How to measure

(Under a vehicle's belly, ) I attach a sensors (which act like a Tx) to the
vehicle's tires- one sensor for each tire. As the vehicle moves, tire
revolves and makes a circular "region". I have a dedicated receiver (Rx)
somewhere in the middle from all (multiple-- thinking 18 wheeler, etc.)
tires!

Multipath and NLOS are expected.

Question:
(1) Is the channel between the Tx and Rx time varying when the vehicle is
moving?
(2) Do I have to consider fading?
(3) As the tire moves, the position of the Tx moves along the circular
region. Do I need to consider doppler?
(4) I want to use Energy Detector (for non-coherence detection) to detect
the UWB signals. Google tells me "energy samples" are received form the
energy detector. but how does one collect "energy samples"?
(5) I want to measure and store a signal metric (like that "energy
samples") and store in a database, tagging the name of the region as well.
Basically, "region" here means tire.
(6) Furthermore, I want to use ray tracing to simulate this kind of
environment. DO I need to define/assume some kind of channel model? Indoor
UWB Channel model? S-V, etc?

Please provide me some hints. Does anyone know one or two (or more) books
that talk about these topics?

Thank you.

>(Under a vehicle's belly, ) I attach a sensors (which act like a Tx) to
the
>vehicle's tires- one sensor for each tire. As the vehicle moves, tire
>revolves and makes a circular "region". I have a dedicated receiver (Rx)
>somewhere in the middle from all (multiple-- thinking 18 wheeler, etc.)
>tires!
>
>Multipath and NLOS are expected.
>
>Question:
>(1) Is the channel between the Tx and Rx time varying when the vehicle is
>moving?
>(2) Do I have to consider fading?
>(3) As the tire moves, the position of the Tx moves along the circular
>region. Do I need to consider doppler?
>(4) I want to use Energy Detector (for non-coherence detection) to detect
>the UWB signals. Google tells me "energy samples" are received form the
>energy detector. but how does one collect "energy samples"?
>(5) I want to measure and store a signal metric (like that "energy
>samples") and store in a database, tagging the name of the region as
well.
>Basically, "region" here means tire.
>(6) Furthermore, I want to use ray tracing to simulate this kind of
>environment. DO I need to define/assume some kind of channel model?
Indoor
>UWB Channel model? S-V, etc?
>
>Please provide me some hints. Does anyone know one or two (or more) books
>that talk about these topics?
>
>Thank you.
>
>
>

I don't know all the answer but I will stab at a few.

3) your statement in question 4 should answer this (non-coherence
detection)
4) hint, how do you plan on detecting your tx signal on your rx end?
5) you need to really clarify this question

On Fri, 25 May 2012 13:40:06 -0500, "x201s"
<tienshanjp@n_o_s_p_a_m.yahoo.co.jp> wrote:

>(Under a vehicle's belly, ) I attach a sensors (which act like a Tx) to the
>vehicle's tires- one sensor for each tire. As the vehicle moves, tire
>revolves and makes a circular "region". I have a dedicated receiver (Rx)
>somewhere in the middle from all (multiple-- thinking 18 wheeler, etc.)
>tires!
>
>Multipath and NLOS are expected.
>
>Question:
>(1) Is the channel between the Tx and Rx time varying when the vehicle is
>moving?

Yes.   Since the distance between the Tx and Rx are constantly
changing, and the vehicle is moving in constantly changing reflective
and inteference environments, the channel can be expected to change.
That being said, much depends on the power levels, signal bandwidth,
and expected range of the system, which should be known, I would
think.

>(2) Do I have to consider fading?

That depends on some things like what I have mentioned above, but if
there's a channel model that's been developed for your system you
should be able to sort it out from that.   Since you mention that the
system is NLOS and has multipath, it's almost a given that you have to
consider fading.

>(3) As the tire moves, the position of the Tx moves along the circular
>region. Do I need to consider doppler?

That depends on a lot of things, like the modulation type, symbol
rate, and the type of receiver.   ;)

You probably also have to worry about the doppler due to reflections
from moving objects around the vehicle.

>(4) I want to use Energy Detector (for non-coherence detection) to detect
>the UWB signals. Google tells me "energy samples" are received form the
>energy detector. but how does one collect "energy samples"?

I'd start with an antenna.   ;)

Seriously, though, energy is represented by the samples of the signal
you get from the A/D converter.

>(5) I want to measure and store a signal metric (like that "energy
>samples") and store in a database, tagging the name of the region as well.
>Basically, "region" here means tire.
>(6) Furthermore, I want to use ray tracing to simulate this kind of
>environment. DO I need to define/assume some kind of channel model? Indoor
>UWB Channel model? S-V, etc?

An indoor UWB model would likely be inappropriate if the vehicle is
not expected to be indoors much.  A ray-tracing system needs to
reflect the expected usage model, vehicle moving through an
environment of reflectors, the Tx and Rx motion relative to each
other, etc., etc.  Any relevant channel models should take such
expected usage models into account.


>
>Please provide me some hints. Does anyone know one or two (or more) books
>that talk about these topics?
>
>Thank you.
>
>

Eric Jacobsen
Anchor Hill Communications
www.anchorhill.com

hint: The "ray-tracing" is a typical homework-style problem. 
Look at the definition of Rayleigh- and Rician fading. You don't need any
fancy 3D modeling, just a few time-varying path lengths, count the
wavelengths and turn into a complex phasor, put some fixed attenuation on
top of each and add the paths.  

The good news is, your delay spread is short => high coherence bandwidth =>
flat channel (which sadly isn't very useful in non-coherent detection). 
The bad news is, if it fades, you get a wideband notch - all frequencies
fade coherently. 
This will be a problem at slow speeds. 

PS before you draw any conclusions from raytrace experiments, note that
reflection on a metallic surface -flips the sign- of the reflected path. 

Intuitive explanation: the E-fields of incident and reflected wave must
cancel on the metal (=>short circuit, E=0) => needs opposite sign. 

In practice it would lead me to the conclusion that a very shallow
reflection tends to interfere destructively, not constructively.

That's at least what I observed when I did a one-tap multipath experiment
with measurement hardware years ago:

http://www.dsprelated.com/showarticle/32.php

The relevant part here is 
>> "footnote": The reflection itself may cause a sign change, but it
doesn't really matter here.

Well, I think it matters very much when your main reflection is shallow,
such as from the underside of a truck. 

Here are the electromagnetics behind the sign change (page 3)
http://www.ece.msstate.edu/~donohoe/ece3313plane_wave_reflection_transmission.pdf

On 5/25/2012 2:40 PM, x201s wrote:
> (Under a vehicle's belly, ) I attach a sensors (which act like a Tx) to the
> vehicle's tires- one sensor for each tire. As the vehicle moves, tire
> revolves and makes a circular "region". I have a dedicated receiver (Rx)
> somewhere in the middle from all (multiple-- thinking 18 wheeler, etc.)
> tires!
>
> Multipath and NLOS are expected.
>
> Question:
> (1) Is the channel between the Tx and Rx time varying when the vehicle is
> moving?
> (2) Do I have to consider fading?
> (3) As the tire moves, the position of the Tx moves along the circular
> region. Do I need to consider doppler?
> (4) I want to use Energy Detector (for non-coherence detection) to detect
> the UWB signals. Google tells me "energy samples" are received form the
> energy detector. but how does one collect "energy samples"?
> (5) I want to measure and store a signal metric (like that "energy
> samples") and store in a database, tagging the name of the region as well.
> Basically, "region" here means tire.
> (6) Furthermore, I want to use ray tracing to simulate this kind of
> environment. DO I need to define/assume some kind of channel model? Indoor
> UWB Channel model? S-V, etc?
>
> Please provide me some hints. Does anyone know one or two (or more) books
> that talk about these topics?

I don't know what you want to measure, how the sensors transmit their 
signals, or how signals from differentiated. What happens to your 
transmission path when the vehicle traverses a wet road or splashes 
through a puddle? Does it matter?

Jerry
-- 
Engineering is the art of making what you want from things you can get.
�����������������������������������������������������������������������

First, thank you to jacobfenton, Eric Jacobsen, mnentwig and Jerry Avins.
I appreciate your time and interest.

Allow me to explain what I want to do. 

I am interested to localize (=find) a region, rather than finding a precise
point from a certain (known) position.In a typical tire pressure monitoring
system, identifying which tire (which region) has low tire pressure would
be sufficient.

The idea is to attach a UWB Tx in wheel rim and rotate it at certain
predefined angles (say, at 10, 20, 30.....360 degrees, 0/360 being 12
o'clock position) and measure (simulate in my case) "some signature" of the
UWB as received at an Rx.

I suspect that it is "amplitude" of the received signal I am after, and
when I use Energy Detector (ED), and in case of multi-path, I need to
collect energies from all the paths.. at least all the "dominant" paths.

This is only the first half. On the other half, I need to compare those
"signatures" that I just collected above with an unknown (real life signal;
when I am actually driving and say a tire's sensor starts sending some
pulses..), minimizing some kind of cost function (which I have no idea as
of now, but someone suggested that using a correlation comparison is the
easy way out). When I get a match, that should give me which tire (again,
which "region") was sending that signature. This is possible because in the
first half, I had stored not only the signatures of various locations, but
also the tire id along with the signatures, say, in a matrix.

Somewhere I believe I read that for ED, I don't need to worry about delay
profile. In case of ED, the time difference between the template signal and
the received signal is zero, so I can forget about delay profile.

mnentwig: I am sorry that I don't have enough knowledge to implement what
you wrote. I googled and have found some research articles about ray
tracing in the context of UWB propagation channel.So I want to use ray
tracing.

Eric: Yes, indoor channel would not be sufficient.Luckily I found that two
researchers (G. Lasser and C. Mecklen brauker of Vienna Tech) have propsed
a channel model that may be useful in my case.
Check: www.publik.tuwien.ac.at/files/PubDat_186175.pdf  

Problem: They developed that channel model for narrow band. I don't know if
I can use it for wide band, and UWB in particular.

My current progress: I struggling to come up with a plan. What "output" do
I get from the ray tracing? What do I do with it? I need to store those
values (amplitudes?) at different positions of the Tx (10, 20, 30....360
degrees..), tagged as "region_1". I will do that simulation for all tires.

Next, in localization phase, when some unknown signal arrives, I will get
the best match, and when a match is found, I will also know the problematic
tire.

Some answers to Eric again:
I want to use Matlab for a receiver simulation. 
type: Binary Pulse Position Modulation/Energy Detection.
Chip period: 2 exp -9
pulse width: 1.4 exp-9
2048 symbol/packet

Jerry:
I want to use the tuwien people's model. I expect only changes required is
changes in electrical properties of the road surface (dielectric loss,
permittivity, anything else?)

mnentwig: 
>problem at slow speed
How slow? I could probably argue (not with you :) but in my paper) and
"assume" that the vehicle in questions is a high speed, long distance
*huge* truck.

>You don't need any fancy 3D modeling, just a few time-varying path
>lengths, count the wavelengths and turn into a complex phasor, put some
>fixed attenuation on top of each and add the paths. 

Do you have some matlab/C/java/python codes that does what you wrote? Or do
you know any resources that may have it?

Anyway, getting longer. Let me stop here this time.

Thanks again,and Cheers,
x201s    

 

Where do you all go? :)



>First, thank you to jacobfenton, Eric Jacobsen, mnentwig and Jerry Avins.
>I appreciate your time and interest.
>
>Allow me to explain what I want to do. 
>
>I am interested to localize (=find) a region, rather than finding a
precise
>point from a certain (known) position.In a typical tire pressure
monitoring
>system, identifying which tire (which region) has low tire pressure would
>be sufficient.
>
>The idea is to attach a UWB Tx in wheel rim and rotate it at certain
>predefined angles (say, at 10, 20, 30.....360 degrees, 0/360 being 12
>o'clock position) and measure (simulate in my case) "some signature" of
the
>UWB as received at an Rx.
>
>I suspect that it is "amplitude" of the received signal I am after, and
>when I use Energy Detector (ED), and in case of multi-path, I need to
>collect energies from all the paths.. at least all the "dominant" paths.
>
>This is only the first half. On the other half, I need to compare those
>"signatures" that I just collected above with an unknown (real life
signal;
>when I am actually driving and say a tire's sensor starts sending some
>pulses..), minimizing some kind of cost function (which I have no idea as
>of now, but someone suggested that using a correlation comparison is the
>easy way out). When I get a match, that should give me which tire (again,
>which "region") was sending that signature. This is possible because in
the
>first half, I had stored not only the signatures of various locations,
but
>also the tire id along with the signatures, say, in a matrix.
>
>Somewhere I believe I read that for ED, I don't need to worry about delay
>profile. In case of ED, the time difference between the template signal
and
>the received signal is zero, so I can forget about delay profile.
>
>mnentwig: I am sorry that I don't have enough knowledge to implement what
>you wrote. I googled and have found some research articles about ray
>tracing in the context of UWB propagation channel.So I want to use ray
>tracing.
>
>Eric: Yes, indoor channel would not be sufficient.Luckily I found that
two
>researchers (G. Lasser and C. Mecklen brauker of Vienna Tech) have
propsed
>a channel model that may be useful in my case.
>Check: www.publik.tuwien.ac.at/files/PubDat_186175.pdf  
>
>Problem: They developed that channel model for narrow band. I don't know
if
>I can use it for wide band, and UWB in particular.
>
>My current progress: I struggling to come up with a plan. What "output"
do
>I get from the ray tracing? What do I do with it? I need to store those
>values (amplitudes?) at different positions of the Tx (10, 20, 30....360
>degrees..), tagged as "region_1". I will do that simulation for all
tires.
>
>Next, in localization phase, when some unknown signal arrives, I will get
>the best match, and when a match is found, I will also know the
problematic
>tire.
>
>Some answers to Eric again:
>I want to use Matlab for a receiver simulation. 
>type: Binary Pulse Position Modulation/Energy Detection.
>Chip period: 2 exp -9
>pulse width: 1.4 exp-9
>2048 symbol/packet
>
>Jerry:
>I want to use the tuwien people's model. I expect only changes required
is
>changes in electrical properties of the road surface (dielectric loss,
>permittivity, anything else?)
>
>mnentwig: 
>>problem at slow speed
>How slow? I could probably argue (not with you :) but in my paper) and
>"assume" that the vehicle in questions is a high speed, long distance
>*huge* truck.
>
>>You don't need any fancy 3D modeling, just a few time-varying path
>>lengths, count the wavelengths and turn into a complex phasor, put some
>>fixed attenuation on top of each and add the paths. 
>
>Do you have some matlab/C/java/python codes that does what you wrote? Or
do
>you know any resources that may have it?
>
>Anyway, getting longer. Let me stop here this time.
>
>Thanks again,and Cheers,
>x201s    
>
> 
>
>

On Tue, 05 Jun 2012 18:29:45 -0500, x201s wrote:
(top posting fixed)
>>First, thank you to jacobfenton, Eric Jacobsen, mnentwig and Jerry
>>Avins. I appreciate your time and interest.
>>
>>Allow me to explain what I want to do.
>>
>>I am interested to localize (=find) a region, rather than finding a
> precise
>>point from a certain (known) position.In a typical tire pressure
> monitoring
>>system, identifying which tire (which region) has low tire pressure
>>would be sufficient.
>>
>>The idea is to attach a UWB Tx in wheel rim and rotate it at certain
>>predefined angles (say, at 10, 20, 30.....360 degrees, 0/360 being 12
>>o'clock position) and measure (simulate in my case) "some signature" of
> the
>>UWB as received at an Rx.
>>
>>I suspect that it is "amplitude" of the received signal I am after, and
>>when I use Energy Detector (ED), and in case of multi-path, I need to
>>collect energies from all the paths.. at least all the "dominant" paths.
>>
>>This is only the first half. On the other half, I need to compare those
>>"signatures" that I just collected above with an unknown (real life
> signal;
>>when I am actually driving and say a tire's sensor starts sending some
>>pulses..), minimizing some kind of cost function (which I have no idea
>>as of now, but someone suggested that using a correlation comparison is
>>the easy way out). When I get a match, that should give me which tire
>>(again, which "region") was sending that signature. This is possible
>>because in
> the
>>first half, I had stored not only the signatures of various locations,
> but
>>also the tire id along with the signatures, say, in a matrix.
>>
>>Somewhere I believe I read that for ED, I don't need to worry about
>>delay profile. In case of ED, the time difference between the template
>>signal
> and
>>the received signal is zero, so I can forget about delay profile.
>>
>>mnentwig: I am sorry that I don't have enough knowledge to implement
>>what you wrote. I googled and have found some research articles about
>>ray tracing in the context of UWB propagation channel.So I want to use
>>ray tracing.
>>
>>Eric: Yes, indoor channel would not be sufficient.Luckily I found that
> two
>>researchers (G. Lasser and C. Mecklen brauker of Vienna Tech) have
> propsed
>>a channel model that may be useful in my case. Check:
>>www.publik.tuwien.ac.at/files/PubDat_186175.pdf
>>
>>Problem: They developed that channel model for narrow band. I don't know
> if
>>I can use it for wide band, and UWB in particular.
>>
>>My current progress: I struggling to come up with a plan. What "output"
> do
>>I get from the ray tracing? What do I do with it? I need to store those
>>values (amplitudes?) at different positions of the Tx (10, 20, 30....360
>>degrees..), tagged as "region_1". I will do that simulation for all
> tires.
>>
>>Next, in localization phase, when some unknown signal arrives, I will
>>get the best match, and when a match is found, I will also know the
> problematic
>>tire.
>>
>>Some answers to Eric again:
>>I want to use Matlab for a receiver simulation. type: Binary Pulse
>>Position Modulation/Energy Detection. Chip period: 2 exp -9
>>pulse width: 1.4 exp-9
>>2048 symbol/packet
>>
>>Jerry:
>>I want to use the tuwien people's model. I expect only changes required
> is
>>changes in electrical properties of the road surface (dielectric loss,
>>permittivity, anything else?)
>>
>>mnentwig:
>>>problem at slow speed
>>How slow? I could probably argue (not with you :) but in my paper) and
>>"assume" that the vehicle in questions is a high speed, long distance
>>*huge* truck.
>>
>>>You don't need any fancy 3D modeling, just a few time-varying path
>>>lengths, count the wavelengths and turn into a complex phasor, put some
>>>fixed attenuation on top of each and add the paths.
>>
>>Do you have some matlab/C/java/python codes that does what you wrote? Or
> do
>>you know any resources that may have it?
>>
>>Anyway, getting longer. Let me stop here this time.
>>
>>Thanks again,and Cheers,
>>x201s

I think they're all in the dark, holding their heads and moaning.

Even after clarification, your questions are still pretty unclear.  If 
anything, the clarification muddied the waters.

You're tossing out a lot of buzzwords, but you have yet to say what you 
want to _do_:

"I want to make a system that detects when an 18 wheeler runs over a dead 
animal, analyzes the collision, and informs the driver if he should drive 
on, or turn around and carve off a steak."

"I want to make a system that keeps track of the wheel positions on an 18 
wheeler to use for (illegal alien detection, tire condition monitoring, 
generation of really oddball pornography, etc.)"

So (particularly for Jerry and me) start at the VERY TOP LEVEL.  Be like 
they taught you in journalism class.  First, go through your dictionary 
and take out all the words of three syllables or more.  Then:  tell us 
What you want to do; tell us Where you want to do it; tell us When you 
want to do it.  Then, go over to the editorial page, and tell us Why you 
want to do it.

We will then try to assist you with How -- because I'm pretty sure you're 
on a long windy road that stops at a dead end.

-- 
My liberal friends think I'm a conservative kook.
My conservative friends think I'm a liberal kook.
Why am I not happy that they have found common ground?

Tim Wescott, Communications, Control, Circuits & Software
http://www.wescottdesign.com

On 6/5/12 10:48 PM, Tim Wescott wrote:
>

>
> I think they're all in the dark, holding their heads and moaning.

ohhh, <moaning>...

>
> "I want to make a system that detects when an 18 wheeler runs over a dead
> animal, analyzes the collision, and informs the driver if he should drive
> on, or turn around and carve off a steak."

that perked me up a little...

>
> "I want to make a system that keeps track of the wheel positions on an 18
> wheeler to use for (illegal alien detection, tire condition monitoring,
> generation of really oddball pornography, etc.)"

... getting more interesting.  i'm already typing one-handed...

lessee... semi-truck, roadkill, food, porn...  what's left?

not much...

maybe, flatscreen TV?

>
> So (particularly for Jerry and me) start at the VERY TOP LEVEL.

and i'm a bottom-up sorta guy and i can't figure anything out either.


-- 

r b-j                  rbj@audioimagination.com

"Imagination is more important than knowledge."