joe smith
March 6th 07, 01:28 PM
After years of good service from our ATD-200, am planning to upgrade to a
Zaon MRX collision avoidance unit. (Good review in March's Aviation
Consumer. Hopefully the SureCheck days are over.) It's nice knowing there's
traffic around, but altitude info is better.
But what I REALLY want is something that also detects azimuth. So far the
only option is Zaon's XRX. But I'm not enamored with the XRX. It's too big,
and not very ergonomic. [The last thing you want to be doing when the
traffic alarm goes off is studying a display heads down. It's bad enough you
have to look at the MRX for altitude info. How about a man's "Traffic alert"
voice for traffic below and a woman's voice for traffic above?] And true
TCAD is too expensive.
Anyway, I'm not the entrepreneur type and will never get around to this, so
I throw it out for someone else to develop. When I was a grad student, I was
working with Adaptive Arrays. (Anti-jamming / Military stuff.) These are
simple affairs, using something called the "LMS" algorithm to adapt to
incoming signals. (Least Mean Squared) The array would automatically find
and null the strongest incoming signal.
The rule was that an N element array could adapt to N-1 signals. So, for
example, the simplest array, a two element array, would null out the
strongest incoming signal from any direction, giving a text book cardioid
pattern with the null pointing to the offending signal. It seems to me that
we might only need a two or three element array to keep things small and
simple. It's not like we have to continually lock on to continuous wave
signals. We just need something fast to detect and adapt to different signal
bursts as they arise.
By taking the weights from the array, one can compute and display the
position of an incoming signal with good accuracy. There's a matter of
processing time. You would want the weights to settle down enough and I
don't know if that can be done fast enough for a typical xpndr reply. But
maybe close enough to give a quadrant solution. Otherwise, the concept is
simple. The array automatically picks and adapts to the strongest signal, so
even if there's a collision between two offending xpndr signals, the
strongest one (presumably the closest one) wins the attention of the array.
I did a cursory Google search and I see the technology has moved right along
from 30 years ago. They're using this stuff now in cell phones and GPS's.
Presumably, the technology has become cheap and maybe off the shelf. I don't
know if the XRX is using this technique already or is simply doing a brute
force, "let's continually rotate the array and scan" radar approach.
Hope someone smarter and more driven than I can come up with something using
this simple technique. I never could figure out how to write a DO LOOP in
Assembly in school, and this type of product will sink or swim on post
processing to make it usable and ergonomic.
Mike Palmer <><
Excellence in Ergonomics
Zaon MRX collision avoidance unit. (Good review in March's Aviation
Consumer. Hopefully the SureCheck days are over.) It's nice knowing there's
traffic around, but altitude info is better.
But what I REALLY want is something that also detects azimuth. So far the
only option is Zaon's XRX. But I'm not enamored with the XRX. It's too big,
and not very ergonomic. [The last thing you want to be doing when the
traffic alarm goes off is studying a display heads down. It's bad enough you
have to look at the MRX for altitude info. How about a man's "Traffic alert"
voice for traffic below and a woman's voice for traffic above?] And true
TCAD is too expensive.
Anyway, I'm not the entrepreneur type and will never get around to this, so
I throw it out for someone else to develop. When I was a grad student, I was
working with Adaptive Arrays. (Anti-jamming / Military stuff.) These are
simple affairs, using something called the "LMS" algorithm to adapt to
incoming signals. (Least Mean Squared) The array would automatically find
and null the strongest incoming signal.
The rule was that an N element array could adapt to N-1 signals. So, for
example, the simplest array, a two element array, would null out the
strongest incoming signal from any direction, giving a text book cardioid
pattern with the null pointing to the offending signal. It seems to me that
we might only need a two or three element array to keep things small and
simple. It's not like we have to continually lock on to continuous wave
signals. We just need something fast to detect and adapt to different signal
bursts as they arise.
By taking the weights from the array, one can compute and display the
position of an incoming signal with good accuracy. There's a matter of
processing time. You would want the weights to settle down enough and I
don't know if that can be done fast enough for a typical xpndr reply. But
maybe close enough to give a quadrant solution. Otherwise, the concept is
simple. The array automatically picks and adapts to the strongest signal, so
even if there's a collision between two offending xpndr signals, the
strongest one (presumably the closest one) wins the attention of the array.
I did a cursory Google search and I see the technology has moved right along
from 30 years ago. They're using this stuff now in cell phones and GPS's.
Presumably, the technology has become cheap and maybe off the shelf. I don't
know if the XRX is using this technique already or is simply doing a brute
force, "let's continually rotate the array and scan" radar approach.
Hope someone smarter and more driven than I can come up with something using
this simple technique. I never could figure out how to write a DO LOOP in
Assembly in school, and this type of product will sink or swim on post
processing to make it usable and ergonomic.
Mike Palmer <><
Excellence in Ergonomics