comment on this article

Sharing basic algorithms will save innovators from 'reinventing the wheel'

It's been a busy few months for Steve Whalley. Having retired from Intel after 28 years, where his last post was director of sensing, Whalley is now chief strategy officer for the MEMS Industry Group (MIG), where he is leading efforts to prepare the MEMS and sensors ecosystem for what is anticipated to be a decade of explosive growth.

One of the first moves Whalley has made in this respect is to establish the Accelerated Innovation Community, or AIC. This open source algorithm cooperative is intended to help smaller companies to reduce time to market, start up costs, risk and barriers to entry by encouraging collaboration across the MEMS/sensors supply chain.

"The idea started from a discussion with Ian Chen from Freescale. I was finding out from him what he saw as the issues for the sensors industry and what MIG could do to bring people together.
"He raised the point that many start ups were reinventing the wheel. If they wanted, for example, to do three axis sensor fusion, they had to develop an algorithm to handle the data. It was the same if a company wanted to develop a pedometer and so on (see fig 1). That seemed to be stifling innovation."



So Whalley set off to investigate whether it would be possible to create an open source community in which low level software could be made available. "Companies may pay for complex software," Whalley noted, "but they won't pay for the low level stuff."

The result was the AIC, with Freescale being the first company to 'sign up'. "It is getting the ball rolling by contributing some algorithms and collateral," Whalley continued.

Ian Chen, marketing manager with Freescale's Sensor Solutions division, explained the move. "It's very simple," he said. "Sensor adoption is a key driver of the Internet of Tomorrow (IoT). This is bringing a lot of different applications; ones that people hadn't previously thought of or ones putting a different twist on conventional activities.

"Both require a different kind of intelligence in the sensor, particularly if you're talking about monetising the innovation. You have to address things like accuracy and security. It all boils down to having good algorithms."

Since the launch of AIC, Freescale has been joined by other sensor vendors and interested parties. "There has been a positive response," Whalley said. "Companies have been asking 'why haven't we done this before? Let's get on with it'. The feedback is that most major players will get involved and will contribute."

Chen said AIC was something Freescale knew it had to do. "The MIG's charter calls for it to facilitate everything around MEMS and sensors. It's a body around which we can all rally. But I don't think it will have the monopoly; while it will start things off, I'm hoping to see more open innovation around sensors."

Alongside semiconductor companies, Whalley says interest is being shown by universities such as Berkeley and Carnegie Mellon. "And some pure research institutes are interested."

One of the reasons why MIG and its collaborators are keen to develop AIC is that they believe it will enable creativity. "AIC could help a start up to build a prototype," Whalley suggested. "It could then put that prototype in front of an investor, get a bit of money to help it make a more robust device and to license more software. If we only do that, we will have been helpful."

Whalley also believes that AIC will get start ups focused on their true idea. "And that isn't sensor fusion," he asserted. "It may be a health device, such as something that monitors chronic obstructive pulmonary disease. It might monitor how many steps the users takes and how that affects their breathing. But the developer should be focusing on that and not on software."

It appears that AIC is a timely development. Whalley noted that, with the advent of the smartphone, more sensors are available at lower cost. "They are now cheap enough for companies to use in their apps, but they need algorithms to bring those sensors to life. There will be the need for a lot of sensors over the next decade and anything we can do to break through some of the challenges will help the whole industry."

In his opinion, sensors are being applied across the board. "All segments," he said. "IoT, wearables and health care – even agriculture – are all starting to use sensors. Big challenges for the future, such as clean water, transportation and energy will all need new types of sensor."

Chen agreed. "We see the need to fundamentally change the cost of developing algorithms. We want to provide basic algorithms with good support to those entrepreneurs developing IoT applications, but we also want to create a forum in which people can engage in open innovation."

One other area where it is believed that AIC will make a significant difference is in data standardisation. Whalley explained: "One thing MIG would like to do is to collect data sets; these are important to determine a profile of users. If we can put basic algorithms out there and get them used, we can construct data sets."

Chen fleshed out this concept. "Say a start up is looking to develop an app that monitors whether a geriatric patient has fallen. For less than 1% of the time, they won't have, but if you want to create a robust algorithm, you have to spend time collecting data when they're in bed or sitting in a chair and so on.

"Collecting all that information would be prohibitively expensive, but it is needed in order to create differentiated IP. There is a huge opportunity for competitors to come together to lower the barriers to entry, much like the pharmaceutical companies have done to drive down cost."

Chen acknowledges that supplying software is one thing; applying it is another. "Freescale's sensor fusion software is a starting point and it's all very well having free code, but people have to see how it works. We think customers will be interested in using the K64F Freedom platform as a sensor fusion starter kit. It's an inexpensive way for them to start developing. And if a company wants professional help, we're offering commercial support for the code."

He noted that, since the code was made open source, it had been downloaded more than 200 times in two weeks.

Whalley sees only an upside to AIC. "It won't do any damage at all to the algorithm community; if anything, it will be a shot in the arm. We're looking to create a pie that is big enough for existing suppliers, but also an opportunity for new companies to develop products and sensors we haven't thought of as yet."

Concluding, Chen said: "Everyone who has invested time and energy developing algorithms knows that it doesn't pay to reinvent the wheel. If there are ways to change this, it will benefit a lot of people."

Author
Graham Pitcher

Related Downloads
66414\P20-21.pdf

Comment on this article


This material is protected by MA Business copyright See Terms and Conditions. One-off usage is permitted but bulk copying is not. For multiple copies contact the sales team.

What you think about this article:


Add your comments

Name
 
Email
 
Comments
 

Your comments/feedback may be edited prior to publishing. Not all entries will be published.
Please view our Terms and Conditions before leaving a comment.

Related Articles

ST buys Atollic

STMicroelectronics is acquiring the software-development tools specialist ...

Close to the edge

There is a tension that lies at the heart of the software defined network ...

Under exposure

Chipmakers are set to break the 10nm barrier as they move from the test-chip ...

Core competence

It will probably be no surprise to discover that applications are becoming more ...

Audio power amplifier

Toshiba Electronics Europe has developed a 4-channel, high-efficiency class-AB ...

Auto grade op-amps

ROHM Semiconductor is now offering automotive-grade op-amps featuring superior ...

MSP430FR4x MCUs

The MSP430 ultra-low-power microcontroller family now includes the ...

Shaping the future

Alexander Everke, the CEO of ams, started his career in the semiconductor ...

A man with a plan

Cypress Semiconductor was formed in 1982 and went public in 1986. Until 2016, ...