Wednesday, December 14, 2011

CASARA, Arctic First Response and Electronic Searches

This is the first in a series of postings on my recent dealings and experiences with the Civil Aviation Search and Rescue Association. Other articles in this series are:

If I had seen this article from the CBC, or this one about two and a half years ago, when I was still a happy member of the Civil Air Search and Rescue Association, I would have been ecstatic. But that was before I learned a very hard lesson about what level of technical correctness is acceptable by CASARA, and what CASARA management is prepared to do to people who dissent from the officially accepted (if not officially published) line. Now I am just very concerned.

My concern stems from a set of electronic search techniques that have been used by at least one CASARA unit when tasked by the Joint Rescue Coordination Centre in Trenton. I have written before on the procedure for performing the electronic ELT homing technique know as the Aural Null, and the confusion in CASARA about how it works. Earlier this year I received a paper from another member of CASARA titled IMPROVEMENT IN POSITION ACCURACY For ELT and VISUAL SEARCHES by Anne Barr, Mike Casey, Langley R. Muir. This is a long paper with several technical assertions in it so let's take them one at a time. In this post I will address how the Aural Null procedure works.

When discussing the Aural null in their paper Barr, Casey and Muir state:

The basic assumptions on which these techniques work are that there is only one ELT operating in the area and that the radiation pattern of that ELT is very nearly circular at the search altitude.

In practice the second of these assumptions can be severely mistaken. Reasons for the radiation pattern to depart from circularity could include:
  • deformation of the antenna or ground plane of the ELT
  • the presence of one or more reflectors, such as the metal of the crashed aircraft or hanger walls
  • the presence or absence of trees and vegetation of possibly variable conductance, reflectance, and spacing leading to variable absorption and reflectance, refraction, and diffraction of the signal
  • the presence of one or more topographical reflectors leading to complete or partial constructive or destructive interference patterns
  • possible wave guide issues if the ELT is near or within buildings, structures, or severe topography and
  • the presence of high tension power lines. 

Any or all of the above can lead to a non-circular radiation pattern and consequently make the simple geometry of the Aural Null procedures give rise to incorrect positions. Nonetheless, an understanding of how radiation patterns can be distorted as well as training and practice in a variety of circumstances can allow a CASARA crew to locate an ELT in many difficult circumstances.
...
Some of the difficult circumstances noted above are discussed in the CASARA Training Manual1 in the section on the use of DF techniques with the L-PER (p 139 ff), the CAP Manual (p 179 ff), and they are well-known within the Electronic Search Specialist (ESS) community. It is not too difficult to devise circumstances in which no method, other than the Gridded Mapping technique, would enable a crew to determine the position of a radiating ELT. In the remainder of this paper however we will leave these complications aside and assume that the task is to determine the location of a well-behaved ELT which is radiating in a circular pattern with no other complicating factors.

Emphasis is mine.

More than one of the scientists and engineers who have reviewed the paper have noted that, when dealing with life and death situations, one does not have the luxury to decide what environmental factors to consider and what to set asside. Other than that, the only real problem with this statement is in the first paragraph "... the radiation pattern of that ELT is very nearly circular at the search altitude" and similar assertions. They are correct with most of their reasons why the ELT radiation pattern is not circular, but the Aural Null technique does not depend on the radiation pattern being nearly circular; rather it depends on the area in which the ELT signal may be detected being nearly circular.

The signal that CASARA, or any other searchers, will use to track an ELT is transmitted on 121.5 MHz at a power of 0.1 Watt (or 100 milliwatts, mW) for the older style TSO C91 units and most TSO C126 units. Engineers and technicians will usually refer to the power output of a transmitter in decibels (abbreviated dB) relative to one Watt (dBW) or relative to one milliwatt (dBm). This makes the mathematics much simpler when asked to calculate things like what distance a particular transmission my be detected.

If you are already familiar with dB arithmetic you may wish to skip this paragraph, and the next. Decibels are a way of stating a ratio. As above, an ELT will radiate at a power of 100 milliwatts. We can express that as a ratio of 100 mw / 1 mw. This seems to be a pointless operation, but it will make sense in a moment. If we perform the division implied by the ratio 100 ÷ 1 = 100, the units cancel each other out. To convert this ratio to the units of bel we use the common logarithm: log(100) = 2. Since a decibel is one tenth of a bel, 2 bels would equal 20 decibels and since the original ratio was in relation to milliwatts 100 mW is 20 dBm.

The reason we convert values to decibels is one of the features of logarithm functions. Most of the processes that affect radio or audio energy are either multiplications or divisions. For example an amplifier may double the power of a signal so that the power output of the amplifier is twice the input power. On the other hand if a signal has to travel some distance to a receiver the power at the receiver may be one half of the power transmitted. By using logarithms these multiplications become addition, and divisions become subtraction. If we convert the amplifier doubling into decibels we get 10 × log(2 ÷ 1) = 3dB. Similarly if we convert the loss over distance to decibels we get 10 × log(1 ÷ 2) = -3dB. So if we are far enough away from the ELT that the power has been divided in half, we could say we have 100mW ÷ 2 = 50mW, or we could say we have 20dBm - 3dB = 17dBm. If we put the received signal through our doubling amplifier we could say we are amplifying 50mW × 2 = 100mW, or 17dBm + 3dB = 20dBm. Addition and subtraction are easier than multiplication and division so you will find radio engineering formulae use dB and dBm.

As Barr, Casey and Muir state there are many things that can affect the radiation pattern of any transmitter including an ELT, and the effect on the radiation pattern will affect how far away the signal may be detected in each direction. The most important factor affecting the reception of the signal though is distance. In their paper Barr, Casey and Muir correctly state "...the signal strength will radiate outwardly (both horizontally and vertically) in a very nearly spherical pattern and will vary inversely with the square of the distance from the antenna." Unfortunately they then completely ignore the affect of the inverse square law on reception distance.

When radio engineers and technicians compute how much power a particular receiver will see from a transmitter they use a link budget equation:
PRX = PTX + GTX - LTX - LFS + GRX - LRX
Where:
  • PRX = received power (dBm)
  • PTX = transmitted power (dBm)
  • GTX = transmitter antenna gain (dB)
  • LTX = transmitter losses (coax, connectors, etc.) (dB)
  • LFS = free space loss or path loss (dB)
  • GRX = receiver antenna gain (dB)
  • LRX = receiver losses (coax, connectors, etc.) (dB)
For now let's just worry about  LFS the free space loss or path loss. This is where the inverse square law is applied.We will use the simplified equation for frequency in MHz and distance in kilometers:
LFS = 20log(dkm) + 20log(fMHz) + 32.45
The frequency is fixed at 121.5 MHz so the equation further simplifies to:
LFS = 20log(dkm) + 74.14
If we ignore the other terms in the link budget formula for now we can generate a chart showing the signal strength in dBm the receiver will see given the distance from the transmitter, and transmitted power.
And for clarity here is the same chart to a maximum distance of 100 km.

The final piece of information we need is what signal strength the receiver needs in order that we can detect the signal from the ELT. That is often called the noise floor. Signals with a strength below the noise floor can not normally be detected. The noise floor depends on a number of factors. For VHF signals and receivers a value near -140 dB is normal. But aviation receivers use amplitude modulation and have slightly wider bandwidth so I'm using a conservative -107 dBm based on a sensitivity of 1μV into 50 Ω. We can see from the charts that a transmitter radiating an effective power of 10 dBm (3 milliwatts) would be detectable up to 100 km. A transmitter radiating 0 dBm (1 millwatt) could be detected up to 35 km. To look at it another way, if the radiation pattern of an ELT is non-circular, even if the radiation strength is 20dB less in one direction than in the ideal case, the signal could still be detected by a search platform up to 35 km, that is 19 nautical miles, away. To put that in perspective, a log periodic antenna (often used for TV reception in rural areas in past decades and still seen on top of towers around the country) may have an attenuation of 25dB in the direction opposite of the 'forward' direction. A log periodic is a fairly complex device designed by engineers. So, while it is possible that something might affect the radiation pattern enough to reduce detection range below 50 nautical miles, it would have to be something significant such as a large geologic feature, catastrophic physical damage, a nearly exhausted battery, etc.

This is a good time to go back to the link budget equation to talk about the remaining four parameters: GTX, LTX, GRX and LRX . The two gain parameters GTX and GRX represent the antenna gain of the transmit and receive antennas respectively. For example, the log periodic antenna of the last paragraph would have a gain of 9dB in the forward direction, or a gain of -25dB in the backward direction. The two loss parameters LTX and LRX refer to the total losses in the transmitter or receiver system, cables, connectors, etc. These parameters would be very important if the antenna was some large distance from the transmitter or receiver. For the situation of an ELT and an airborne search platform these values would be a few dB, or less.

Isotropic approximation of a mono-pole
Toroidal approximatin of a mono-pole
Up to this point I have been assuming approximately isotropic radiation patterns (left). An isotropic radiator sends equal amounts of power in all directions and has a gain of 0dBi.The antennas in use with ELTs and communications radios installed in aircraft are ground plane whips (or mono poles) which have a toroidal (right) radiation pattern and a typical gain of 5.2dBi. This means that in practice, when computing the link budget equation we should add approximately 10dB of gain to account for the antennas in use. The lower half of the radiation pattern is truncated for clarity, and due to the ground plane of a whip antenna (mono-pole).

In a previous posting I shared this table from the RCAF National SAR Manual with you. It gives ELT detection ranges for search platforms flying at various altitudes:
Altitude feetRange nm
100030
200045
300055
400067
500085
10000100
15000127
20000150
30000200

So why does the RCAF give ELT detection ranges of less than 50 nautical miles when search platforms are below 3000 feet? Did we not just show that the ELT should be detectable for a much greater distance than that? Indeed we did. So what is going on? ELTs use Very High Frequency or VHF signals. In the same band but slightly higher than the frequencies used by FM radio stations. VHF signals only travel line-of-sight. They can not bend around the curve of the earth the way short wave or AM radio station signals can do. In fact the light we see by follows the same rules. So, just like you may be able to see a short distance while standing, but a greater distance when in a tall building, or from a hill top, a search platform may "see" an ELT (that is receive a signal from it) only if the ELT is nearer than its radio horizon.

Line of sight - Radio Horizon. Aircraft at A and C detect the ELT, Aircraft at B does not.
As you might expect there is a formula for calculating the distance to the radio horizon:

d = 1.41 ×√ h 
Where:
  • h = height in feet
  • d = distance in miles 
or:
d = 1.07 ×√ h 

Where:
  • h = height in feet
  • d = distance in nautical miles

If you do the math you will find your figures are very close to those provided by the RCAF. So the Aural Null does not depend on the radiation pattern of the ELT. So long as the link budget calculation shows the receiver will receive a signal above the noise floor, it depends on the radio horizon of the search platform. Since the Earth is very nearly spherical, the radio horizon will be very nearly circular. This was the official position of the RCAF when I discussed it with them not too long ago.

I don't know why these experienced members of CASARA would make this mistake. I do know that the first flight (with which they attempted to collect the data used in their paper) had to be repeated because they could not detect the training ELT beacon even when flying directly overhead. They replaced the beacon and flew again. I tested the ELT used for the first flight and found that the battery had been completely depleted. Perhaps for a number of training missions over a protracted period the batteries in the training beacons had insufficient charge to produce a strong enough signal to be detected to the radio horizon and this caused confusion.

A more likely reason comes from the paper. They describe an electronic search that they call an Aural Null, but the points they use for the calculation were taken with the receiver tuned to a frequency 150 kHz above the ELT frequency. Spurious signals from an ELT can be detected at frequencies that far away from the design frequency but only well within the radio horizon. On the flight documented in their paper the reception range was about 2.5 km. Under those conditions, a non-circular radiation pattern would affect the accuracy of the calculation. This is why the approved Aural Null technique does not specify tuning off the ELT frequency.

I have not been able to get a satisfactory answer as to why they would use an incorrectly performed search procedure, which produced predictably inaccurate results, to attack the reputation of an approved electronic search technique; nor why persons in leadership positions in the provincial and national organizations would support this. I am very concerned that the ideas set out in their paper have become widely accepted within CASARA. If CASARA is going to be providing first response SAR in the arctic, it is important that their operations be based on proper, scientifically valid techniques.

So whether by accident of poor battery management, or naive use of off-tuning to home spurious transmissions or some other mechanism members of CASARA have become convinced that radiation pattern rather than radio horizon is the controlling factor in how the Aural Null technique works. In my next article I talk about what can happen if a search platform actually performs a search in the way recommended by the Barr, Casey and Muir paper as the product of these misconceptions.



Some graphics reproduced with permission from SARMobile.ca

No comments:

Post a Comment