Saturday, November 17, 2007

The weird world of ITS, VII, IVI

Let me just clarify right away:

ITS: intelligent transportation systems
VII: vehicle infrastructure integration
IVI: intelligent vehicle initiative

It’s an insular and esoteric world that is being created by Department of Transportation and industry experts. Excellent goals are behind the acronyms. Vehicles on the highway will collect and transmit data to each other and to centralized databases so that we can: avoid collisions in intersections; avert hundred-car pileups; avoid congested highways: ultimately be able to have cars on autopilot that can squeeze more vehicles onto the same roads and move them at higher speeds safely.

I’m all for these goals. Clearly this is what the future will bring us. The thing I find insular, esoteric, and weird is the method being used to accomplish these goals.

All of these systems are based on the wireless frequency of 5.9 GHz, specifically allocated for DSRC (oops, another acronym: Dedicated Short Range Communication), specifically dedicated to wireless use among automobiles.

Now here is the question, and I ask it honestly, why do cars need their own frequency and therefore their own hardware and software?

The rest of society – that is financial, education, health, consumer, retail, emergency services – are all happily communicating on other frequencies, using equipment that is interoperable, more or less. Data bits are data bits. Right now we mingle voice, data, video, graphics, music over the same devices, and increasingly one single device. In what way is parking data, or congestion data different?

Is the security needed for transmission of financial data less demanding than that needed for transportation?

Are the wireless mobility demands required by the cell phone in my car, or the laptop in police cars different when encased in the car itself?

Curiously, this requirement for a different standard means that we can’t use the cell phones, laptops, and wireless gateways that permeate our environment. The VII, IVI, and ITS system demands a whole new network and set of investments be made before the system starts throwing off benefits. Why go this route?

It would cost much less money, be much less complicated, and be a reality much sooner if the goals of car-to-car networking were achieved with the devices and standards used by the rest of the wireless world.

Hmm. Maybe that is the point. Weird.


david said...

Hi Robin,

You make the point that there are many other wireless technologies around, and hence ask why these shouldn't be used for car-to-car communication.

From my perspective, the answer is that there isn't anything that's overly suited. Cellular communications are (at present) inherently single caller to another single entity. All communication is via fixed base stations, and round trip times are commensurately long compared to broadcast by a car to others in its immediate area. Standard WiFi (802.11b) uses a chunk of spectrum that can be used by anyone for roughly anything: that's not ideal for safety-critical applications because (nevermind the malicious transmitters) I could set up my cordless phone and suddenly the platoon of cars driving by my house might not get receive each other's messages (NB that's unlikely, but so goes the thinking).

DSRC is actually making use of the IEEE 802.11 standard, in the form of another ammendment known as 802.11p. So it is using existing technology, just in a frequency band that is less congested.

Yes, we do need to deploy a whole load of new infrastructure. But unless we were using an already-deployed ubiquitous network like the cellular networks (which isn't really possible, see above), we need to do this anyway: at present we don't have WiFi (or similar) nodes on traffic lights.

So: yes, it's true that we could use 802.11b/g WiFi, 802.11p has some features which are more beneficial (e.g. a dedicated safety channel, theoretically guaranteed Quality of Service/Priority for certain messages) for a vehicular scenario. The nice thing about the 5.9 GHz band is that it's not far from where 802.11a already operates (5.7 GHz), so the hardware isn't all that different anyway.

My guess is that you'll see the DSRC radios that get deployed in vehicles will be capable of 802.11a/b/g/p/n all in one (multiple radios in the same package at any rate), and probably cellular too. The cost of vehicular deployment is in purchasing the boxes and fitting them: what goes in the box as regards radios is unlikely to be particularly significant.


Best wishes,


Robin Chase said...

I'm just impatient. We could build a lot of these applications today
on the back of broadly available wireless technology. The market
forces behind 802.11a/b/g/n are huge, so the costs of that equipment drops and that technology improves. DSRC isn't compatible with any of that so while
it might benefit a bit from falling technology costs due to spin offs
from consumer electronics, it won't have the breadth of implementations and competing technology providers that exist in consumer electronics -- and they are what really drive the costs down and the functionality up. DSRC also doesn't seem to be moving very
quickly, and will end up being a technology targeted at a narrow set
of applications that it seems doomed to very slow evolution.

The benefits you site for DSRC aren't compelling and can be had using other frequencies.

I do hope that 802.11p (and 802.11s for that matter) become standards
quickly, and that we're able to build open platforms based on them.
But if that doesn't happen quickly, we're not waiting. We're starting
to see some mobile devices and applications that aren't -- lots of
devices including phones and in-vehicle navigation devices that are dual cell/wi-fi network compatible.

David Cottingham said...

I think there may be a misunderstanding here: DSRC is just another name for 802.11p, in the same way that WAVE (Wireless Access for the Vehicular Environment) is yet another name for it. The waters are muddied because people also use DSRC to refer to earlier generations of microwave tag ("tag and beacon") systems, such as E-Z pass.

Hence, DSRC _does_ build on widely available 802.11 technology.

The intention behind DSRC (aka 802.11p) is that it will useful for safety critical applications. Currently there is no widely deployed technology that caters for such applications.

Meanwhile, there are indeed lots of applications that are NOT safety critical and could use existing wireless deployments. Indeed, we're starting to see those, given that, as you point out, devices with multiple radio technologies are available.

As regards frequency, DSRC utilises an dedicated portion of spectrum. It could use (almost) any other portion of spectrum instead. The allocation is purely one of convenience. Hence, you are correct in saying that the benefits could be had at other frequencies. However, that does NOT negate the need for A technology of some variety that caters for automotive safety critical applications. We don't have one at present, other than in the development of DSRC.

It may well be that we end up with DSRC _only_ being used for safety critical applications and existing ones used for (e.g.) entertainment. One good example is HSPA in the latest deployments of cell networks, where multi-megabit throughputs are achievable. That's available right now. However, it's not ubiquitous, and (to reiterate) it's not suitable for safety critical apps.

If the goal is to use existing infrastructure for, say, tolling then that's fine. It's already done (look at pay as you drive insurance products that use GPRS as provided by cell networks). Parking payment machines in some cities now also use the cell network, or WiFi mesh networks.

In summary, we already use commodity wireless infrastructure for many transport applications, and that usage is growing fast. But we don't have anything suitable for safety critical functions. Hence, we need something new. If that new technology can ALSO be used for, say, tolling, then great.



Mike said...

This is just to supplement David's comments. Let's start with a bit of history: back at the origin of the Intelligent Transportation Systems program, even before it was called ITS, the approach was (and continues to be) the use of widespread commercial communications technologies (both fixed and mobile), wherever possible. The phrase "Dedicated Short Range Communications" originally referred to those very few areas where something "dedicated" to ITS was a better approach (e.g., current and past generation electronic toll collection). An example of the use of commercial technology is the use of cell phones as traffic probes. While this technology has had its problems (the first several generations failing to achieve accurate results), the current systems are showing improvement. This is part of ITS. Similarly, the 511 programs throughout the country provide traffic and road conditions via practically any landline or mobile phone, by simply dialing 511. About 1/2 the US population currently live in areas with 511 service. This same information is available via websites and in some locations it's been optimized for mobile web browsers as well.

Turning to VII and DSRC/WAVE/802.11p, I agree that many of the applications that may use 802.11p at 5.9 GHz can (and probably will) use other media. In fact, the coalition of federal, state, and automakers working on VII have said as much.

However, for localized safety of life applications, handling both vehicle to roadside and vehicle to vehicle communications, you need very quick access as each vehicle comes into range (it can't take seconds to aquire and register with a network), very low latency (cars operate in close proximity at high speed, and reaction time must be factored in), high reliability, preferably good security (to keep hackers out of wireless safety systems), and preferably privacy protection (because many of us would prefer neither industry nor the government to keep massive databases of all our movements). 802.11a/b/g/n, WiMax, cellular, etc. don't fit that bill. Hence the need for some modifications to 802.11 (802.11p) and some additional functionality on top of 802.11 (IEEE 1609 series of standards, currently approved for trial use), at least for localized safety applications.

To say the 802.11p isn't compatible with 802.11a/b/g/n is technically true, but then, 802.11a and b aren't compatible with each other (but dual mode or even tri-mode a/b/g systems could be cheaply built), and the same holds true now with n, and soon with p.

802.11p uses a slightly different frequency because for safety of life systems it is desirable to have co-primary status, rather than accept and live with interference from thousands of licensed and unlicensed devices. 5.9 GHz is where spectrum was available from the FCC in the United States.

In conclusion, yes, we can and should do even more with available commercial technology. But the ITS community has focussed primarily on this approach since day one. For some applications, however, the requirements mandate a different approach. This same approach can, technically, support the other applications. Whether or not these other applications use 802.11p in the long run is primarily a question of relative cost.

Anonymous said...

Nice blog as for me. I'd like to read a bit more concerning this matter. The only thing I would like to see here is a few pics of some gadgets.
Katherine Karver
Cell phone blocker