Slashdot Mirror
LightSquared Disrupts 75% of GPS Connections In Government Test
Freddybear writes with this quote from BusinessWeek:
"Philip Falcone's proposed LightSquared Inc. wireless service caused interference to 75 percent of global-positioning system receivers examined in a U.S. government test, according to a draft summary of results. ... The tests worked off an 'extraordinarily conservative' threshold and didn't show the devices' performance was affected, [LightSquared exec Martin Harriman said]. 'If we're affecting the performance of the device — my goodness, we'd like to be sure that doesn't happen,' Harriman said. The laboratory testing was performed for the National Space-Based Positioning, Navigation, and Timing Systems Engineering Forum, an executive branch body that helps advise policy makers on issues around GPS. It found that 69 of 92, or 75 percent, of receivers tested 'experienced harmful interference' at the equivalent of 100 meters (109 yards) from a LightSquared base station."
This came out during the week, but was overshadowed by the news that Falcone And Friends got Wells Letters, SEC notices that are basically game-over. Investors in Harbinger Capital Partners, Falcone's hedge fund, are likely to flee, but they'll be limited in their ability to withdraw funds. This has happened before to Harbinger in 2009, and Goldman Sachs seems to have gotten preferential treatment in exiting.
The LightSquared bit is juicier, though, because of the hints of corruption that have squeaked out through the press. Air Force General William Shelton, testifying before Congress about LightSquared and the interference that its plans could cause GPS, complained that the White House had told him to change his testimony to make it seem that he was less opposed to LightSquared's plans. There are also allegations of $30,400 donations being given to the Democratic Party by Falcone and LightSquared's CEO on the days of meetings and on days when meetings were arranged.
Yes, but the towers are broadcasting signals that are orders of magnitude more powerful to ground receivers than the gps satellites. If Lightspeed was a satellite phone system (so if it was another satellite system producing the crosstalk), it would not interfere even if the frequencies were directly adjacent.
It's not so much that the nearby frequencies need to be silent. I believe it was actually the case that those frequencies were originally licensed for low-power signals. So when the hardware engineer was designing his GPS receiver circuit, he would use the expected max power that could be licensed for that band in his calculation for determining how many -dB/Octave his filter needs. Now LS comes in and wants to relicense that spectrum for signals of many orders of magnitude more power; the circuits were simply not designed to handle this because such signals were illegal at the time of manufacture.
It's not the hardware engineer's fault that the adjacent bands were "zoned" to be residential and now LS wants to come in and build an airport.
:(){
Nope, it's not adjacent channel interference -- the precision positioning people use a correction signal (from a satellite) in L band, below the GPS L1 signal, that is completely swamped by the LightSquared system -- these precision positioning systems, which are also used in highway construction and other large developments as well as in large scale agriculture, are among the systems that the study identified as impossible to make work even with a redesign -- these bands were meant for weak signal reception of satellite signals, NOT for multi-kilowatt ground stations.
And when you talk about adjacent channel, remember that GPS boxes aren't so much receivers as correlators -- and they are working with signals that are effectively below the noise floor -- that's why correlation techniques have to be used. What might be acceptable as adjacent channel in other modes is devastating to correlator-based designs.
See for example the FAA report at:
http://scpnt.stanford.edu/pnt/PNT11/2011_presentation_files/09_Bunce-PNT2011.pdf
Look at the docs posted for the recent symposium at Stanford:
http://scpnt.stanford.edu/pnt/
Opening comments on how LightSquared destroys GPS:
http://scpnt.stanford.edu/pnt/PNT11/2011_presentation_files/01_Parkinson-PNT2011.pdf
the FAA report on testing:
http://scpnt.stanford.edu/pnt/PNT11/2011_presentation_files/09_Bunce-PNT2011.pdf
The LightSquared idea is a good one, but not on the frequencies they've selected!
I'm an electrical engineer. I did my doctorate in a GPS lab working on safety-of-life applications (landing planes and such). The LS issue has been a very hot topic of discussion in the technical community for most of the past year. At the annual ION GNSS conference this past September, there was a panel discussion on the preliminary test results described in TFA. Out of approximately 600 people in the room, there were exactly two who expressed opinions supporting LS's contention that the interference to GPS would be insignificant: one guy was the LS General Counsel, and the other was a guy who is claiming he has retrofit kits (RF notch filters) that will eliminate the interference. It should be noted that, when asked how his kits would be fitted to the millions of GPS receivers already in the field, the latter person had absolutely no answer.
It is not at all a stretch to say that very nearly 100% of the people who have done LS testing, or evaluated the results from an engineering perspective, conclude that the effects as proposed will be somewhere between "significant" and "catastrophic."
--Jake
The nearest GPS freq is 1575.42 MHz but is the L1 freq
Very close but not quite. The L1C signal is not a simple continuous carrier like the old transit sats from the 50s/60s. The data rate is somewhere around ten megabits and it modulation is BPSK. The exact answer requires more detail but the actual transmitted BW will end up maybe 10 megs higher and 10 megs lower than the center theoretical carrier. Which is getting uncomfortably close to the lightsquared signal.
So... that's 1565 or so, vs the interference at 1559. So you head over to minicircuits.com (a seller of many microwave components, including the high pass filter you are trying to purchase) and look for a coaxial filter with a curve showing almost 0 dB attenuation at 1565 and up to keep your noise figure usable, and at least 60 dB out of band attenuation at 1559. Then you realize why the EE types claim "its a law of physics" that this simply cannot be worked around. Oh and note the ones that don't even come close to making the grade are roughly the size weight and cost of a very small cell phone. Generically building a filter in that frequency range with those specs is impossible, but building the device to that exact frequency spec and stable over any temperature range makes it even more impossible.
Before the sorta knowledgeable DSP types jump in, yes, you can get filter curves like that using DSP. However you need a analog input clean enough to do the DSP on it... So, again, you're screwed. Just plug your 60 dBm 3rd order IMD preamp into your 32 bit A/D 10 GHz A/D converter and then process it. This is technobable of the finest level, components with specs like that Might exist in just 50 years or so, but they sure don't now.
"Science flies us to the moon. Religion flies us into buildings." - Victor Stenger
Run the numbers on the Q factor required and the maximum possible passband attenuation to keep the noise figure of the front end usable... If you know what "snake oil" is WRT crypto it sounds like this guy's offering sounds suspiciously like "frequency grease" WRT RF.
Note that if the problem is front end overload, his snakeoil/freqgrease might be a simple 10 dB attenuator, probably being sold at an immense markup. If would be easier to duct tape aluminum foil to the existing antenna until the incoming signals are knocked down enough that the FE is not overloading but optimistically there is still enough RF signal left to decode.
This is assuming its not at the RF technology level of those stickers you put on cell phones to magically do things that sound good.
I am intimately familiar with the RF arguments. The proposed notch filter is indeed snake oil, but not for reasons of insufficiently steep stopband rolloff. Rather, it's a relatively bulky thing which will work just fine for the receivers produced by the guy's company (Javad), and maybe even other receivers that could be retrofitted. But it is totally unusable for most embedded receivers (handhelds, etc.) due to size (and cost), and there are a lot more of those deployed in the world. The "snake oil" part of the argument is that he is being spectacularly disingenuous about it: when asked how he intends to retrofit every TomTom, Garmin, GPS-enabled wristwatch, and mobile phone already out there in the field, he just waves his hands and says, "those devices won't be affected."
While I'm on the subject, people seem to be unaware of a further bit of deception on LS's part. Their initial proposal included two bands just below GPS (the so-called "Low 10" and "High 10"). When testing showed that the resulting interference would make the proposal a non-starter, LS submitted a modified proposal in which they would only the lower of the two bands (farther away from GPS), and at a lower broadcast power level. The thing is, LS never stated that this was their intended final configuration. Indeed, upon further discussion, it emerged that this revised proposal was intended only to placate objections in the short term, and that LS fully intends to use both bands and the higher power levels (as in their original proposal) eventually. In other words, the potential interference problem was never addressed, just kicked down the road a bit.
In short, the proposed filter was far more sophisticated than the bits of aluminum foil you might see for sale on QVC on channel 179 at three in the morning... but it's virtually meaningless in any practical sense.