Part 2: Where do those Embedded Forecasts come from – and why this question should make you nervous
January 14th, 2010 | 
Author: You are currently browsing comments. If you would like to return to the full story, you can read the full entry here: “Part 2: Where do those Embedded Forecasts come from – and why this question should make you nervous ”.
Posted in Best Practices, Market Intelligence, Marketing and Sales, Uncategorized | 
Tags: Forecast, Marketing
I’m not sure why mil/aero vendors targeting medical is laughable. Are you saying the safety/security assurances given by certifications such as DO 178-B are not valuable enough for any medical devices to be designed with a system powerful enough to support a certified OS?
No all medical products – just those patient monitoring products that have minimal frequency response requirements (under 100HZ). Patient monitoring safety, over the years, has been most effective by the design of alarm systems – many of which can distinguish between a patient disconnect, 60 cycle noise and an actual cardiac incident. They can also detect a systems failure – or power loss so that a patient will not go unmonitored. Also, there are OSes (ThreadX, MontaVista Linux and Nucleus – Micrium can be included but has 178B) that are used in hundreds of millions of applications – most of which having more severe requirements than patient monitoring. Having built 4 medical device companies and taken 2 of them public, I feel qualified to comment in this regard. Thanks for your comment – Jerry
I regret that I do not know much about the medical device industry, which may be part of my confusion. There are still 2 points I am unclear about: 1) Are the majority of medical devices considered patient monitoring products with minimal freq. resp. reqs? If not, isn’t there still a large market for certified OSs in the medical industry? 2) I can understand how applications could have more severe requirements than patient monitoring, but if these are devices detecting cardiac incidents, wouldn’t having something like a separation kernel (which I believe is part of the Common Criteria certifications you mentioned) provide safety/security that is a critical requirement?
I’ll be posting a medical device paper soon. To respond to your question, one has to look at patient attended (e.g., ultrasound, dialysis, CT scan, etc.) and patient non- attended (e.g., monitoring – even remote) devices for specific requierments (e.g., realtime, high speed, etc.). So – if I need to measure ECG (100HZ), pulse (4 HZ), temperature (minutes or more), respiration (3-20/minute), oxygen and CO2 concentration, etc., why would I need a separation kernel? MILS, Common Criteria and EAL, as well as DO 178B offer assurances specific to catastrophic events in unusual environments (avionic, etc.). Unless I have to use multiple OSes in a single application virtualized capabilities are unnecessary. ThreadX has been ceritifed by Welsh Allyn for certain applications but Express Logic hasn’t taken the trouble to incur the cost for other certifications as they are focused on different markets. Same for Nucleus, MontaVista Linux and Micrium. EMF data shows that these OSes enjoy as good or better design outcomes as do the more powerful OSes such as Integrity, VxWorks and LynxOS. Are they less capable for non-mission critical applications? I think that the fact that they are deployed in hundreds of millions of applications speaks for itself.