Help me find a UWB providor that can track my trucks.

September 21, 2006 9:08 AM

TechFilter: Anyone familiar with Ultra Wideband?

In some of my recent research on RFID / RTLS / GPS as methods for Yard Management (tracking trucks / trailers / containers at a port or shipping facility, for example), I've stumbled upon the possibility of Ultra Wide-Band as a solutiton. Apparently this is a relatively new technology, and I'm having trouble finding much decent research on the technology, namely as it is applied for functional purposes (ie. yard management).

I'm kind of vainly hoping there's an RF / UWB expert out there somewhere that can give me some more insite into the technology, or perhaps even recommend some companies that are exploring this field for practical applications.

posted by allkindsoftime to technology (15 comments total)

I think UWB has not been approved for use by the FCC, so it's still just research stuff. It broadcasts over waaaay too much spectrum to be allowed under current regs. At least that's my understanding. (The flip side is that it's at extremely low power over huge banndwidth - thus, ultrawideband. It's spread spectrum on steroids.)
posted by GuyZero at 10:26 AM on September 21, 2006

http://www.ultrawidebandplanet.com/


It's cool stuff. I am not sure it's ready for primetime, yet.

I looked into it for a wireless video app last year, and will go dig up some more info and reply again, later, if I find anything.
posted by FauxScot at 10:30 AM on September 21, 2006

Time Domain Corporation sells eval kits. Did not look up current state of regulatory issues, but they were substantial last year.

http://www.timedomain.com/products/P2101E2EVK.pdf


These kits are for professional developers, not end users. UWB product design takes some god-awfully expensive test equipment. Even these eval kits were super expensive, as I recall. Not for the faint of heart.

It does look like magic. TDC also has a 'vision through a wall' imaging system that they sell to spooks and government agencies, if you really want to see something wierd.
posted by FauxScot at 10:40 AM on September 21, 2006

I am not an expert on this, but an enthusiast. As I understand it, RFID may be the best option for you. RFID doesn't need an on-board power source (the chips can be powered by the scanning radio signal), and I think UWB would. RFID technology is mature-ish, and there are established service providers out there.

I don't know how big your company is, how much you want to spend, or how much of the overall work you want/need to do yourself, but one established player is IBM. They have an SMB-sized offering, as well.

Here's the IBM info about it, and here's what one of their partners, CDW, offers of theirs. If I were you, I would call CDW and talk to them about what you need done.

Because it's not just the transmitters that you need to stick on stuff... it's the software that analyzes what the transmitters are doing and where they are going, and the integration with your supply-chain-management system if you have one, etc. Complicated.
posted by dammitjim at 10:47 AM on September 21, 2006

It's unlikely that the FCC will ever approve wideband for broad use. As GuyZero says, it's Direct Sequence Spread spectrum (DSS) on steroids.

What it's trying to do is to play in the cracks, to take advantage of guard bands in other people's licenses. But it also adds a small amount of background noise to the parts of the spectrum that licensees are using. (DSS looks like undifferentiated hiss to anyone who doesn't have a rake receiver with a properly synchronized Pseudo-Noise (PN) generator.)

As long as the total bandwidth used by the wideband is small (e.g. 1% or less of the carrier bandwidth), the amount of interference it creates is negligible. But if the wideband uses a lot of bandwidth (10%), then that interference becomes significant and interferes with the ability of narrow-band licensees to use the parts of the spectrum they are entitled to.

HOWEVER...

It used to be the rule that you could transmit at up to 50 milliwatts without any license. And because of the ferocious coding gain that wideband offers, it might well be possible to use it practically in an area the size of a port district at power levels of 50 milliwatts.

CDMA can carry a mile with a 50 milliwatt signal and that's with just 2^7 chips per bit. If you had 2^12 chips per bit or even more, it might carry a hell of a lot further.

What I don't know is whether the FCC has revised, or will revise, that 50 milliwatt rule if this kind of thing becomes more wide spread. (As it were.)
posted by Steven C. Den Beste at 12:42 PM on September 21, 2006

Thanks all for your input so far.

dammitjim - I actually am very familiar with the RFID capabilities (both passive and active) and the typical software solutions that are implemented with such systems (I've spoken with a number of vendors).

My issue is with spot-level accuracy. Apparently most RFID will only get you spot-level accuracy down to the 3-7meter level (ie. the system will tell you that tag is in a certain spot, and the tag will physically be within 3 to 7 meters of that spot). Most RSSI, as I understand it, is within 5-10 meter accuracy. GPS even less accurate, at 8-12m.

What I am looking for is a solution that guarnatees accuracy down to 3m or less - as I understand it, only (the possibility of) UWB could currently do this.
posted by allkindsoftime at 1:41 PM on September 21, 2006

A UWB solution must be powered. RFID systems are very low energy, and don't have to have batteries. The detector hits the RFID with a big electromagnetic pulse, and the RFID unit uses energy from that to respond.

There's absolutely no way a UWB system could be made to do that; the electronics simply is too complicated to be able to operate off that miniscule an amount of energy for long enough to do what you want it to do. So UWB units would all have to have batteries -- which, from the sound of it, would disqualify it from consideration for your purposes. (Do you really want to have someone walk around the yard once per week to replace batteries in every single box car?)

Also, a UWB unit will be expensive. In large quantities I'd think they'd cost about $100 each, or maybe even more. (In small quantities the price would be a hell of a lot higher.) UWB is going to require at least two ASICs (one digital, one RF), neither of which is going to be easy to design.

DSS is inherently tricky and difficult, and UWB makes it even more difficult. In IS-95 CDMA the chip rate was 1.2288 MHz. In CDMA2K it's 3 times that. In a UWB system it's going to be much higher, and making that work is nontrivial. (Making that work even at the speeds CDMA2K and UMTS run is very tricky, as the Europeans have been finding out.)

There are some other technical issues which I can go into if you want me to get nerdy about why any kind of DSS system, including UWB, would be problematic (e.g. the problem of synchronizing the PN generators) but I think the point has been made.
posted by Steven C. Den Beste at 2:23 PM on September 21, 2006

By the way, it's not logically required that UWB use DSS, but every proposal relating to UWB I've heard of does use DSS.

The reason is that if you rely on something like simple FM with a UWB carrier, the noise level is prohibitive because of all the narrow band users living in the same spectrum space.

You need DSS to make it work, because DSS is designed to handle a huge noise floor.
posted by Steven C. Den Beste at 2:27 PM on September 21, 2006

One more post: In many kinds of RF location systems, what they do is to measure the signal amplitude and calculate the distance based on how much the signal has decayed. For a number of reasons that's has inherent errors since there can be other reasons besides distance why the signal amplitude varies.

You can also do a challenge-response and measure the response time, but that's more tricky because the reaction time of the remote has to be absolutely consistent, and that's really tough to do.

In a DSS system you measure range by calculating the PN phase delay. Signal amplitude doesn't matter. In IS-95 CDMA (which I used to work on) we had to synchronize our PN generator to a 16th of a chip. The cell did, too, which meant that it implicitly knew the RF round-trip time accurate to 51 nanoseconds, about 15 meters.

The problem is that the RF path is rarely a straight line, and the signal to the phone isn't always the same as the path back to the cell. So when the FCC forced all the cell systems to implement a remote position sensing system (for 911), the only real way to do it was to put GPS receivers in the phones.

In a more limited and controlled area (like a switching yard) where the fixed transceivers operate from the tops of towers, then the RF paths will be much more controlled since there won't be buildings and hills and aircraft in the way. In order to make it cheaper, the system probably doesn't synchronize as accurately, say to a quarter of a chip (much easier) instead. So you can get positioning pretty much as accurate as you want it to be as a function of how wide a carrier (and thus the chip rate) that you use.
posted by Steven C. Den Beste at 2:45 PM on September 21, 2006

Steven -

Thanks for the info.

Actually most of the RFID systems that I've obtained information on do work off of powered tags. The battery life is typically dependent on a customizable ping-rate for active RFID tags - if they are set to ping the system every 10 seconds = a 2 year life, every 30 seconds = a 6 year life, etc.. Passive RFID systems the battery life is potentially even longer than 6 years. These batteries are usually small (ie. equiv of 2AA's), and with the number of trailers I'm looking to tag, the cost would be fairly neglible - I assume around $50 a tag.

So I assume I would be powering a UWB solution.

And I definitely assume its possible, because I found this just now.

I must confess that I'm not huge on the technical stuff - I work in a more functional aspect for my company but because of my dabblings I was tasked to look into this.

Can you explain to me the difference between "IS-95 CDMA" and "CDMA2K" and define what the chip rate is? I know that CDMA is the cellular network used in North America for cellular connections, but that's about it.

Also, I'm not familiar with DSS and how / why it plays with UWB.

Appreciate any further definition you can provide me with.
posted by allkindsoftime at 4:20 PM on September 21, 2006

You done it now, boy... (heh)

IS-95 is the "2G" CDMA standard which was used by Verizon and Sprint. CDMA2K is the "3G" upgrade which they're both deploying now. "IS-95" refers to a particular ISO standard which fully describes the system. It was designed to work with 1.25 MHz carriers.

CDMA2K uses a 3.75 MHz carrier. UMTS (the 3G replacement for GSM) is slightly different in order to be incompatible. (I'm not kidding.)

A "chip" is a piece of a bit. The way DSS works, a single carrier (i.e. the cell) can theoretically carry up to 64 simultaneous and independent channels. Three of those are reserved for system functions (the pilot, the sync channel, and the paging channel) and the other 61 can be used for traffic channels. In practice it isn't possible to carry that many traffic channels simultaneously, but the number that can be carried varies as a function of conditions, which is why CDMA is said to have "soft capacity".

Anyway, in IS-95 the carrier carries 1,228,800 chips per second even if only 9600 bits per second are being transmitted. The receiver takes groups of chips and processes them to rederive the bits they're intended to carry. In IS-95 usually 128 chips are used to transmit a bit.

Why? Because the same 128 chips can transmit different and independent bits to multiple receivers at the same time -- and that's where the magic happens.

How? I once posted a visual analogy of it here. If you want to see my attempt to explain the real thing, it's here. But either you'll get an "Aha!" from it, or you'll be completely bewildered.
posted by Steven C. Den Beste at 6:13 PM on September 21, 2006 [1 favorite has favorites]

DSS = Digital Spread Spectrum, being the algorithmic foundation for most UWB low power techology systems, whereby you trade power and frequency exclusivity for full spectrum use, at low power, trading on the fact that interference will not typically appear simultaneously across the entire RF spectrum.

The FCC typically takes the common sense stand that an unlicensed use of spectrum is permissible, so long as it causes no interference to other licensed uses, and if served by RF emitting equipment, that the equipment used is type accepted or certificated, and unmodified. Some spectrum ranges (frequency bands) are explicitly reserved for unlicensed operation of type accepted or certificated equipment, mostly to harmonize with international agreements for services where significant quantities of equipment are manufactured and shipped internationally, for short distance services employed by the general public. Microwave ovens, WiFi data networks, citizens band radio, and similar uses in the ISM bands are examples of this.

But even in licensed bands, if your unlicensed use causes no interference to licensed services, no one is going to initiate enforcement proceedings. I used to follow up interference reports for viewers when I worked for a GE Broadcasting affiliate, and I'm quite familiar with the FCC complaint process. A CB operator running an illegal (because it was never type accepted for CB use) but clean 1 KW linear amp wouldn't raise my eyebrows or the hackles of our viewers in fringe areas, whereas a guy with a badly souped up 10 watt CB rig in an urban area could expect to use his toy about 2 or 3 days on average, before being visited. The guy with the clean 1 KW linear might get complaints from other CB'ers he was stepping all over, but they pursued their claims on a different footing, being unlicensed users of a reserved public service band.

UWB applications are under technical review by the FCC for a variety of uses, and some early commercial systems have been deployed in fairly demanding environments. You might contact Parco Wireless with your requests.
posted by paulsc at 6:19 PM on September 21, 2006

DSS stands for "Direct Sequence Spread Spectrum". The third "S" is often omitted from the acronym just to keep it manageable.
posted by Steven C. Den Beste at 8:53 PM on September 21, 2006

There is a lot of confusion about some terms here. Digital spread spectrum (DSS) is a generic term that applies to many wide bandwidth techniques. The two most important are frequency hopping spread spectrum (FHSS) and direct sequence spread spectrum (DSSS). Steven has been talking mostly about DSSS and in particular the Qualcomm proprietary scheme called CDMA which is used for cell phones. Neither of these have anything to do with ultra wide band (UWB).

UWB is a completely new spread spectrum communications method that the FCC has assigned the 3.1GHz to 10.6GHz band. There are two basic UWB techniques -- impulse and multi-channel. The UWB RFID linked by Paul uses the impulse technique. UWB is new enough that there are no industry wide protocol standards yet. Each company is developing its own techniques. It is way beyond just the research stage. The FCC approved commercial applications back in 2002.
posted by JackFlash at 12:23 AM on September 22, 2006

I was not aware of those things. All the proposals for UWB I'd heard of in the past had been based on DSS, and were proposed to lay on top of existing licenses.
posted by Steven C. Den Beste at 1:48 AM on September 22, 2006