Breaking free of the past

Every RF engineer learns early on that it takes signal power to make wireless information recoverable. It goes without saying that “turning down” the modulation makes it harder to understand the message. That’s why certain axioms develop, such as, “an efficient signal moves as much power from the carrier to the sidebands as possible” and terms like spectral efficiency refer only to the number of bits we can pack into a Hertz of spectrum.

It’s no wonder then that modulation technology today is based upon the same foundations as methods almost a century old. While our preferred embodiments have improved and evolved, the RF engineering concepts remain rooted in the past. XMax is the result of taking a different path; one where the object is to place as little power in the sideband as possible. This clearly seems contradictory to conventional wisdom, so let’s work from the objective forward.

In our case, the objectives were:

* Design a signal profile that could transmit “out of band” with little to no impact on legacy spectrum users. In our case, a “legacy user” is anyone operating any radio service of any type other than xMax.

* Having created such a benign signal, use it where spectrum has efficient propagation characteristics, i.e., spectrum below about 1 GHz, which is well known to have good penetration and better multi-path characteristics, resulting in better range.

* Receive the extraordinarily weak sidebands and recover the information even in the presence of strong interferers.

Immediately, the skeptic will determine that the first two goals are doable, but the third would be a problem. Hopefully, by the end of this narrative, you’ll understand not only that all three goals are doable, but also, in many cases, xMax represents a new category of modulation and demodulation.

Early on in this project, we saw a correlation between two facts. The first is that the most basic element of information is Yes or No, or, in other words, the single bit. The second is that the most basic element of a radio signal is the single photon or single cycle of the radio wave. While seemingly un-related, it occurred to us that if we could cause a single cycle of a continuum of cycles (the carrier) to represent a single bit, interesting technical exploitations could be realized. (Later we found that through coding many bits could be represented.)

We spent many months exploring the various ways single cycles could be accurately modulated. Because of the recent availability of extremely fast logic devices, a simple and accurate method eventually was developed that could operate to about 12 GHz without needing custom silicon. This type of frequency modulation became our standard method.

In this example, the modulation index is 0.125 and the pulse repetition rate is no more than one modulation event in 2048 RF cycles. As you can see, the result is a strong carrier with a low power but broad sideband profile. Left un-filtered, the complete sideband width is about 1.15 times the carrier frequency. This large width might not be desirable or necessary unless the data rate is commensurately high. In many applications, the data rate will be only a few megabits per second and filtering to a width of about one bit/Hz can reduce the bandwidth. In this example, we chose a 10 MHz-wide filter and located the sideband 10 MHz from the carrier Bear in mind that the non-filtered portion of the signal could be anywhere under the 1.15X original curve. This affords the opportunity to strategically place the sideband energy anywhere we choose. This can be useful in frequency planning.

Now that we have an essential understanding of the power spectrum profile, and we understand that the sidebands, weak as they are, can be placed over a wide range and can be as wide as necessary to accommodate the symbol rate, we can look to the reception problem.

Most engineers will point out that reception is the real problem. This is likely why nobody has used single-cycle modulation before. Sidebands that are so weak leave little to recover. Or so we thought.

Here are assumptions that reasonable engineers will make:

* The signal will need a very high power to achieve any range. This is not true. For example, with sideband power of only about 0.5 mw (1/2000th the power of Wi-Fi) and ground-level unity-gain antennas, we can achieve range of more than a mile.

* The information must be spread over a very wide spectrum, like ultrawideband (UWB). This makes a system with processing gain possible. Nor is this true. As we pointed out above, the information can be restricted to bandwidth as little as 1 bit /Hz. Systems such as UWB typically use bandwidth that is proportionally orders of magnitude wider.

* Highly directional antennas must be used, as there is no other way to gather enough power to bring the signal above the noise floor. Not true again. We have successfully used unity-gain omni-directional antennas in our field tests.

* The carrier is somehow used to correlate with the pulse, creating system gain. Again, not true. The system can be operated without the carrier for certain applications, especially those that are battery powered, albeit at a reduced data rate.

The real answer can be found by changing our perspective of how we view the radio signal. If we use any of the traditional reception techniques, we find that we’re really treating the signal as a tone, which must be either processed by analog filters and discriminators or by digital signal processors (DSP), which involves use of the fast fourier transfer (FFT) to detect signal properties.

Both of these time-tested processes require that the information carrying sidebands must exist long enough, i.e., for many RF cycles to form either an FFT result or to stimulate a tuned filter system into resonant oscillation. However, the xMax system produces a sideband for only the duration of one RF cycle, because modulation lasts only one RF cycle; after all it’s single-cycle modulation. Therefore a detection system that requires multiple RF cycles to work can’t detect a single-cycle event. What is needed is a detection system that can process the RF signal on a cycle-by-cycle basis, i.e., a tuned circuit or DSP that relies on an FFT simply doesn’t have time to react. In our experience, little knowledge existed that could point the way.

As so often happens, however, an idea came out of nowhere to devise a remarkably simple circuit that could literally sort RF cycles on a real time basis. While we can’t yet divulge the exact nature of the device, (sorry… we pay lawyers too much money to ignore them) we can give a “black-box” description.

Imagine a black box that has an input and an output. The input is a swath of spectrum that contains dozens or hundreds of strong signals from legacy technology systems, plus a weak signal that is characterized by RF cycles that occasionally change on a single-cycle basis. The box is able to eliminate nearly all signal components that change on some other time basis than the single cycle, while reducing far less any signal that changes only on a single-cycle basis. Thus we’re sorting RF waveforms in real time, based upon whether they change on a single-cycle basis.

The net result might as well be called gain. Legacy signals are nearly eliminated while xMax signals are preserved. The difference or gain can range upwards of 100 decibels (dB), thus making the impossibly weak signal fully recoverable.

So, while the xMax information is very weak, the receiver can operate in what seems to be nearly silent spectrum and the information is recovered.

How weak is the xMax signal?

The signal is on average 74 dB below the carrier. That’s pretty weak. I highlighted average because that’s just what we get when we do an FT on the signal. However, now that we know the signal is both modulated and demodulated on a cycle-by-cycle basis, is it really accurate to understand the signal power on an average basis?

Suppose we could do an FT only for the time of the modulation event itself. We would find that at the instant of modulation, depending on the index of modulation, we actually have quite a strong sideband. However that strong sideband only exists for the briefest of time, essentially one RF cycle. Therein lies the requirement for the real-time cycle-by-cycle detector outlined above. When we consider then that the sideband is intense for a brief period of time and that because the pulse-repetition rate is low, the average power is quite low and we can understand that a special detector that is responsive only to brief but intense sidebands makes sense. The result of all this is a micro-power radio link that works as well as traditional links using perhaps 1000 times the power.

If that doesn’t help, think of the zenon strobe lamp. It’s very bright, but only flashes for a very short time. The average power consumed is quite low, yet the flash can be seen for miles.

Joseph Bobier, is president of operations for xG Technology LLC. Bobier formally trained in electronics and communications technology in the U.S. Navy. Since then he has received several patents in the fields of renewable energy, wireless networking and radio modulation technologies.