yet.
However, the entire event lacked one key element: Where are the providers of design tools, the EDA vendors? They need to complement all these ideas and existing capabilities with multi-physics simulators, to streamline edge-node design, minimize unit cost and power dissipation. Then they can also benefit from this high growth area.
I was very glad that Ian Campbell, CEO of OnScale and Mary Ann Maher, founder of SoftMEMS, spent some time with me in San Diego to explain their team’s current multi-physics design, modeling and simulation capabilities. Even better, the day after MSEC, Mentor, a Siemens Business, announced a collaboration agreement with OnScale and SoftMEMS to develop an integrated workflow for digital prototyping of edge nodes.
Finally, the ecosystem for MEMS and sensors is ready to deploy billions of intelligent, ultra-low power and cost-effective edge nodes worldwide. It’s the right time to join these efforts – you don’t want to miss this opportunity!
Thanks for reading, and see you at next week’s Microtech Innovation Summit … Herb