Sensor Network For non-Ionizing Radiation 2

[NavasCOL]

Hi guys,

I am building a sensor network to detect devices in the 800 MHz to 2.4 GHz band using the AD8361. Obviously we want to detect cell phones, bluetooth and wireless connections with our network and be able to see all the power density on the area. We designed an Archimedean spiral antenna and we attached it to the AD8361 using SMA connectors. After that we want to use a microcontroller to digitalize the DC output of the AD8361, then send that digital signal to a data acquisition card as the PCI-1200 and finally use LabView to present the result.

My problem is: I dont know what kind of microcontroller use because the speed of the signals that we are sensing.
The output of the sensor is between 0V and 3.6V, what do you think I should use, 8 bits or 10 bits converter.
Do you have any comments about all this? About the antenna or the connectors?

PS: So far I'm thinking to use the PIC12F675
PS2: We would like to triangulate the radiation emitter too and try to locate the emitter in an area. How difficult is that? I would appreciate comments about that too or any other thing.

Thanks

J

 

[MacGyverS2000]

Do you care about seeing individual signals or do you only care about power densities? If density is all you care about, I can't imagine a need for high-speed processing... you're not trying to decode the traffic, only get a sense of average power.


Dan - Owner

http://www.Hi-TecDesigns.com

 

[VEBill]

"Triangulation" using only wide-bandwidth power density measurements from several locations is susceptible to being fooled by circumstances.

For example, if one sensor is adjacent to an 'extra, unexpected' low power emitter, then it will detect a high power density (thanks to the 1/r^2 law). This data point will throw off your 'triangulation' algorithm.

Basically - by converting the RF envionment to one or more simple scaler power density measurements, you've thrown away 99.999999...% of the information. You shouldn't expect much in what remains.

If you've got only ONE source, then it should work after a fashion. It might be difficult to be sure that there is only ONE source.

 

[VEBill]

Also...

I believe that this chip is more intended for direct connection within low power transmitters. You might want to confirm that it will detect anything when you account for a low power source PLUS all path losses.

 

[NavasCOL]

Hi again guys, thanks for your answers, i do appreciate it.

macgyver: i only care about power densities so i dont need a high speed microcontroller. my real problem about that is that i dont want to lose information so i would have to use the appropriate microcontroller. this will be the first time i work with this devices so i have no idea which one to use. Can you give me a hint please? Thanks for your comments and help mc.

VE1BLL: This is a 3 stages project and now i can assume there is only one source in the measure area. I know we cant be sure there is only one but in this stage we will probe our device in an OATS (Open Area Test Site) and in that way we can be sure our test emitter will have more power than others unknown radiation sources. Do you think under this circumstances an easy triangulation algorithm would work?

Actually i was thinking on do the LAbView program the easiest as possible. Just comparing the power density of each sensor (we will use 4 devices to have an square area)
and giving the biggest power density of the measure as the location of the radiation source could work. OF course the error using that would be huge but it could be a good approximation if there is only one radiation source and a small test area.

I would like to hear your comments about all this and any other comment from more people.

Thanks for your help.
Regards,

J

 

[MacGyverS2000]

One the one hand you say you don't need a high-speed uC, but then you follow it with "I don't want to lose information". What exactly is it that you think you'll be losing? How fast do your density measurements need to be? Even with a low-end PIC I can measure at a 200kHz sampling rate and have plenty of time for math in between. If you're only sampling and storing, the processor will be twiddling its thumbs in between measurements.


Dan - Owner

http://www.Hi-TecDesigns.com

 

[NavasCOL]

Thanks again for your answer Mac.

Since I will work in the 2.4 GHz, I will be sensing Wireless Access Points (WAP) using the 802.11b and 802.11g standards. This devices use FHSS and DSSS modulation techniques. This frequency hops occur really fast and each one of them has information about the total power density of the whole signal.

Besides, there could be more devices (i.e cell phone) in the measure area. Without the triangulation option I would like to have the information of that power density too. So, the DC out put of the sensor can change really fast and with the wrong uC I could lose information.
Please correct me if i am wrong but that;s what i think could happen.

Regards,

J

 

[MacGyverS2000]

I'm not overly familiar with the protocols used with 802.11, so I can't speak authoritatively on the subject.

If you mean the protocol itself includes the capability to hold the power density then you have no hope of using a PIC to do that... it just doesn't have the power to decode the stream (at least by itself).

If you're talking about more simple measurements timed between each frequency hop, then your measurement speed will be based upon the dwell time set up by the system parameters. A system might have a dwell time of 200ms, which is plenty of time for a PIC to make a measurement and do some processing.


Dan - Owner

http://www.Hi-TecDesigns.com

 

[NavasCOL]

Thanks so much Mac, that was exactly what i was talking about. I was wondering if the speed of the changes in the frequency hops could be too much for the PIC to detect them. So you think i wont have problems if i use the PIC12F675? I just need to digitalize the DC signal, then save them for a little while and send the package to the data acquisition card.

Besides that, what do you think about the "triangulation" problem that i talked about in the 5th post. Do you think it is a good approximation use that kind of comparison?

PS: We are simulating the Archimedean spiral antenna using CST simulation software. When i get final results i would like to share them with you and hear your opinion about them, what do you think about that?

Thanks for all your comments,
Regards

J

 

[VEBill]

Triangulation based on power density might work well if the source is isotropic (or at least omnidirectional in 2D), and the path consistent. Typically, these conditions are not true in the real world. If this system is intended for a limited situation for a specific purpose, then it might work.

 

[NavasCOL]

Hi VE1BLL, thanks for your comment.

I want to sense cell phones, bluetooth and wireless connections working on the 800 MHz to 2.4 GHz band. As far as i know, all this devices work with omnidirectional antennas so "isotropic" sources wont be a problem. What is the best solution (or the problems) trying to triangulate this kind of devices using my system? Because my solution so far is the trivial one hehehe...
Is there any standard to do that? I would like to find an algorithm to use it in LabView.

Thanks for all your comments an support. I appreciate it

J

 

[VEBill]

Those sorts of devices might be nominally omnidirectional, but one shouldn't be surprised to see variances of several dB in their real-world azimuth pattern depending on how much effort was expended in their design and how the local environment interacts with the antenna system. If the antenna inside the black plastic sleeve is (for example) a raised colliner dipole and is well decoupled from the feedline, then it might be pretty close to a perfectly circular pattern.

The normal triangulation method is to get several relative bearings and then, well, triangulate. My favorite Direction Finding technology is 'Doppler DF'. You can Google Doppler DF to learn more.

Time of arrival method is how GPS works.

 

[Higgler]

NavasCol,
What is the end use of your product? I realize you want to sense cell phones, but then what do you do with the information?

Is this for commercial use?

kch