Sunday, January 8, 2012

CASARA, Electronic Searches - A Radio Tutorial

This post continues a series I began in my post on CASARA, Arctic First Response and Electronic Searches.

I have sent individual invitations to the three authors of the paper I'm reviewing (Anne Barr, Mike Casey and Langley Muir) to read what I am posting, and to submit their own comments. So far the only response I've gotten was from Anne Barr who referred to an email sent out by Mike Daniels (CASARA Ontario President) in which he said:
There is an issue here of two very knowledgeable people at odds regarding the Theory of the Aural Null Procedure. I doubt whether either of you will ever agree on this issue. Therefore the best resolution is to agree to disagree and move on. Quite frankly we are all far too busy to carry this discussion further, I have instructed CASARA Ottawa to close their book on this issue once and for all. By doing so they may get on with their business on hand.
...

As stated I have requested CASARA Ottawa close their book on this issue of Aural Null/Cardinal Pass and not to initiate or respond to further correspondence on this issue unless it originates from CASARA National or from one of the above mentioned agencies. Further, I have suggested to John Davidson, President, CASARA National and the Executive that they do the same.
The business on hand is, of course, searching for lost aviators. To do that they will on occasion need to use aircraft without Direction Finding equipment to perform electronic searches. This will require that they use the Aural Null technique. If, as Mr Daniels asserts, "There is an issue here of two very knowledgeable people at odds regarding the Theory of the Aural Null Procedure", then the other knowledgeable person should be able to provide valid scientific, engineering and technical data substantiating that the techniques documented, and that have been used on actual search and training missions are sound, effective and safe. Alternatively they could acknowledge that the techniques are unsound, ineffective and unsafe. I would also appreciate a full explanation of what instructing CASARA Ottawa to "close their book on this issue" really means. So I extended the same invitation to John Davidson, President of CASARA National. Until they respond, I can only assume that because they consider "two very knowledgeable people to be at odds" they consider the material in the Barr, Casey and Muir paper to be correct and it will continue to influence CASARA operations. So I will continue with my review.

A good portion of the Barr, Casey and Muir paper is given over to Off-Tuning methods, so it seems appropriate to go into off-tuning in some depth. Off-tuning is the practice of tuning away from the actual frequency on which a transmitter is sending a signal. Due to a number of factors a transmitter will always emit some of its energy on frequencies other than the one intended. One mechanism for this is harmonic radiation. Harmonics are frequencies that are a multiple of some fundamental frequency. So an ELT designed to transmit on 121.5 MHz will also transmit on harmonics at twice, three, four, etc times that frequency or 243 MHz, 364.5 MHz, 486 MHz, etc. However since modern transmitters are designed to transmit on many evenly spaced frequencies, or channels, (even ELTs) their design and construction gives rise to spurious transmissions on those channels adjacent in frequency to the intended channel. So an ELT transmitting on 121.5 MHz may also transmit on 121.525 MHz, 121.55 MHz, 121.575 MHz, 121.6 MHz, etc. It may also transmit on 121.475 MHz, 121.45 MHz, 121.425 MHz, 121.4 MHz, etc. Or on some subset of these frequencies depending on the transmitter design. These spurious transmissions are so common that they may cause interference to users of adjacent channels so regulatory agencies such as the FCC enact regulations to control them. For ELTs these are:

   (h) For ELTs operating on 121.500 MHz, 243.000 MHz and 406.0-406.1 MHz
   the mean power of any emission must be attenuated below the mean power
   of the transmitter (pY) as follows:

   (1) When the frequency is moved from the assigned frequency by more
   than 50 percent up to and including 100 percent of the authorized
   bandwidth the attenuation must be at least 25 dB;

   (2) When the frequency is removed from the assigned frequency by more
   than 100 percent of the authorized bandwidth the attenuation must be at
   least 30 dB.
 
Industry Canada Regulations are:

5.6 Unwanted Emissions

The power of unwanted emissions measured by an averaging meter of 300 Hz resolution bandwidth shall be attenuated below the level of the mean transmitter power (TP) by:
  1. at least 25 dB on any frequency removed from the centre of the authorized bandwidth by more than 50% up to and including 100% of the authorized bandwidth;
  2. at least 30 dB on any frequency removed from the centre of the authorized bandwidth by more than 100%.
The authorized bandwidth is 25 kHz.

These regulations are much less restrictive than normal aviation band communications transmitters intended for routine communications. An ELT, as emergency equipment is given a regulatory "break" in the operating parameters it has to meet. Normally the amount of power on each spurious emission will continue to be reduced as the spurious frequency is farther from the intended frequency. Even though the regulations would allow all spurious emissions removed more than 100% of the bandwidth to be 30 dB lower than the emission on the intended frequency, there would be some reduction of power for each step away in frequency.

This property allows people hunting transmitters to narrow down the location by tuning off the intended frequency when they are very close. Using this technique may, for example, allow a CASARA ground team to identify which airplane of several located at an airport has a transmitting ELT. The technique also works in the air, but there are some other factors that make the situation not quite that simple. If you read my previous post on the subject, these other factors are some of the complications that Barr, Casey and Muir chose to leave aside. I will address them in due course. If you have not already read that post I suggest you do so before continuing with this one.

In their paper Barr, Casey and Muir state:
There are essentially four factors that will determine the audible volume of the ELT signal; the relative orientation of the two antennae, the aircraft altitude, the receiver frequency, and the volume setting on the receiver. When prosecuting a search it is important that one changes only one variable at a time, observes the effect, and only then change one of the other variables. After one is in the near vicinity of the ELT, using the standard techniques, one can then refine the search further using the other techniques described in this paper.
Unfortunately this completely ignores the action of the automatic gain control, and propagation effects. It is also unclear where they came up with a relation between the aircraft altitude and the audible volume. The effect of the volume setting needs no explanation. Changing the receiver frequency is off-tuning; tuning to a spurious emission with more or less power will have an effect on the amount of power received. Changing the orientation of the receiver antenna with respect to the orientation of the transmitting antenna will cause polarization effects and change the amount of power received. However the orientation of the ELT antenna, in the case of an actual search, is unknown. The orientation of the receiving antenna is determined by the attitude of the airplane as it maneuvers. So there is little a search crew can do about or with polarisation effects other than to be aware of their potential. Propagation effects, especially multi-path induced fading, will also change the amount of power received. The automatic gain control is built into the receiver to compensate, as much as possible, for what ever affects the amount of power received.

It may help to think of the automatic gain control as a robot that has its own volume (gain) control that works on the signal before it gets to the pilot's volume control. The robot never shows the pilot what setting it is using. The robot turns its volume control up or down as the amount of received power decreases or increases trying to keep the signal amplitude it sends on at the same constant level. The pilot's control then increases or decreases the output volume from the one maintained by the robot. This is why a receiver is able to deal with a transmission from the control tower only a mile distant that may have a 50 or 100 watt transmitter, and the reply from an airplane with a 10 watt transmitter 25 miles away without the pilot changing the volume control. Of course there are limits to the amount of control the robot has. A very weak signal may be so weak the robot turns its volume control up to the maximum. Any reduction in power beyond that point could result in a reduction of output volume, or the loss of signal reception. A very strong signal may be so strong the robot turns its volume control down to the minimum. Any increase in power beyond that point will result in distortion of the output signal. Except at these two extremes, it is not possible to determine with any degree of certainty what the strength of the signal was before the robot applied its adjustment. This also underscores a problem with the technique proposed by Bar, Casey and Muir. How can the crew only change one variable at a time, when an automated system is constantly changing one of the variables behind the scenes? Clearly a technique that does not have this limitation is needed.

Without knowing the location, power output and the transmit antenna radiation pattern of the ELT, there is no way to operate the search airplane, other than by chance, to put the receiver in the position of dealing with a signal that forces the automatic gain to the minimum. In fact, the ELT output may not be enough to force the automatic gain to a minimum regardless of how close the search airplane gets to the ELT. It is possible to operate the airplane to put the receiver in the position of dealing with a signal that forces the automatic gain to the maximum. Off-tuning far enough, flying far enough away that path loss attenuates the signal enough, or flying at the cusp of the radio horizon so the signal is attenuated can all put the receiver into a condition where it can not apply any more amplification through the automatic gain. Tuning further off frequency, flying further away or continuing over the horizon respectively will result in the loss of the signal. In fact crossing the radio horizon from beyond to within the horizon and noting the position of the cusp as the signal is detected is the basis of the Aural Null technique. The Aural Null procedure that Barr, Casey and Muir propose as an augmentation to the standard procedure is a combination of off-tuning and using distance to reduce received signal strength until the ELT signal is not detectable.

If we assume the received signal strength decreases monotonically with increasing distance from the ELT, on the intended frequency and all spurious emissions, then their procedure does sound logical. Succinctly it is: once the ELT signal is first detected, the crew begins to off-tune the receiver. At first off-tuning may result in a loss of signal if the spurious emission is not strong enough. Eventually though, as the airplane continues in a direction that reduces the distance to the ELT the signal will be detected on the off-tuned frequency. As the searchers continue to close the range to the ELT, they will be able to off-tune the receiver by greater amounts and still detect the signal. Once the signal is lost on an off-tuned frequency where it was previously heard, the crew can conclude that they have passed abeam the ELT, change direction by 90 degrees and continue the procedure. But can we reasonably make that initial assumption?

One of the things that Barr, Casey and Muir leave aside from consideration are propagation factors. As a radio signal propagates from transmitter to receiver it is affected by the environment. This will have an effect on the signal strength at the receiver that is called fading. The dominant form of fading is multi-path induced fading, often shortened to multi-path. Multi-path affects consumer reception of broadcast radio and television as well as cellular telephone so it has be studied extensively. Radio engineers are primarily interested in fade depth which is the amount that received signal strength is reduced by propagation under various conditions. Fade depth is factored into the link budget equation to determine how much power and gain is needed to ensure reliable communication. Most of the work in the field has been done for terrestrial or satellite communications but S. Loyka, A. Kouki and F. Gagnon published a paper in the proceedings of the IEEE on the computation of fade depth for ground to air communications links. Their results show that, in the presence of specular reflection, for clearance angles (the angle the search aircraft is above the horizon of the ELT) between 0.1° and 10° fade depth in dB is well estimated by:
-20 × log(θ) + 25
where θ is the clearance angle in degrees. For clearance angles of about 5° fade depth can be 10dB, 1° gives 25dB and 0.1° 45dB. Search aircraft at an altitude of 500 feet above the ELT will have a clearance angle less than 1° when further than approximately 5 nm away; at 1000 feet the same clearance angle is achieved beyond approximately 10 nm. Fading, and in particular multi-path induced fading, does not affect received signal strength uniformly. Since the fade depth depends on the relative lengths of multiple propagation paths from the ELT to the search aircraft it will vary as the aircraft moves about or changes altitude. If we consult the received signal strength charts from my previous article:
it is easy to see that the 25dB attenuation possible at 1°, or even 45dB at 0.1° will have little effect for the 20dBm budget curve at ranges less than 70km. However when combined with the Barr, Casey and Muir off-tuning aural null method, the total effect of off-tuning and fading can take the signal below detectable strength at large distances from the ELT. But since the fading is not uniform the received signal strength does not decrease monotonically as distance from the ELT increases. Since a deep fade would be indistinguishable from passing abeam the ELT; following their procedure a search aircraft encountering a deep fade while off-tuning would erroneously conclude they had passed the ELT and turn. They do not provide any procedure for detecting or reacting to this misguidance due to fading.

The graphic on the left (from www.sarmobile.ca which has a more in-depth description of ELT propagation effects) one can see how fading in combination with off-tuning can create areas of local signal peaks that can quite easily seduce a search aircraft away from the ELT. During my time as a member of CASARA Ottawa I observed this happen on at least three separate training flights, two while I manned a training beacon on the ground, one as aircraft pilot following guidance from a navigator that, as it turns out, was using this technique. In each case I reported the situation during the exercise debrief. It wasn't until I was appointed unit training officer that I learned Barr, Casey and Muir were purposefully experimenting with this technique and using it when dispatched on actual search missions.

When designing critical systems, especially if health or safety depend on the critical systems, one strives to design in a fail safe or at least fault tolerance or graceful degradation. The standard aural null is fault tolerant. If the ELT performance is degraded to the point that the transmitted signal is not detectable at the radio horizon, and the radiation pattern is non-circular for that or some other reason, there will be errors in the geometry. Those errors are detectable, correctable and are small when compared to the total search area. The augmented aural null procedure proposed by Barr, Casey and Muir is not fault tolerant. As shown above, errors introduced by fading are more likely to be large than small because the greatest fading occurs at lower clearance angles and therefore greater distances. No procedure for detecting and correcting large guidance errors is provided. Clearly a technique which instead of guiding searchers to the vicinity of the ELT, guides them to a location 10 nm or more away from the actual location presents a safety problem to any survivors awaiting rescue. Rather than augmenting the operation of the standard aural null, this procedure destroys the fault tolerance of the standard procedure, and introduces unnecessary risk to survivors.

Also in their paper Barr, Casey and Muir state:
At the time of writing the methods described below have not yet been extensively tested, but both logic and preliminary testing would indicate that considerable improvements in time required and the precision of the location are possible. More practice and more experimentation is, of course, desirable and reasoned criticisms and discussion of potential pitfalls are always welcomed. Again, these methods are meant to augment and not to replace the usual CASARA procedures, which are still necessary and fundamental.
Well here are some the reasoned criticisms I have been trying to get them and the rest of CASARA to understand from the time before they wrote their paper, from the time when I was unit training officer, in fact. They appeal to logic to justify their procedure, but their starting assumptions are not true, so their logical deductions fail. They have not done extensive testing, even though the technique was used on actual search and training missions. Neither, it seems, have they examined the math or physics already well established in the field. In fact they claim:
The overall topics of wave propagation and the mathematics of the various aspects of finding maximums and minimums are the subject of a vast literature in the fields of mathematics, physics, and engineering. Nevertheless, in the case of CASARA actually looking for an ELT we have little need to understand either advanced mathematics or complex physics.
I would concede that when following an established search procedure there is little need to understand the math and physics. But when modifying existing procedures, or developing new ones, the need to understand the science is crucial.

It is possible that I have made a mistake along the way, but the number of scientists, engineers, technicians and signals analysts who have reviewed and agreed with my work and conclusions is very long. Surely, as I said at the start, if this is a difference of opinion between two very knowledgeable people, then the other person should be able and willing, if not eager, to present a valid scientific rebuttal.

One of the cornerstones of flight safety is the public acknowledgement of mistakes. As one of my high school teachers was fond of saying "learn from the mistakes of others, you can't possibly live long enough to make them all yourself". That would serve well as the motto of any safety program. It is not possible to learn from the mistakes of others if they hide them; neither is it possible to learn from your own mistakes if you don't acknowledge them. How then can we rely on an organization that refuses to accept valid criticism from qualified professionals? Aren't the lives of people who may go missing on a flight in Canada worth what ever time it takes to get this right? What are they doing that makes them too busy to critically examine search techniques used in training and on actual missions which have come under professional criticism?

The next post is this series is Off-Tuning - Cardinal Pass - What's in a Name?

No comments:

Post a Comment