The truth of the matter is that the US is extremely vulnerable on so many fronts in this cyber war. We have so much of our infrastructure that is accessible through the various networks. air traffic control and electrical generation and distribution systems are all vulnerable. In fact it is a good bet they are already penetrated and sleeper code is in place to do harm when the controlling organization or country wants to initiate an attack. I venture it is a safe bet that the US is the most vulnerable of any country.
Another long overdue consideration is that defense in cyberspace is far behind offensive capabilities. Countries like North Korea are not as susceptible to cyber attack as we are. They just don't have that much infrastructure to protect.
It will take a great deal of attention and money to bring this situation under control. Embedded chips made in other countries may not be safe from malicious code being designed into the system fromt he beginning. Detecting this and preventing it use will require additional efforts that might not be possible with the existing systems.
I agree, Ivan. This underscores the importance of recent "safety microcontroller" rollouts by TI, Freescale, and Renesas. The Embedded Systems Working Group is one more sign that we are collectively paying attention to vulnerabilities of power plants, air traffic control systems, financial systems and, yes, train yards.
The way to protect factory and any other important machine control networks is to not allow the capability of external modification to exist at all. Of course it is more convenient and cheaper to change the program and the parameters over the network. It is also not possible to have this ability and have it be secure, we all know that. But real security does have a real cost, which is that somebody would need to actually visit the controller and alter the program or settings. Any outside access is not completely secure, only fairly secure, and we all know that any security measures only last untill somebody cracks them. And that always happens.
So it becomes a trade-off of costs-which costs more, manual updates or hackers damage? Each can be calculated, and then a decision can be made.
Thanks Ivan, Chuck, William for great points. William, the ideal case you raised of a physical control over the hardware network might be re-interpreted by others to say a hard-wired physical-layer network, preferably fiber, should be used for changes in configuration. Yet someone will always come in and demand wireless updates for reasons of cost, and all the best ideas for trusted systems fly out the window. This TCG work will be interesting to watch.
The fact is that my assertion was that a primary way to be secure was to not have the capability present, not wired of fiber or wireless. Tha capability of remotely changing the program would not be present in the system. No, there is no question about it being less convenient, but a disaster is more inconvenient. But if the way to change a calibration or a program requires physicaly operating a switch at the machine, then all remote hackers are kept out.
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.