Coupling & Crosstalk: A Trillion Sensors?

Coupling & Crosstalk is my column in the MEPTEC Report. This column appears in the Winter 2014 edition on pages 10-11.

Electronic coupling is the transfer of energy from one circuit or medium to another. Sometimes it is intentional and sometimes not (crosstalk). I hope that this column by mixing technology and general observations is thought provoking and “couples” with your thinking. Most of the time I will stick to technology but occasional crosstalk diversions may deliver a message closer to home.

Name Calling

Whatever you desire “there’s an app for that!” Dream big or small, it is very likely a software program is already available. But what if you dream in hardware?  Hewlett-Packard is building applications with one MILLION sensors. Robert Bosch is dreaming of 1,000 sensors per person, i.e. seven TRILLION. Janusz Bryzek is aiming for one TRILLION per year.

Standardized smartphone hardware platforms and application “stores” have significantly lowered the cost and time to develop and sell applications. One might argue the barrier to entry is too low – some applications are developed in a weekend for pennies and it shows… What works well for software, where larger investments can be delayed until market interest is confirmed, is simply not practical for most hardware applications. Hardware has high development costs especially if they are micro-electromechanical systems (MEMS) based products. MEMS devices on average cost as much as $30-45 M and can take 25-30 years for full commercialization. For hardware applications it is essential to understand the product and market requirements (including likely customer demand) beforehand.

The inaugural TSensors Summit recently held at Stanford University was focused on how to identify the markets and requirements for high-volumes of sensor based hardware. Most, if not all, of these hardware sensor solutions would be MEMS based to provide high functionality at low costs in extremely high volumes. The organizing committee, on which I serve, has the goal to build a process to generate an industry roadmap to achieve a trillion (“T”) sensors. As the driving force behind TSensors, Janusz Bryzek (Vice President Development, MEMS and Sensor Solutions of Fairchild Semiconductor and Chairman of the event) wants to achieve large volumes not for their own sake but for “abundance”.

In Abundance; The Future Is Better Than You Think (2012), Peter Diamandis and Steve Kotler describe solving global problems using technology to achieve balance between supply and demand. This abundance – providing of “a world of nine billion people with clean water, nutritious food, affordable housing, personalized education, top-tier medical care, and non-polluting, ubiquitous energy” – is what Bryzek wants to achieve. And Diamandis and Kotler know they need high technology including sensors, possibly needing as many as forty-five trillion sensors in the next twenty years. (Well in excess of 1 T per year at the high end of the estimate…)

The end markets are clear: medicine, food production, energy, and water. Therefore, what is needed is to define the specific applications in order to focus companies on building the necessary software and hardware. Yes, there will be software only solutions that will make a difference but the biggest impact will come from hardware-centric or hardware-enabled (combined with software) systems. The challenge is how to develop this hardware as quickly as possible and build at scale while providing the proper return on investment. The race is on to have these solutions before the world adds another billion or two people…

Clearly the current MEMS development paradigm is too long at 25-30 years and too expensive at $25-30M per MEMS device and even more for the entire application system. Therefore, the industry needs to accelerate this work through “coopetition” (cooperation with competition) including the generation of roadmaps. Roadmaps help an entire industry identify needs and timing to ensure a healthy ecosystem of suppliers and customers. As an example, the semiconductor industry has flourished through the annual publication of International Technology Roadmap for Semiconductors (ITRS) since 1998 identifying the needs and challenges of the next fifteen years.

The goal of the TSensors community is to build a similar roadmap for sensors. A comprehensive roadmap would identify the technology required for the desired functionality and timing. This would enable suppliers and customers to align their schedules and expectations so neither would get too far ahead of the other. A mismatch is not good – having a technology that no one is ready to buy and not having the technology available when customers need it are equally problematic.

Over three days, the TSensors Summit had forty-six excellent presentations from leading visionaries of the sensor world covering a very wide array of markets and applications. Challenges discussed included everything from basic technology to business issues to customer education to regulatory approval.

What quickly became apparent is the need to converge solutions by creating general-purpose sensor platforms. Steve Nasiri (Founder of InvenSense, now running Nasiri Ventures) described how InvenSense disrupted the market by shifting from a single-axis inertial measurement device (gyroscope or accelerometer) to a motion-sensing platform. Previously to build a product solution required multiple single-axis devices (one for each direction) that had to be properly mounted and calibrated into the end product (video game system, smartphone, etc.).  The end product developer also needed to write the software to read the raw data from each sensor and interpret the results. InvenSense moved from single-axis MEMS sensors to providing 6-axis (and greater) MEMS devices along with the software required to provide meaningful motion data. This greatly reduced the development efforts and manufacturing costs while increasing the functionality and value of the InvenSense solution. The shift to supplying platforms that are general purpose moved InvenSense from being a component supplier to a solution provider.

The need for sensor platforms can also be clearly illustrated by examining the area of chemical sensing. If we presume there are two hundred different chemical sensing applications in the next ten years all with the potential for high volume, which scenario is easier: designing and building two hundred different sensor technologies or building a generic integrated sensor platform that can handle many or all of them? There are several companies, including Hewlett-Packard, currently working on MEMS based Raman spectrometers to cost effectively identify a large number of chemicals. The ultimate goal is to make these MEMS sensors inexpensive enough that they become disposable.

Volunteers are working to identify the number of distinct platforms (and requirements) to service all the different sensing needs discussed at TSensors and those already known to them. For each type of platform identified a working group will be formed. The working groups will identify the technology requirements along with the development work and ecosystem support required to rapidly commercialize the platform. Estimation of the current progress towards commercialization and gaps will be made for each platform type.

It is true that some platforms may initially be more expensive as “general purpose” versus application specific. However, it is hoped that platforms shorten the time to market for a wider array of products with significantly lower development cost across all applications. The platforms should also have lower overall costs due to economies of scale and competition between different suppliers building platforms for the same markets. And as platforms provide higher value solutions versus simply supplying components they should also speed development while lowering product costs.

Now is a great time to get involved in this roadmap effort from providing input, writing requirements, or determining your company’s market strategy! We welcome participation at one of the six or so events in 2014 including additional Summits in Japan, China, and Germany. And we look forward to adding working group members to drive the platform definitions. Our goal is to complete a first revision of the roadmap before the next United States TSensors Summit in October 2014. Please see http://www.tsensorssummit.org/ to engage in the process and for additional details.

 

As always, I look forward to hearing your comments directly. Please don’t hesitate to contact me to discuss your thoughts.