Back when I used to work at a chemical plant, we had two networks: one for regular corporate/internet access, and one for plant control. The later was physically isolated from the former, which is why scenarios like Die Hard 4 are so hilarious. You can't hack a plant control system from outside because that network is physically isolated.
Anyway, I hope this same practice is used in this standard, where critical control systems are physically isolated from these external communication systems.
They could, although it'd be needlessly elaborate compared to turning it off, or driving an old car without such functionality. I see more and more luxury cars have things like radar-based collision avoidance built in, which will probably become standard and eventually mandatory on other vehicles. So cars will have to distiguish between what another car is telling them and what their own senses indicate.
Also, but a fit farther down the line, even if a vehicle is falsely sending 'I'm slowing down and driving totally safely, nothing to see here!' to other vehicles, if that message is inconsistent with the vehicle's telemetry (approaching rapidly), other cars are going to know there's something amiss.
My experience is different - I help develop systems like that and what I often see terrifies me. Yes the system may be on a different physical network, but almost always there is at least one point of entry i.e backup server, domain controller etc. And I can't believe how many times I worked in front of the computer that had usb storage turned off, but had a cd writer installed ;)
Did you miss the part in that Die Hard movie where they have to physically break into a power plant because the network is not accessible from the outside? http://www.imdb.com/title/tt0337978/synopsis?ref_=tt_stry_plThe third stage will be to take out the power grid; this cannot be done remotely but requires intervention at the actual physical plants.
Hmm, I guess I'm remembering it wrong (I thought they were controlling stuff in a van or something). Luckily, the scenario would still be impossible since every gauge/pump/etc. has a hardcoded interlock that shuts it down if it goes past that limit. You physically can't over-pressurize a steam pipeline because the once the pressure gets above a certain point, the pump is hardcoded in its local circuit board to stop.
There were other good tech fails in the movie, though. My favorite is when they break into a secure NSA data storage facility and steal all its contents... in a portable hard drive.
1. Tell that to Iran... They thought the same thing and look what happened to their centrifuges.
2. I hope so too, and it likely will be. Many current car designs already have a separation of interests, with things like engine control systems only giving "read only" access to entertainment systems and similar (so you can get diagnostic information without the entertainment system having the chance to alter engine settings).
Some cars have "over the air" updates which CAN update the engine control systems however, that can only happen when the engine is not running and they often use traditional digital signing to confirm the authenticity of updates.
I do suspect you'll see people try to exploit these V2V systems (e.g. next blackhat?). Let's hope they're designed with security in mind. We've already seen the security services using hacked cars as lethal weapons to kill journalists.
I suppose. Things like Stuxnet require government-scale resources, though, so I'm not sure you have to worry about that kind of attack as much for V2V standards.
That was the point. GP said, "You can't hack a plant control system from outside because that network is physically isolated." What happened in Iran proves you can (or more strictly, that systems don't remain perfectly physically isolated even with an air gap).
Conspiracy theory slip in of the week. 'Not only did the gpvernment totally kill him, they did it with a method no one has yet shown is even technically possible'.
Where there is a will, there is a way. There will always be ways around an air gap.
The Iranians had their centrifuges isolated by an air gap. Until someone (cia? mossad?) dropped USB thumbdrives in the parking lot that contained stuxnet and people being curious plugged them into computers on the air gapped network to see what was on them.
Target theoretically had isolated the network that their cash registers where on, but someone mis-configured a VLAN and it was connected to the HVAC network which was vulnerable.
I don't think that is an excuse to not do some super cool shit regarding inter-vehicle networking, but you just have to be careful as hell. Based on how horrible most car's entertainment/gps systems are, I'm not sure I put a whole lot of trust in the auto manufacturers to put as much scrutiny on it as they should.
I think this is fantastic news (and I suspect many people around here will agree). However I am worried this might cause a certain demographic (anti-government types) to come out of the woodwork about "the government spying on our cars" and similar.
It will only take one talking-head to get that ball rolling against it and once it starts I don't foresee it stopping. For comparison, like the "issue" with students using a fingerprint for their school libraries (and the huge negative backlash against that because "something something our kids aren't criminals! Government spying! Privacy!").
That's a very anti-safety attitude. If your assumption of safety comes from the concept that everyone should be perfect all of the time then it is deeply flawed.
Even pilots need systems like ACAS to avoid collisions, and that is on top of on-board radar, significant training, and ATC who are there to make sure such things never occur. Yet how many times has there been a collision averted by ACAS since its introduction? Dozens? Hundreds? Who knows (FAA maybe?). A lot.
If our attitude to driving was more similar to our attitude to airline safety then our roads would be significantly safer today. In the airline industry it is all about continuous improvement and learning from even minor accidents, on the roads we just ignore such things as a "regular problem we have to put up with."
Its possible to turn the argument around to "If your assumption of safety comes from the concept that all safety equipment should be perfect all of the time"
Given the inherent immense liability problem, it seems almost impossible to release this as a product without some kind of government blanket immunity to criminal/civil prosecution, otherwise the mfgr will be dragged into every legal dispute tangentially involving the car and a crash. Therefore with no motivation to produce a perfect product, perfect products will not be released.
The end result will be drivers with a careless attitude because they purchased something to take responsibility, causing the overall accident / death rate to increase, despite society as a whole spending more money. I'm not seeing much net good coming out of this.
Road design can also have a lot to do with it. There's a rather dangerous intersection close to my house, where there's a serious right-angle collision about every 2 months on average (over a >2 year observation period). It's close to a freeway on-ramp (so you get people who are in a hurry), visibility is poor in some directions due to large trees and bus shelters, and you have a 2 lane street crossing a 6 lane one (zoning boundary that's also a transit artery).
I couldn't agree more. We have roads set up in grid, forcing intersections every mile. Not only is this a huge safety issue, the stopping is where a large part of fuel goes and is where half the time spent commuting goes. The urban drive cycle has an average speed of just under 20mph due to all the stopping.
I'll be the first. I'd be very concerned about this type of communication system and my vehicle. My concerns aren't limited to government surveillance, but that certainly is one. I'm also concerned about malware or other security vulnerabilities. I'm also concerned about the government "requiring it" and the related costs associated with it. It will certainly drive up costs to consumers.
I really don't think there's any reason for car's to have inter communication systems. Using statistics and light signals, cars can have the theoretical capacity to guess what others cars are about to do.
Yes, I also believe this a great news. I can't want for the day for me to jump in the car from the bar drunk as hell and my car drives me home. But I also agree with the anti-government types in the respect that the government should not be forcing technology into law. I mean this not because the 'OH MY GOD THE GOVERNMENT IS SPYING ON ME'. But I mean in the way that then there is one a monopoly on the technology and how the law is written. If the law states 'contractor A' will supply the hardware, or that the law states 'must use bluetooth technology' or 'must use wifi or radio frequencies x,y, and z' I do have a problem with that. If anything it should be 'optional' because I don't want more features in my car. I also know plenty of people do NOT want this because they don't want all the 'features' and just want a car to get from point A to point B. Not only that. But this will also increase the cost of the car for development and hardware, etc.
This is actually one of the few reasons I wouldn't be totally against this kind of technology. (For security reasons, I think that putting software with open on the air connections into a 1-2+ ton metal machine powered by explosions that moves ridiculously fast is mostly a bad idea.) Particularly in America, even fairly 'bicycle friendly' cities can be pretty dangerous to ride a bike in, and often for design reasons that force bikers to either endanger themselves or put themselves in a place where cars will make dangerous maneuvers near them. (Example below) A system that you could attach to a bike that would basically give your bike a warning system for nearby cars (and them a warning system for bikes) would be fantastic. Another good thing about it would be that it could be used to give motorcycles the profile of a car, which would make riding them much less dangerous: no matter how safe a motorcycle rider you are, you're smaller than a car and that lower visibility can cause accidents.
The promised example: A common dangerous design for bikers in cities is bike lanes right up against a lane of parked cars. Depending on how fast a biker is going and the local terrain, someone opening up their door into the bike lane can be lethal: you might not have time to stop yourself, and your options might be to slam into the door and maybe a person getting out of the car or to swerve into traffic, both of which are very dangerous. The safest thing to do when that's the case is actually to ride in the car lane, but that has its own dangers: many cars will try to pass you by going part of the way or all of the way into the lane to the left of them, which might have oncoming traffic or traffic coming from behind that the drivers don't check for, which can lead to them swerving back towards you. That means that the most safe thing you can do for everyone in that situation is to ride dead center of the lane so that passing you by going only halfway into a lane isn't possible (since many car drivers seem to treat that kind of passing differently from an actual lane change), but car drivers absolutely hate that and will do all sorts of other dangerous things like following super close.
"Please use your turn signals." "No, if you get in that lane you block the bike lane in about 100 feet." "Your brights are on!" "If you would be so kind as to reduce the bass just a small bit..." "Are you blocking the post office drop off lane because you're driving a Lexus or have you suffered some sort of heart attack?" Yeah, this'll be good.
To get any network security right, you have to start with the assumption that each node is fully open to being controlled/modifiable by its end user, and that there is a non-negligable amount of unaccountable trolling nodes that will actively seek to disrupt the system. In the world of corporate embedded design and agency approvals, how likely do you think this is?
Look at flaws in aviation communication systems like ADS-B. This is highly regulated territory and full of agency approvals. With car owners as customers, the manufacturers will need to build the onboard units as cost effective as possible. I doubt that they will develop hardware and software from scratch in order to employ maximum security.
I wasn't using "agency approvals" as a positive point for security :P. Real security properties just aren't legible to that kind of environment.
Most likely they'll start off with some "tamperproof" microcontroller platform to assume away user control, further rely on keeping the protocol documentation secret and other obfuscation, and then write naive code that trusts every node to behave because they can't conceive of how brittle that kind of system is.
This sounds really cool. Unfortunately I don't know if I will get any direct benefits unless they add a way to upgrade old cars. I don't plan on buying another car. Ever.
I'll still get indirect benefits (safety while walking etc).
I really wish there were exceptions for low-volume production. Among other things, a car sold in the US today must have anti-lock brakes, airbags, tire pressure monitoring system, and electronic stability control. These are required even if you want to make a total of 10 cars. This kills the enthusiast market and makes it hard for car manufacturers to experiment with low-volume models.
Meanwhile, anyone can buy a 1000cc superbike that can go from 0-60 in 2.8 seconds. (And it gets to use the carpool lane.)
Those are requirements because you share the road with others - if you opt out of anti-lock brakes you might slam into the side of my car and kill me. So personally, I'm quite glad these things exist.
Want to make a car to take out on a test track? Go crazy. Want to drive it on public roads? Follow rules.
Those requirements did not always exist, and now that they do, are we really that much better off? Some obviously have more impact than others.
The concern of the GP comment is that regulation is stifling potential innovation, so we should always be on guard about nickel-and-dime regulations that come down from the govt.
But the incredibly regulated market also has its costs. A great example exists in China, which is currently experiencing a revolution in low-cost electric vehicles[1].
The low-speed electric vehicle revolution has exploded in China's countryside, so most people living in cities in China do not know these vehicles exist. The industry literally grew up while regulators in the cities weren't looking.
That article is worth taking in as a whole, for its implications are broad. First and foremost, this unregulated and illegal transportation revolution is becoming a means for China to remedy some of its severe pollution problems. Second, these networks of new manufacturers will become the experts in a manufacturing domain which will certainly have reach well outside of China.
In the US (and the West more broadly) we're progressively adding regulatory burdens to the auto manufacturing market, but little comprehensive cost-benefit analysis. This itself is a substantial inhibitor to innovation which might completely upend transportation as we know it.
For more really excellent discussion along these lines, I highly recommend Horace Dediu's Asymcar podcast[2]. Asymcar digs into the nature of the transportation market, its players, and the conditions under which transportation market disruption might arise.
Very interesting. If I happened to live in a gated community, with beach access or something, I'd rather drive one of these around than a golf cart--although servicing it might be a problem.
But by that logic we should ban every single car older than 30 years - they have no anti-locking breaks or airbags or electronic stability control. As long as these cars are a tiny minority it's not a problem - the same with enthusiast self-built cars - as long as they pass all the inspections(I know most US states don't have the equivalent of European MOT, but you should), then I have no slightest problem with them.
What is needed is a standardized set of these required by law components, they should be mass produced and very cheap. Did I just find a niche - a hobbyist pack for car makers?
As far as I know my car (a 2011 Chevy Aveo) does not have anti-lock brakes. Wikipedia indicates that they are mandated in the EU but not the US.
My car does have a tire pressure monitoring system though. I believe it was also one of the last cars produced for the US without electronic stability control.
You're right that ABS is still an option on some cars. I have to ask, though... Why on earth would you not spring for it? Anti-lock breaks are enormously effective.
The short answer is that I bought the cheapest new car I could find.
The long answer is that I want my big expensive stuff to be simple and easy to maintain. It has a manual transmission, no air conditioning (I could have added it for $2k USD), and crank windows. I was disappointed to find that it has electric locks and lock/unlock buttons on the key. Power assist steering was something I didn't really want either. I wouldn't have paid for a radio if it was optional.
My other car is a '76 VW Beetle (currently undergoing reconstruction). Classic Beetles don't even have radiators.
I feel nearly certain that this is just lobbying by companies that have patents on certain aspects of the communication and would like to have a mandated market for some electronic modules on the car.
Same thing happened with tire pressure monitoring. In-wheel systems were being developed, but there was really no market until they were mandated. Meanwhile, your ABS/traction-control system can passively detect low tire pressure with no additional hardware.
For a govt mandated feature, FRAND rules would apply, so it wouldn't be any different than all of the patents Samsung or Nokia hold on tech like LTE (for instance)
It still creates a market for more hardware, weather patents apply or not. For the vehicle manufacturers it creates another barrier to entry in the market.
But the barriers for entry are already high due to regulatory concerns, safety concerns, marketing, and the cartel-like nature of the automotive business.
This has great implications for emergency vehicles. The current methods of inter-vehicle communications used by emergency vehicles (loud ass sirens and flashy lights) are ridiculously error prone, and becoming less and less effective as manufacturers get better at insulating cars.
Even when the driver does hear it, it's often hard for them to localize where the sound is coming from.
"An ambulance is approaching from the left" would be fantastic.
"The current methods of inter-vehicle communications used by emergency vehicles (loud ass sirens and flashy lights) are ridiculously error prone, and becoming less and less effective"
Yes, you could solve that problem with complex, interdependent systems that will fail in unpredictable ways.
Or you could just mandate better driver training and education which is the clear failure point here, at least in my experience.
My experience is both as a driver and as a firefighter who regularly drives a fire truck.
You can blame individual people, but the mfgrs will simply purchase immunity from their hired politicians. Its a question of responsibility. Individuals have it. Corporations do not.
I'm not sure that's true. Auto manufactures are a really big target for liability in accidents (I'd likely get more from GM then an individual in a won law suit), and a quick search [1] seems to back my feeling that there is sizable law and case law that define this area of liability. I'm not saying that it's simple to win against any sufficiently funded entity, but it's not impossible that anyone has immunity.
I guess you missed the CEO of General Motors getting grilled by congress earlier this year and tha massive writedowns they took due to the settlements they entered into with the families of people hurt/killed by faulty ignition switches. I could cite a slew of similar examples.
Training will not help a driver localize where a EV is approaching from. I'm not suggesting this system would supplant traditional warning systems, but would simply augment them with additional information. It would allow drivers to make better decisions about how to yield, rather than having the decision forced upon them when I'm suddenly right on top of them.
If it's relevant, I'm firefighter/paramedic, been driving EVs for a decade or so.
I live in Chicago, and sometimes it's really hard to tell where the EV is coming from. They're ear-piercingly loud, and the sound bounces everywhere. You hear the siren, then frantically start checking all your mirrors every 0.5 sec, which probably creates more of a distraction to the driving you should be doing.
Wouldn't surprise me if wireless hardware companies are involved in this. I'd be much more comfortable with passive sensors being used, but then you couldn't really do traffic optimization. If it's lives you're trying to save go passive. Animals cause accidents too. If traffic congestion is the goal, then we can start talking about this.
> The information sent between vehicles does not identify those vehicles, but merely contains basic safety data. In fact, the system as contemplated contains several layers of security and privacy protection to ensure that vehicles can rely on messages sent from other vehicles.
With all of the stuff talked about in the article I don't see how's that supposed to work.
Next up: Mandated GPS tracking in the black box with dial-in ability. For 'public safety'. O ya, and we should have new cars spy on the other cars and automatically report back to the mothership when they misbehave.
78 comments
[ 2.7 ms ] story [ 124 ms ] threadAnyway, I hope this same practice is used in this standard, where critical control systems are physically isolated from these external communication systems.
This by its very nature cannot be a closed network.
Also, but a fit farther down the line, even if a vehicle is falsely sending 'I'm slowing down and driving totally safely, nothing to see here!' to other vehicles, if that message is inconsistent with the vehicle's telemetry (approaching rapidly), other cars are going to know there's something amiss.
1. Tell that to Iran... They thought the same thing and look what happened to their centrifuges.
2. I hope so too, and it likely will be. Many current car designs already have a separation of interests, with things like engine control systems only giving "read only" access to entertainment systems and similar (so you can get diagnostic information without the entertainment system having the chance to alter engine settings).
Some cars have "over the air" updates which CAN update the engine control systems however, that can only happen when the engine is not running and they often use traditional digital signing to confirm the authenticity of updates.
I do suspect you'll see people try to exploit these V2V systems (e.g. next blackhat?). Let's hope they're designed with security in mind. We've already seen the security services using hacked cars as lethal weapons to kill journalists.
Like a "See, the government even makes massive, obvious mistakes" type of comment.
What is this in reference to?
See here:
http://en.wikipedia.org/wiki/Michael_Hastings_(journalist)
The Iranians had their centrifuges isolated by an air gap. Until someone (cia? mossad?) dropped USB thumbdrives in the parking lot that contained stuxnet and people being curious plugged them into computers on the air gapped network to see what was on them.
Target theoretically had isolated the network that their cash registers where on, but someone mis-configured a VLAN and it was connected to the HVAC network which was vulnerable.
I don't think that is an excuse to not do some super cool shit regarding inter-vehicle networking, but you just have to be careful as hell. Based on how horrible most car's entertainment/gps systems are, I'm not sure I put a whole lot of trust in the auto manufacturers to put as much scrutiny on it as they should.
It will only take one talking-head to get that ball rolling against it and once it starts I don't foresee it stopping. For comparison, like the "issue" with students using a fingerprint for their school libraries (and the huge negative backlash against that because "something something our kids aren't criminals! Government spying! Privacy!").
Even pilots need systems like ACAS to avoid collisions, and that is on top of on-board radar, significant training, and ATC who are there to make sure such things never occur. Yet how many times has there been a collision averted by ACAS since its introduction? Dozens? Hundreds? Who knows (FAA maybe?). A lot.
If our attitude to driving was more similar to our attitude to airline safety then our roads would be significantly safer today. In the airline industry it is all about continuous improvement and learning from even minor accidents, on the roads we just ignore such things as a "regular problem we have to put up with."
Given the inherent immense liability problem, it seems almost impossible to release this as a product without some kind of government blanket immunity to criminal/civil prosecution, otherwise the mfgr will be dragged into every legal dispute tangentially involving the car and a crash. Therefore with no motivation to produce a perfect product, perfect products will not be released.
The end result will be drivers with a careless attitude because they purchased something to take responsibility, causing the overall accident / death rate to increase, despite society as a whole spending more money. I'm not seeing much net good coming out of this.
I couldn't agree more. We have roads set up in grid, forcing intersections every mile. Not only is this a huge safety issue, the stopping is where a large part of fuel goes and is where half the time spent commuting goes. The urban drive cycle has an average speed of just under 20mph due to all the stopping.
I really don't think there's any reason for car's to have inter communication systems. Using statistics and light signals, cars can have the theoretical capacity to guess what others cars are about to do.
I'm just reporting what auto manufacturers have been discussing at international standards meetings.
The promised example: A common dangerous design for bikers in cities is bike lanes right up against a lane of parked cars. Depending on how fast a biker is going and the local terrain, someone opening up their door into the bike lane can be lethal: you might not have time to stop yourself, and your options might be to slam into the door and maybe a person getting out of the car or to swerve into traffic, both of which are very dangerous. The safest thing to do when that's the case is actually to ride in the car lane, but that has its own dangers: many cars will try to pass you by going part of the way or all of the way into the lane to the left of them, which might have oncoming traffic or traffic coming from behind that the drivers don't check for, which can lead to them swerving back towards you. That means that the most safe thing you can do for everyone in that situation is to ride dead center of the lane so that passing you by going only halfway into a lane isn't possible (since many car drivers seem to treat that kind of passing differently from an actual lane change), but car drivers absolutely hate that and will do all sorts of other dangerous things like following super close.
Most likely they'll start off with some "tamperproof" microcontroller platform to assume away user control, further rely on keeping the protocol documentation secret and other obfuscation, and then write naive code that trusts every node to behave because they can't conceive of how brittle that kind of system is.
I'll still get indirect benefits (safety while walking etc).
Meanwhile, anyone can buy a 1000cc superbike that can go from 0-60 in 2.8 seconds. (And it gets to use the carpool lane.)
Want to make a car to take out on a test track? Go crazy. Want to drive it on public roads? Follow rules.
The concern of the GP comment is that regulation is stifling potential innovation, so we should always be on guard about nickel-and-dime regulations that come down from the govt.
The low-speed electric vehicle revolution has exploded in China's countryside, so most people living in cities in China do not know these vehicles exist. The industry literally grew up while regulators in the cities weren't looking.
That article is worth taking in as a whole, for its implications are broad. First and foremost, this unregulated and illegal transportation revolution is becoming a means for China to remedy some of its severe pollution problems. Second, these networks of new manufacturers will become the experts in a manufacturing domain which will certainly have reach well outside of China.
In the US (and the West more broadly) we're progressively adding regulatory burdens to the auto manufacturing market, but little comprehensive cost-benefit analysis. This itself is a substantial inhibitor to innovation which might completely upend transportation as we know it.
For more really excellent discussion along these lines, I highly recommend Horace Dediu's Asymcar podcast[2]. Asymcar digs into the nature of the transportation market, its players, and the conditions under which transportation market disruption might arise.
[1] http://green.autoblog.com/2014/07/25/a-window-into-chinas-lo...
[2] http://www.asymcar.com/
My car does have a tire pressure monitoring system though. I believe it was also one of the last cars produced for the US without electronic stability control.
The long answer is that I want my big expensive stuff to be simple and easy to maintain. It has a manual transmission, no air conditioning (I could have added it for $2k USD), and crank windows. I was disappointed to find that it has electric locks and lock/unlock buttons on the key. Power assist steering was something I didn't really want either. I wouldn't have paid for a radio if it was optional.
My other car is a '76 VW Beetle (currently undergoing reconstruction). Classic Beetles don't even have radiators.
Same thing happened with tire pressure monitoring. In-wheel systems were being developed, but there was really no market until they were mandated. Meanwhile, your ABS/traction-control system can passively detect low tire pressure with no additional hardware.
Even when the driver does hear it, it's often hard for them to localize where the sound is coming from.
"An ambulance is approaching from the left" would be fantastic.
Yes, you could solve that problem with complex, interdependent systems that will fail in unpredictable ways.
Or you could just mandate better driver training and education which is the clear failure point here, at least in my experience.
My experience is both as a driver and as a firefighter who regularly drives a fire truck.
Isn't that what you have now by depending on people?
[1] http://www.lawyersandsettlements.com/automotive.html#.U_Jd_2...
If it's relevant, I'm firefighter/paramedic, been driving EVs for a decade or so.
> The information sent between vehicles does not identify those vehicles, but merely contains basic safety data. In fact, the system as contemplated contains several layers of security and privacy protection to ensure that vehicles can rely on messages sent from other vehicles.
With all of the stuff talked about in the article I don't see how's that supposed to work.