xG Technologies -- Physics or Snake Oil ?

James Enck suggests, in this EuroTelcoblog post, that xG Technology may be up for sale. 

If you haven’t heard, xG is a small Florida company that hit the tech press several times in 2005 with claims for a new wireless technology providing 40 Mbps over a 15–20 mile radius using less than 1 watt of transmit power. 

At the time, I didn’t pay much attention but recently I had a discussion with a friend, whose RF engineering expertise I respect, who has met the founder of xG Technology, Joe Bobier, and examined xG’s technology on behalf of some potential investors.  Of course this was done under NDA, so I couldn’t get any details, but my friend did say,

…after taking some time to understand what they were doing I realized they actually have a pretty novel technology.  … it was refreshing to see what an outsider (the inventor is a long time ham radio operator w/ a high-school degree) with a different perspective can do.

This led me to do a little research.  A quick search at the US patent office using “Bobier; Joseph” for “Inventor Name” shows nine pending applications at least six of which discuss modulation and a number of issued patents at least two (US 6,901,246 & US 6,968,014) of which describe modulation schemes.  I haven’t read them all in detail — there’s a lot of repetition between applications and, as is usual with patents, the language is extremely boring — but I did go over two of them fairly carefully.

In each case, he uses a sine wave carrier which gets modulated on a per-cycle basis with changes occurring only at carrier wave zero-crossings.  In US # 6,901,246, he changes the amplitude of the carrier wave on a cycle by cycle basis.  That’s just equivalent to multiplying the carrier by a binary data stream.  The resulting spectrum will have a line at the carrier frequency surrounded by side lobes based on the spectrum of the binary data — for NRZ binary data the side lobes would follow a sinc function, like this.

That’s a lot of side-lobe energy and thus not very useful.  Of course that was his first filing, in 2001.  In subsequent applications he gets more and more creative with the proposed modulations and, since the patent office only posts patent applications 18 months after the filing date, we don’t know everything he’s got in the pipeline.

What interesting, and implied by Q3 in their current FAQ, is the idea of combining a spread spectrum signal, i.e. a low level very broad spectrum, with a pronounced carrier frequency.  Recently enacted FCC rules for Ultra Wide-band (UWB) permit intentional radiation at a very low level, but only in the range 3.1 GHz to 10.6 GHz.  At lower frequencies, where radio signals are subject to less attenuation and thus can cover more distance, part 15 rules govern low level unintentional radiation.  But part 15 rules do not allow intentional radiation, however low level.

Based on these spectrum plots on the xG web site, they have come up with a legal way to transmit very low amplitude spread across a wide swath of valuable spectrum (all well below 1 GHz).  They require a narrow licensed channel for their carrier, while their real information is in the highly attenuated but very broad sidebands.  As an additional benefit, they avoid the otherwise complicated clock recovery system required in most spread spectrum systems by transmitting a solid clock signal — their carrier signal at the center of their narrow, typically licensed, band.

So they get the benefit of operating at frequencies where signal propagation is better and using a very broad swath of spectrum, while reducing the complexity of their receivers by providing a strong clock signal and conforming to current FCC regulations.

If true, it sounds like a good deal to me.

 

Comments

[a friend]

I have been following this technology for months in many forums and webs, it is a clever idea to achieve a communication channel putting low level spreaded spectrum signals around a big carrier. But nobody say the real truth, that is, if you want to carry out another not interfering channel, the distance between carriers will be at least the used bandwidth not the carrier bandwidth, so reducing the efficiency of a feasible cellular system which needs, at least, three frequencies to be operative. In fact, thinking about spectral efficiency (Bits divided by the real ocupied bandwidth, not the 3dB carrier bandwidth) it is just like any other spread spectrum system, very few bits/s/Hz. Moreover, They say a DSL application should improve actual systems. I suspect that it will be imposible because this system should maintain power transmission masks (at most -40 dBm/Hz) in the carrier and then the spreaded spectrum information should be so power lowered and distorted by the cable channel that the system would not improve actual systems at all. Nevertheless I think there are many applications for it.

[brough]

Dear friend, Thanks for the comment. I don't know if xG is for real or not, but the idea of spread spectrum is real and a little different than comes across in your comment. Forget about xG's carrier for a moment and just talk spread spectrum, e.g. CDMA or UWB.

It's certainly true with Frequency Division Multiple Access (FDMA) that adjacent carriers for separate communications channels can interfere. However with Code Division Multiple Access (CDMA) and Ultra-Wide Band (UWB) all of the communications channels share the same spectrum. They are distinguished by coded modulation while sharing the same spectrum.

The metric "bps per Hz" is appropriate for FDMA where communications channels are separated by frequency, i.e. in Hz. It's not a useful metric for spread spectrum where communications channels share the same frequencies.

I was speculating that xG transmits the useful information using spread spectrum and that their carrier is only present to simplify clock-recovery at the receiver (thus reducing the cost of the receiver). IF this is true, i.e. if they are using Ultra Wide Band or another form of spread spectrum modulation across the entire spectrum below 1 GHz, then they have at least one advantage over FCC-approved UWB schemes -- radio waves below 1GHz carry much further than those above 3 GHz. The question then would be how much total power (across many, many MHz)?

I don't know if xG is for real, but I like the idea that by licensing 5 KHz of spectrum, they get formal permission to spread low-level signals through the rest of the spectrum. That beats the FCC-approved UWB which is only allowed to spread it's low-level signals between 3 GHz and 10 GHz.

Thanks,
- rbt

[saal]

Sir, Pls give me an explanation about "single cycle modultion
technique" which is used in xMax technology.

[brough]

The term single-cycle modulation appears to have been coined by Joe Bobier of xG Technologies. As I understand it, from reading his patent applications, he is manipulating individual cycles of his carrier signal. For example, one patent application talks about flipping the polarity of a single cycle of the carrier signal, i.e. flipping the phase by 180 degrees. This is done to individual, perhaps widely spaced, cycles of the carrier wave in response to changes in the modulating signal, i.e. the information signal.

[Dan Plesse]

As far as I know xG Technologies, Inc. is traded on the London Stock Exchange! xG Technologies, Inc. is traded on the London Stock Exchange's Alternative Investment Market (AIM), a global market for growing companies. The stock is listed under the symbol XGT.L I am hoping to report news about them via opticalalert.com my newsletter. XGT.L is the ticker for yahoo.

http://finance.yahoo.com/q?s=XGT.L

[brough]

Yes, they went public on the London AIM on November 20, 2006, i.e. subsequent to my original post, but possibly with a change in business strategy as well. I haven't followed them in detail, especially as there's been nothing new about their single cycle modulation technology, but people interested in their history might find these URLs relevant:
http://disruptivewireless.blogspot.com/2006/03/vowlan-vo3g-yesterdays-news-try-new.html
http://eurotelcoblog.blogspot.com/2006/09/xg-ipo-cojones-de-acero_27.html

[an engineer]

There is no free lunch. Transmitting a pilot tone in a band far away in frequency from the modulation isn't very useful, because the pilot will not be phase coherent with the modulation - fading channels won't allow it (it's beyond the "coherence bandwidth"). While you might be able to make it work when the terminals are fixed, it sounds impossible in a mobile link. In OFDM, you use pilot tones for precisely the same purpose, and they are much more useful because they are embedded in the same channel with the data tones - which is how WiMax works. Also, in a bi-directional link, the receiver will also have to transmit a pilot, because the transmitter won't be able to determine the phase impairments to it's own pilot over the path.
Sending data using xMax sounds like an ok system for one-way data broadcast to a fixed termial, but won't be so great for bi-directional or mobile. It may have some uses, but isn't going to become "4G."

[Dan Plesse of Opticalalert News]

xG posted a few more data items
January 25, 2007 xG Signs Agreement with UK-based Telecommunications Company
January 26, 2007 xG Confirms Positive Lab Beta Trials Prior to Imminent Field Testing

[Phil Karn]

I've written up some initial analyses of xG's claims on my website. Summary: the inventor clearly does not understand the most basic principles of communications theory. Investors should be *very* wary. See:

www.ka9q.net/xmax.html
www.ka9q.net/tristate.html

[mike burenstein]

xgt it sounds like a good deal to me.