The VMSK Delusion

Arthur: You are indeed brave, sir knight, but the fight is mine.
Black Knight: Oh, had enough, eh?
Arthur: Look, you stupid bastard. You've got no arms left!
Black Knight: Yes I have.
Arthur: Look!
Black Knight: Just a flesh wound!
--Monty Python and the Holy Grail

This is a collection of pages about the utterly worthless Very Minimum Shift Keying and related, so-called "ultra narrow band" modulation schemes by Hal R. Walker.

Note added 19 March 2007: After a long hiatus, VMSK has popped up in the news again. Florida company Compress Technologies, Inc, publicly traded on the "pink sheets" as CTLG.PK, says they have acquired the rights to VMSK and are developing several commercial applications including FM SCA.

A story about Compress Technology and VMSK in the broadcast industry publication Radio World attracted this letter from Derek Kumar of Digital Radio Express.

Interestingly, I cannot find any mention on Compress Technologies' website or in their press releases, of any controversy involving VMSK. They do state that an "independent lab", Eagle Comtronics, has "verified" VMSK. (Eagle Comtronics, in Liverpool NY, appears to be a manufacturer of cable TV filters and distribution equipment.) No test reports from Eagle Comtronics are apparently available. One would think that in the Sarbanes-Oxley era, Compress Technologies would disclose the controversy surrounding VMSK if only to say that they believe the criticisms lacks merit.

Note added 26 July 2003: The US Securities and Exchange Commission has filed suit against Alphacom and its principals alleging violations of securities laws. Here are the SEC's news release and complaint.

This action is long overdue. It was especially satisfying to see Snyder specifically named as a defendant after his bogus threats to sue me over these webpages. But it's too bad that the SEC chose only to go after Alphacom just for violating the securities laws and misappropriating shareholder funds. They have not gone after them for hawking a bogus technology that cannot possibly fulfill its claims. Indeed, the complaint is apparently based on Walker's input, and they seem to take his claims at face value.

Notes added 17 Dec 2002: A Google search just found this little gem of a paper: Comments on Spectral Efficiency of VMSK by Saso Tomazic of the University of Ljubljana in Slovenia. Using a different approach, he arrives at the very same conclusion as me: the spectral efficiency claims being made for VMSK are hogwash.

Walker continues to update his web pages occasionally with vague claims of progress. In his Latest Comments page dated 30 November 2002 is the following wonderfully disingenuous comment:

Most critics have recently admitted in private correspondence that it does work, but insist that it is not perfect in every way.
Now he knows that I've never claimed that his hardware couldn't transmit bits from A to B under very carefully controlled conditions, and in that very limited sense it "works". So what? There are many ways to move bits around. Much better ways, in fact. What would make VMSK useful is its claimed ability to carry far more data in the same spectrum than before. And in this much more important sense, far from "not being perfect", VMSK definitely does not work and can never work. All of Walker's claims are based on sloppy labwork, wishful thinking and an utter and willful ignorance of the incontrovertible mathematics of communications.

VMSK is now joined by a veritable alphabet soup of "ultra narrowband" schemes, not all of them Walker's: VPSK, VMSK/1a, VMSK/2a, WPSK, Blip VMSK, 3PRK, FK, MCM, MSB, TSK, etc. Of course, none of them can work any better than VMSK; they all suffer from the same fatal flaw. This is a classic pattern among crackpot inventions: when one doesn't work and the sponsors get restless, the inventor just drops it as "yesterday's news" and starts to sing the praises of his latest new scheme.

And on the business side, some people previously involved in hawking VMSK through International Technical Marketing Ltd have apparently taken over the defunct NASD-registered Nevada "shell" company "BuckTV.com", renamed it to "Multi-Tech International" (symbol MLTI.OB) and purchased the assets of the now-defunct Alphacom, Inc ( AHCM.PK). Now they plan to develop and sell a revolutionary new "bandwidth" scheme that looks suspiciously like VMSK. Walker hotly claims that MLTI has no rights to his patents, and the lawsuits that have been flying between Walker and Alphacom will now undoubtedly spread to MLTI as well.

This seems to be the quite the pattern with crackpot inventions: when demos fail and deliverables aren't delivered, the inventor falls out with the business guy. Pretty soon the two sides are locked in an expensive legal battle over the ownership of a technology that neither side can admit is utterly worthless. As usual, the only winners are the lawyers.

Previous additions

  • The Real Facts of Life (11 June 2001): PDF, LaTex.
  • Answer to Walker's "VMSK Delusion" essay (14 June 2001)
  • Comments by someone who attended the supposedly successful Boston VMSK demo on 22 Sep 2000 (18 Jun 2001)
  • Yet another proof that VMSK's data is entirely in the grass (19 Jun 2001)

    Introduction

    Mr. Walker has invented a family of modulation schemes he calls Very Minimum Shift Keying, or VMSK. He claims VMSK achieves spectral efficiencies of 90 bits/sec/Hz or more while requiring no more power than conventional modulation methods such as BPSK.

    These claims are in direct violation of the mathematical principles of digital communications discovered by Harry Nyquist (1928), Claude Shannon (1948), and others. These fundamental principles are as firmly established as the laws of thermodynamics; neither is at all controversial among competent engineers. Unfortunately, thermodynamics and communications theory are arcane (and related) subjects that many laymen either misunderstand or refuse to accept. Hence we still have cranks trying to build perpetual motion machines. And we have Mr. Walker and his VMSK.

    My interest in debunking VMSK is very simple: people are investing real money in a technology that can't possibly meet its inventor's claims. And since the subject falls into my personal field of expertise, I feel a civic duty to speak out.

    Summary of technical points

    The following summarizes my technical conclusions regarding VMSK:

  • A VMSK signal consists of two components: a very wide, very weak BPSK signal that carries the data and a strong, infinitely narrow spectral line (which I call the "clock") that carries most of the signal power.

  • On a spectrum analyzer, the clock makes the VMSK signal look impressively narrow. It even meets FCC spectral masks that could not be met by the BPSK component alone. That's because the FCC generally requires that a minimum percentage (usually 99%; see 47 CFR 2.202), but not all of your power be in a specified frequency range. By adding a powerful (but totally information-free) spectral line to the wide BPSK component actually carrying the data, VMSK's "legal" bandwidth drops to that of the spectral line. This reveals an amusing loophole in the FCC regulations that should probably be plugged, e.g., with spectral power density limits.

  • Alas, this loophole exists only because of the FCC's artificial definition of bandwidth. There are no loopholes in the underlying physical laws. Because VMSK carries data entirely in the weak wideband BPSK component, the minimum absolute RF bandwidth required to carry VMSK without intersymbol interference is exactly the same as BPSK: 1 Hz per bit per second of data rate. With SSB filtering, this figure can be at most halved to 0.5 Hz/bps, the same as QPSK (and the Nyquist bandwidth of VMSK at baseband), but no further.

  • Because of the narrow equivalent data pulses used, the BPSK component in an unfiltered VMSK signal is actually much wider than that of conventional BPSK at the same symbol rate. Ironically for an "ultra narrowband" scheme, this gives VMSK some spread-spectrum-like properties, such as the ability to tolerate limited co-channel interference.

  • Because the strong VMSK clock consumes most of the transmitter power but conveys no useful information, VMSK requires considerably more transmitter power than BPSK to achieve a given bit error rate in an ideal receiver. For the so-called (7,8,9) VMSK code, exactly 9 dB of additional transmitter power is required. This extra power goes entirely into the clock.

    Walking off a cliff

    Walker's scheme reads like an April Fool's joke, but he's quite serious (as far as I can tell -- perhaps he's been pulling my leg this whole time!)

    I have repeatedly tried to introduce Harold Walker to reality, but it's clear that I'll never succeed. Like any true believer, he's unwilling to let mere facts spoil his fun.

    Walker concedes the existence of the wideband BPSK component, but he calls it "grass" and denies that it is where his data lies. Basic Fourier transform theory and the Nyquist Sampling Theorem show that this is simply wrong; his data is carried entirely in the "grass". Yet Walker merely handwaves the grass away as "noise". He has no alternative explanation for why it exists, why it changes character when he changes his data sequence, or why it goes away entirely when he removes his data source. The only thing in his modulator that even resembles a noise source is his random test data generator.

    Walker concedes that an "ordinary" filter with bandwidth equal to his claimed signal bandwidth "utterly destroys" his modulation. At this point, most engineers would stop, investigate and reconsider their beliefs. But not Walker. He says the "ordinary" filters are to blame and one must use his special proprietary filters instead. A quick look readily shows why they work: they're crap! Their response curves fall off to only -30 dB for at least several hundred kilohertz beyond their nominal passbands. Such filters hardly reduce his effective signal bandwidth even to the Nyquist bandwidth, much less to the tiny fraction of it that he claims as his signal bandwidth.

    It is quite likely that Walker's demos have plenty of excess link margin, enough for the wideband BPSK component to overcome the 30 dB attenuation of his poor filters, reach his demodulator and produce data. None of Walker's demos have been described as having anywhere near the level of control and rigor required to keep this from happening.

    Nor has Walker described any precautions to prevent stray RF coupling from a VMSK modulator to a nearby demodulator by direct radiation or by conduction through power supply leads or other cables. Experienced RF engineers know how insidious this problem can be. Preventing it often requires elaborate measures such as careful filtering of power supply leads and the use of copper "screen rooms".

    These two phenomenona (poor filtering and stray RF coupling) could explain Mr Walker's bizarre belief that VMSK's RF bandwidth is tiny even when he concedes that VMSK at baseband requires the full Nyquist bandwidth. As every competent communications engineer knows, the RF signal leaving a linear modulator can never be narrower than the baseband signal entering it.

    On several occasions, Mr. Walker has used non-random test data (specifically an alternating 0101... sequence) to cause the "grass" to disappear, leaving only a few spectral lines. Such tests prove nothing. BPSK will give similar results, all just as predicted by fourier theory. The only meaningful VMSK spectra are those produced with random data.

    Most recently (March 2001), Walker proclaimed that "the grass is gone". His apparent "fix" was to further narrow the difference between a "0" and a "1". This puts even less power into the data-carrying BPSK component (and even more into the clock) and spreads the remaining BPSK power over an even wider bandwidth. Apparently he has now pushed it well below the noise floor of his spectrum analyzer. And since he can no longer see it, he proclaims it to be gone. I eagerly await the results of real-world testing of this new breakthrough.

    I'm reminded of the old joke about the car dealership that loses $50 on every sale but makes it up on volume. Apparently Walker has discovered that he can increase his profits by increasing his per-car loss to $500.

    Seriously, by further narrowing his data pulses Walker has unwittingly created a signal with spread-spectrum- or CDMA-like properties. More specifically, he has created a "transmitted reference" spread spectrum signal. Here the transmitted reference is the strong narrowband VMSK clock, and the "code" for his transmission is the precise phase of that clock. As Walker narrows his data pulses, his data spectrum thins and broadens to many times the Nyquist bandwidth. This gives it the ability to tolerate a certain amount of interference from other overlapping VMSK signals as long as they use distinct clock phases.

    This explains why he can place the clocks of two VMSK signals near each other, though this hardly proves (as he claims) that the clocks are the only significant VMSK signal components. As long as there aren't too many overlapped VMSK signals, they are of roughly equal amplitude, and the bit clock transitions do not coincide, it remains theoretically possible to recover the data from each stream. But just as a CDMA system's capacity is ultimately limited by mutual interference, Walker will eventually discover that he cannot pack an unlimited number of VMSK signals side-by-side. And because of VMSK's lack of error control coding and its profligate waste of transmitter power on the clock, this limit will occur much sooner than it does with existing CDMA or even TDMA systems.

    Concluding comments

    On his web page, Walker cites the cliche of the theoretical "proof" that the bumblebee cannot fly. Mr Walker must think he has already shown that his bumblebee flies.

    But he has done nothing of the sort. There is nothing remarkable about building a single RF data communication link that carries a few hundred kilobits/sec across a room or a piece of wire; that's been done for decades with many existing technologies. What would make VMSK new, interesting and useful is its claimed ability to increase the carrying capacity of a given chunk of spectrum far beyond what is possible with existing technologies, and despite his claims to the contrary, well in excess of firmly established fundamental physical limits.

    Walker has not demonstrated this, and he never will.

    To extend the bumblebee analogy further, Walker claims to have designed a new kind of bumblebee that can occupy the same physical space at the same time as thousands of other bumblebees. And as "proof" for his claims, he has demonstrated that one of his bumblebees (sort of) flies.

    In sum, VMSK is utterly worthless as a modulation method intended to increase the capacity of the RF spectrum.

    Boring business stuff

    In my original critique of VMSK I said that VMSK was being marketed by AlphaCom Communications, a multilevel marketing company. AlphaCom is listed on Yahoo Finance as AHCM.OB.

    In previous statements, Walker had proclaimed AlphaCom an "embarassment", but he did not dispute their right to market VMSK. Recently, however, Walker has disclaimed any connection between his company (Pegasus Data Systems) and AlphaCom. First he said that they have not met their obligations (presumably financial) to him. Now he says they never had the right to market VMSK in the first place, and that his lawyers are going after them. And in a press release in the fall of 2000, AlphaCom said:

    We know you are excited about what the future holds with our VMSK technology. But, to date we do not have any VMSK products or services for you to sell. I want everyone to know that everyone at AlphaCom Communications is committed to providing these new products as soon as possible and we are dedicated to providing them through our distribution model. We strongly discourage recruiting for new distributors based solely on the deployment of VMSK. We encourage you to focus on the available products and services we have to offer so that you do realize immediate income.

    Then all mention of VMSK disappeared from AlphaCom's web site, and a SEC filing mentioned a dispute with Walker over the ownership of the VMSK patents.

    But now AlphaCom's most recent SEC filings again claim ownership of VMSK. They now call it "AHCNB" - "Alpha High Capacity Narrow Band". On his own web page, Walker hotly disputes AlphaCom's claims of ownership. Another Alphacom 8-K filing discloses that one of their directors resigned in protest over alleged mishandling of shareholder funds. And a filing from August 2000 makes the rather astonishing (to me) statement "WE MAY HAVE COMMITTED A VIOLATION OF SECURITIES LAWS...".

    So the plot continues to thicken. One can only wonder what's going on behind the scenes, especially as more and more people realize that VMSK is snake oil. Walker claims he's now funded by some unnamed west-coast venture capitalists, but this doesn't really surprise me. The Pixelon scam convincingly showed that some VCs have vastly more dollars than sense and are eager to throw tens of millions of dollars into bogus technologies with astonishingly little due diligence.

    Note added 19 Apr 2001: Alphacom's CEO has demanded that I remove these web pages. He threatens to sue if I do not comply. Obviously this person doesn't know me (or free speech law) very well, but he's been told in no uncertain terms that I will stand by my principles and continue to tell the truth about VMSK.

    I encourage anyone interested in mirroring these pages to do so. Just in case.

    Since my personal interest in VMSK is solely technical and educational, I really don't care about the business arrangements and personalities. But I suppose this couldn't have happened to a nicer bunch of guys.

    For more reading

  • The credulous EDN Magazine article that first got my attention
  • My original critique of VMSK
  • Walker's rebuttal
  • My response to Walker's rebuttal
  • A refutation of Walker's thoroughly confused "notes on Shannon's Limit"
  • A refutation of Walker et al's claim that the data cannot be in the "grass"
  • A VMSK simulation that shows what happens when VMSK is filtered to its claimed bandwidth
  • Walker still doesn't get it (10 Apr 2001)
  • Walker's latest "breakthrough" (23 May 2001)
  • The Real Facts of Life: PDF, LaTex. (11 June 2001)
  • Answer to Walker's Delusion essay
  • Comments by someone who attended the supposedly successful Boston VMSK demo on 22 Sep 2000 (18 Jun 2001)
  • Yet another proof that VMSK's data is entirely in the grass (19 Jun 2001)

    And finally, in the category of "things that make you go 'hmm!'", here's a most interesting webpage I found on the US Securities and Exchange Commission website.

    Disclaimer

    Although I am an employee of Qualcomm, my analysis of VMSK is entirely on my own initiative. Although some might accuse Mr. Walker of fraud, I do not.

    Although there are troubling hints to the contrary, such as Walker's staunch refusal to back up his claims that reputable, mainstream entities are supporting his work, I believe it best to give Mr. Walker the benefit of a doubt and assume he's honestly self-deluded.

    Last updated: 20 March 2007