Cellphone 'radar' tracks traffic flow

New Scientist ^ | Oct 3 | Staff

[Diddley]

Signals from cellphone masts can be used to track aircraft, monitor traffic congestion and spot speeding motorists without tipping them off that they are being watched.

The radar-like system, which is still being developed, has provoked media reports of the start of a huge extension of Big Brother-style surveillance - privacy campaigners have complained that it could be used to track individual people. But radar experts say such fears are unfounded.

Conventional radar works by transmitting a signal, listening for the reflection and using the time taken for the round trip to work out the object's distance. More sophisticated systems can work out the object's speed from characteristic changes to the signal's frequency, known as Doppler shifts. But such radar systems are expensive, and the signals they send out are easy to detect.

An alternative technique, called passive radar, gets round these problems. Instead of broadcasting its own signal, a passive radar system listens in to the cacophony of radio signals in the environment and monitors the way moving objects change them. The US defence company Lockheed Martin is developing a system called Silent Sentry which exploits the signals from radio and television masts to spot aircraft and ships (New Scientist print edition, 4 December 1999).

Better accuracy Now two British companies, Hampshire-based Roke Manor Research and the aerospace giant BAe Systems, have done the same thing with signals from cellphone masts. They say their system, known as Celldar, short for cellphone radar, can achieve better accuracy because cellphone masts are far more widespread than television and radio transmitters.

Celldar works out the position of objects in the area by comparing the signals reflected from them with those it receives directly from a base station, whose positions are known. From the Doppler shift in the signal it can also calculate the target object's speed.

Celldar has a number of advantages over conventional radar, says David Salter, a member of the team at Roke Manor Research. "The expensive part in most radar systems is the transmitter, because of the high power requirements." Because Celldar devices do not need their own transmitter, they can be made cheaper, smaller and more portable.

Roke Manor Research is currently testing a prototype system, and says it will be two to three years before a fully operational Celldar goes on sale.

Stealthy shadows Because the system is passive, drivers will have no way of telling whether they are being monitored. It is this characteristic that makes passive systems so attractive to the military, says David Bebbington, a radar expert at the University of Essex in Colchester, UK.

Another advantage of passive systems is their ability to spot "stealthy" aircraft and ships, which are designed to fool conventional radar systems by absorbing signals or reflecting them away from the source. To passive radar, these objects show up as shadows that can be spotted.

Civil liberties groups are concerned that the system could be adapted or combined with other technologies to produce a device for tracking people. "I can see profoundly worrying aspects to the technology," says Simon Davies, director of Privacy International in London.

A document on Roke Manor Research's own website has fuelled speculation that the technology could be used in this way, stating that it "can detect vehicles and even human beings at militarily useful ranges". But Bebbington points out that Celldar will be virtually useless for following individuals because its resolution is simply not good enough. And Roke Manor Research now says the information on its website will be removed

[Diddley]

Big brother?

[lelio]

Signals from cellphone masts can be used to track aircraft, monitor traffic congestion and spot speeding motorists without tipping them off that they are being watched.

Huh, I thought of this idea several years ago (I guess I should of applied for a government grant!). If you can figure out from a signal what phone it is coming from, all you need is a single receiver mounted by the side of the road that notes when the signal from the phone was at maximum power. That's most likely when it was closest to your receiver.

From that you can tell how many cars go by in a certain time frame, which is really all you need to monitor traffic flow. Put a couple of more sensors on the side of the road to get a more accurate picture. Granted the phone can change its signal's power during the call but you can get an idea of the number of cars passing by.

[Diddley]

Sounds to me as if you had it nailed! :-)

[Boot Hill]

From the article:   "From the Doppler shift in the signal it can also calculate the target object's speed."

This story is pure BS and here's why.

An automobile traveling at 65 mph will only produce a 90 Hz Doppler shift in a cellphone frequency of 900 MHz. Even if their "Celldar" instrument was measuring the frequency shift of an unmodulated cellphone transmission (and that will never be the case), the primary frequency accuracy of cellphones is, at best, only about 1 ppm or +/- 900 Hz (most cellphones actually have only 2.5 ppm accuracy). In other words, the 90 Hz of velocity information from the cellphone signal, will be drowned out in a sea of noise from the +/- 900 Hz variation due to primary frequency inaccuracy.

Worse yet, we haven't even begun to add in the additional noise from cellphone frequency instability due to temperature, battery voltage, vibration, etc. Nor have we considered the difficulty of accurately resolving the cellphone frequency to within +/- 1.5 Hz (+/- 1 mph) while the carrier frequency is being modulated (changed) by the voice communication on the phone.

To overcome these major "aw shucks" in their plan, they probably will attempt to use differential Doppler detection, but that would require the installation of multiple receivers in and around the cell tower. In differential detection, each receiver would first acquire the cellphone frequency and share that data with the other receivers so that a difference frequency could be determined.

In this scenario, the precise cellphone carrier frequency is no longer important, only the difference in frequency received by two Celldar systems. Moreover, the difference frequency will be 2X that from a single receiver (or about 180 Hz for 65 mph).

But here's the hole in that method. Each receiver must be able to resolve the cellphone frequency down to +/- 1.5 Hz (out of 900 MHz) in order to get any accuracy that a court would accept (+/- 1 mph) and that just is not possible to do with a signal whose frequency is constantly and rapidly changing because of the encoded voice (or data) communication.

Another problem is that whether you use the Celldar system in either single or differential mode, the data is only accurate if the vehicle is driving straight at, or away from the receiver. Any other path would result in an decrease in signal at a cosine rate.

And lastly, the current advent of third generation cellphones using the CDMA frequency-hopping, spread spectrum technology, will put the final nail in the coffin of the Celldar idea. In summary, while this proposal is interesting, it would be too expensive, require too many installation sites beyond each cell station and would not provide enough accuracy or resolution to convince a court to take the data seriously.

Thanks to its shortcomings, this technology offers no big-brother threat, but it does provide a welcome and humorous respite from life's real problems.

Regards,

Boot Hill

[snopercod]

I think this article should have clarified that cell towers can only track cell phones, not all moving objects. At least that's what I belived it means.

As part of the Telecommunications Act of 1996, cell phone companies were directed to figure out a way to locate the position of individual cell phone users, and so they did, somewhat.

They use propagation delay and triangulation from two or more towers to figure out where you are broadcasting from.

They can tell how fast traffic is moving simply by monitoring a randomly-selected cell phone on the freeway.

[Diddley]

It's good to have input from a knowledgeable person in the field.

Thanks.

[Boot Hill]

What caught my attention and what caused my lengthy reply is that I smelled "tech-scam" written all over this article. I sure would be interested in just who is funding Roke Manor Research and their work on "Celldar"!

--Boot Hill

[Diddley]

[7]
Sorry for my late reply.

You do bring up some good points.

Diddley

[j_tull]

cell phone companies were directed to figure out a way to locate the position of individual cell phone users, and so they did, somewhat.

They've done it more than somewhat by embedding GPS receivers in the new cell phones. While my Motorola T720(POS) has an option to "turn it off" who really knows?

[_Jim]

Signals from cellphone masts

NOTHING comes off/out of 'the mast'.

One can expect quite a bit, however, to be seen coming out of the antennas found facing (usually) three different 120 degree sectors ...

[_Jim]

I think this article should have clarified that cell towers can only track cell phones, not all moving objects.

'fraid the article *is* talking about using the signals emanating from BS (base stations) as the signal by which position and speed of objects is deduced ...

[_Jim]

the primary frequency accuracy of cellphones is, at best, only about 1 ppm or +/- 900 Hz (most cellphones actually have only 2.5 ppm accuracy). In other words, the 90 Hz of velocity information from the cellphone signal, will be drowned out in a sea of noise from the +/- 900 Hz variation due to primary frequency inaccuracy.

Absolute accuracy/control of the signal being used is not necessary; this can be mesured (if needed) and used in any calcs (if required) ...

I dare say the purity of a 'cell phone' emission is several orders cleaner/more controlled than the simple free-running 'Gunn' diodes used in X-band Police speed radars I'm familiar with ...

[_Jim]

the current advent of third generation cellphones using the CDMA frequency-hopping

It's either Code-Division Multiple Access "direct-sequence spread spectrum" or it's "frequency-hopping packet transmission" - not both ...

The key: look at the 'IF' filter BW. The direct sequence spread-spectrum system will have the w-i-d-e IF - roughly the width of the 'chipping' rate.

[_Jim]

One would gain some appreciation for what these folks are attempting to achieve if they had some knowldge of the earliest, primitive use of broad-beam, illuminating sources with separate receivers -

- as used in the British Chain Home system. Though not a Doppler-based system, the elementary principles of RADAR are utilized in this system.

[_Jim]

Home page: www.roke.co.uk/sensors/stealth/celldar.asp

Concept: www.roke.co.uk/sensors/stealth/cell_phone_radar_concept.asp

Field Trials www.roke.co.uk/sensors/stealth/celldar_trials.asp

Initial CELLDAR Trials

The first CELLDAR trials were conducted back in November 2001. It was from the success of these trials that a CELLDAR development programme was started. During these trials both aircraft and cars were successfully detected and tracked.

CELLDAR Trials Equipment

Equipment                   Comments

2 x Siemens GSM Phones      Acting as receivers for the base station 
                            signal and the reflected signal from the target.

2 x Yagi aerials

1 x PC                      Standard desktop machine.

AD Converters               200KHz AD converters to digitise signals receive from the GSM phones.

[Boot Hill]

Hi _Jim, its been a while since I've heard from you. How are you doing?

_Jim says:   "Absolute accuracy/control of the signal being used is not necessary; this can be mesured"...

Every cellphone in use will have (at least) a +/- 900 MHz variation in the absolute carrier frequency of whatever channel that cellphone happens to be transmitting on at any given moment. To do as you suggest and measure that absolute frequency would require there be no velocity induced Doppler shift of that frequency. How do you propose to ensure that your measurement is done at zero velocity?

If you mean to obtain that data as part of a factory calibration, then you'll find that the frequency instabilities (not just the primary accuracy), from such sources as changes in temperature, battery voltage, etc., will render that factory calibration meaningless once the customer began using the cellphone under vastly different conditions. Frequency "push" can create as much error as the primary frequency inaccuracies, thus swamping out any Doppler information.

_Jim says:   "I dare say the purity of a 'cell phone' emission is several orders cleaner/more controlled than the simple free-running 'Gunn' diodes used in X-band Police speed radars I'm familiar with... "

Yes, but keeping in mind that in the case of "Celldar", spectral purity of the cellphone carrier is not a big show stopper, I would offer the following observations...

Your are correct about the Gunn diodes and for good reason. Gunn diode currents are notoriously noisy and that noise is reflected in lots of phase shift in the transmitted signal. Moreover, the resonant cavity associated with the Gunn diode has a relatively low Q (when compared to the Q of crystal oscillators). The result is a poor spectral purity and lots of power in the sidebands.

However, cellphones have their own Achilles' Heel. All makes and models of cellphones use a PLL to generate and control the transmitted frequency and the primary frequency generating circuit of the PLL is the VCO, which have notoriously low Q's and consequently poor spectral purity. (But as you say, not nearly as bad as Gunn Diode oscillators.)

While spectral purity is not a big deal, in the case of "Celldar", measurement of the carrier frequency to extreme resolution (i.e., +/- 1.5 Hz out of 900 MHz) is a big deal, especially when the frequency you are trying to measure is being FM, FSK or PSK modulated, as it is in cellphone transmissions.

_Jim says:   "It's either Code-Division Multiple Access "direct-sequence spread spectrum" or it's "frequency-hopping packet transmission" - not both ..."

You are correct that there are two implementations of CDMA. However, the direct sequence implementation (DS-CDMA) and the frequency hopping implementation (FH-CDMA) are are both true spread spectrum techniques, so it would be incorrect to suggest that you can't have both spread spectrum and frequency hopping.   [SOURCE]

Regards,

Boot Hill

[Boot Hill]

Oops,

"Your are correct about the Gunn diodes..."

should have been...

You are correct about the Gunn diodes..."

--Boot

[snopercod]

I have a vague familiarity with how J-STRS works. You have that 100' tall tower with the radar source on it, and the aircraft up above looking for reflected signals.

But the J-STRS antenna/receivers are super-sensitive.

It's hard to understand how those crappy little cell antennae and off-the-shelf receivers could be sensitive enought to track a Honda Civic at 10 miles from reflected signal only.

Furthermore, the ground clutter - reflections from nearby metal buildings, billboards, transmission towers, etc. - would almost certainly overwhelm the tiny reflections from cars on the ground.

Heck, what do I know?

[KantianBurke]

- shrugs shoulders - not using a cell phone at all should do the trick. Never understood the need for one. If the calls important they'll leave a message on your regular machine.

[snopercod]

2 x Siemens GSM Phones - Acting as receivers for the base station signal and the reflected signal from the target.

This sheds new light on the technique. Apparently, they are using a mobile station (or maybe I should say a remote station) - not the base station - to track moving objects.

So if you wanted to monitor freeway traffic, you would set up these remote stations along-side the freeway at appropriate intervals.

That makes more sense to me. It would place most of the ground clutter returns after the target returns, which could be just cut off with an appropriately-timed gate window.