rwb and William K, precisely my point! Define by vertical application and functional intent. I would add that as desktop computing functions are all but replaced by handheld devices, the standalone function of IT "disappears through the skylight" (admittedly, not completely, as there will always be client-server back ends), making it all the more difficult to define embedded - or come to think about it, to define the non-embedded.
I agree. The vague catch-all phrase is confusing. As a manufacturer of military embedded computers (think advanced radar, satellite communications, missile fire control, etc.), we (Acromag) find it is difficult for engineers to find the right type of embedded computers in a Google search. Imagine having to specify the engine and body configuration for your next vehicle if the auto industry didn’t have defined segments like car/truck and pickup/SUV. Auto industry analysts don’t bundle the sales of engines for earth movers with those for station wagons, so why does the electronic industry lump embedded systems in an iPod with those in a B-2 bomber. We need a way to neatly segment all these embedded system components by application (consumer, commercial, industrial, rugged) and by form factor (chip-level, microcontroller, mini SBC, rack SBC, box system, chassis).
First, the difference between a microprocessor, minicomputer, and a mainframe: The micro is a chip IC that will fit in your shirtpocket, the minicomputer you can carry with two hands, and the mainframe took a forklift, or two men with dollies to move it.
An embedded system would be the microcontroller controls for some package that is not primarily a computer. So a dektop tower computer does not have an embedded processor, but the toaster and the cell phone do, so they have an embedded system. A definition based on functional intent would seem to be more logical than anything else that I have heard recently.
I agree Loring from a software and networking sense. From a hardware sense, automation components like robots, PLC's and stepper motors are generic can be reused rewired and reprogrammed over and over for various tasks. The control board from the coffee maker however is inflexible and thus embedded...
Thanks, everyone, for the thoughtful comments. John, a hardware engineer may think of embedded semantics as something only a market analyst cares about, but I think it makes a difference to a device programmer or RTOS kernel developer, albeit from the reverse perspective of saying "What is non-embedded any more?" It is becoming less and less interesting to ask how many Windows vs. Mac vs. Linux seats are in an enterprise, as the offline information processing becomes almost a subset of real-time data acquisition or transaction processing. In that scenario, what we think of as client-server IT may fade into the background as much as a CAN or Fieldbus network. Chuck, the implications for the monetary size of the embedded market under that assumption are indeed mind-boggling. And Greg, your definition is great, except that as more and more devices are networked, the embedded system is arguably the network in its totality.
Embedded Systems are clearly defined in the product and equipment world as I see it. A factory automation system, usually built in low volume, can consist of PLC's (Programmable Logic Controllers) Stepper Drivers, Servo Amplifiers, Sensors, Temperature Controllers, HMI's etc. These components are off-the-shelf and modular. They can be combined and programmed for many purposes and, as such, are Autonomous.
Products built in high volume however require dedicated and cost reduced control systems. Coffee makers and dish washers are good examples of products with embedded controllers. They have all the similar functions, however they are usually contained on a few chips and printed circuit boards...
No matter how we count them, the numbers are mind-boggling. 1.8 billion embedded systems shipped in 2011 and 4 billion (counted the same way) in 2015? Most amazing of all -- 14.5 billion processor cores by 2015. Now we know why MCU makers spend so much time figuring out ways to market their devices to so many different segments.
The question about what label to afix to electronic equipment with some intelligence sounds a lot like debating when grains of sand becomes large enough to count as gravel. Companies that run market surveys have a vested interest in quantifying the number of "embedded systems" so they can show growth or contraction in market segments and sell more reserach reports. I doubt most engineers care. Much like the argument about art and pornography, engineers know an embedded system when they see it, but they might not always agree. Numbers seem meaningless.
In an age of globalization and rapid changes through scientific progress, two of our societies' (and economies') main concerns are to satisfy the needs and wishes of the individual and to save precious resources. Cloud computing caters to both of these.
For industrial control applications, or even a simple assembly line, that machine can go almost 24/7 without a break. But what happens when the task is a little more complex? That’s where the “smart” machine would come in. The smart machine is one that has some simple (or complex in some cases) processing capability to be able to adapt to changing conditions. Such machines are suited for a host of applications, including automotive, aerospace, defense, medical, computers and electronics, telecommunications, consumer goods, and so on. This discussion will examine what’s possible with smart machines, and what tradeoffs need to be made to implement such a solution.