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.