New idea for radar.. maybe.

[richardandtracy]

I've an idea for a cheap radar system I'd like to float on this forum.
There may be something similar in existance already, but I've not come across it yet if there is.

The radar systems I see currently in use have a rotating antenna and high power/ narrow beam pulse sprayed out in a single direction at a time. This strikes me as illogical, unreliable, wasteful of power and expensive.

I want to propose a radar that works on a cheaper/ lighter/ simpler/ wholly electronic/ lower absorbed power system.
You could have an omni-directional pulse and three omni-directional recievers placed at convenient positions on the boat. The three time delays in the recieving of the return pulse can be used to calculate the position of the source of the echo. [This time delay/ position calculation is done with GPS - so it's not impossible.] Having an omnidirectional transmitter would need to be more powerful for the pulse, but it could be done every few seconds rather than at milli-second intervals.
If there were mini transmitters in each reciever, the system could calibrate itself each time it's turned on, making sure that the owner's not moved the reciever since last time the system was used. The 'send' pulse could also be used to re-check the position of the recievers every pulse too - ensuring that the receiver's not come detached.
Is there a system that uses this idea [or even anyone developing the idea]?
I tried to write a bit of software as proof of concept myself, but ran into logic problems after 3 targets - however I'm not actually involved in imaging software so I'm sure it's possible. The processing power needed would be high relative to current radars - but it's obviously possible otherwise cheap GPS would be impossible.

Any comments, any sources..? Basically, does anyone else think the idea's possible?

Regards

Richard.


<hr width=100% size=1>

 

[Evadne]

It is possible, in theory, but I doubt it's practical and I'd be very surprised if it's cheaper. There are underwater acoustic systems that work on this principle. You take the signals from your receiver array (3 transducers is the minimum size) and use DSP to form the beams. The big advantage of such a system is that you can reduce the beam width to less than that of an unweighted warray.

The first expense is in having 3 receivers instead of one. The receivers must be less than half a wavelength apart, otherwise you get "aliasing" where a signal from one angle gives the same result as a signal from a different angle. With radar this is of the order of millimetres, and the positioning accuracy must be of the order of microns, radar receivers are generally too big for this. There may be ways around this, but it will involve more receivers and some very fancy mathematics.

Despite first appearances, having a single, directional rotating receiver is by far the simplest approach.

<hr width=100% size=1>

 

[bedouin]

A very ingenious idea but there are a number of practical problems...

The signal returned from a Radar pulse is essentially an analogue stream rather than a digital pulse. Processing the analogue streams to try to identify the signal from different locations is at the least very difficult trying to do it in real time would require massive computational resources.

Secondly it would in practice be difficult to get accurate bearings. You would need to be able to measure time differences accurately to the pico-second level - that technology does not come cheap - and the delays in signal propagation through the kit would need to be very accurately catered for.

Thirdly you would need at least 3 (and for any accuracy 5 or 6) aerials mounted at a reasonable separation and a reasonable height. Most yachts have difficulty finding a single mounting point for their radome, finding 5 or 6 would be impractical (they couldn't be at deck / rail height as that would give a maximum range of only a couple of miles.)

<hr width=100% size=1>

 

[Talbot]

would only work if there was a single return, but with an omni directional pulse, there are going to be a number of semi-simultaneous returns from around the boat which would be impossible to distinguish which was which, and thus get time seperation at different receivers.

Another minor problem would be the size of the generator you would need to send out an omni-directional pulse of suficient power for it to be received. Even if you did some very clever pulse shaping and compresion, you would still need a big beast:

assuming normal radiation 12 deg beam width of 2kw (lowest power radar available) there are 30 x 12 deg beams for 360, therefore absolute minimum power required would be 30 x 2 = 60Kw and this ignores loss of power achieved by beam shaping. Now I dont know how big a gennie you have, but if it produces 60Kw I would be amazed./forums/images/icons/smile.gif

<hr width=100% size=1>

 

[Evadne]

I think the multiple return problem is one of the reasons why you couldn't use widely spaced antennae easily, whereas with the phased array described above you can vary the "look" direction. You are not after identifying each pulse, from the machine's perspective you want to measure the strength of return in each of the beam directions around the 360 degree circle. This builds up your picture.

The transmitter power is more fundemental. Not being a radar engineer I'd assumed an omnidirectional Tx signal with a narrow angle receiver anyway, as that's usually how I'd do it in acoustics. You could use the same sort of algorithm to phase a transmitter array, but that would require three or more transducers at less than half a wavelength spacing as well.
Like Trillian said, this is getting needlessly messy.


<hr width=100% size=1>

 

[Talbot]

You cant do phased array style signal processing from an omnidirectional transmission. The outgoing signal has to be phased as well to be able to assess the actual angle and range.

<hr width=100% size=1>

 

[richardandtracy]

It was the simultaneous return problem that had my software 'proof of concept' model struggling. With two targets close together the overlap of signals is very hard to distinguish. More than 3 receivers actually improves the situation by reducing the chance of getting indecipherable returns [but also makes it more difficult to calculate].

I don't think the 60 kW output is likely to be a problem - at least not from the power generation side. The idea is that you really need only one pulse every few seconds, so a low average power source can be utilised. If the output is 60 kW for a microsecond, that's 0.06J per pulse. Doing one pulse every 3 seconds gives an average absorbed power of .02W. The advantage of a short pulse is that it improves the resolution.. but I do know it will cause additional difficulties in generating the pulse.

Regards

Richard.


<hr width=100% size=1>

 

[Evadne]

What I meant was, strictly speaking, a weighted circular array. You vary the weights applied to each element to vary the "look" angle and, in this case, sweep the look angle around 360 degrees continuously, thus determining the phase (and thus the angle) of any incoming signal. As you're looking at the returns, it doesn't matter what the transmitter is doing, as long as signal is being transmitted in the "look" direction. If you apply weights to a transmitter array in the same direction they will also form a beam, giving you some signal gain in that direction.

<hr width=100% size=1>

 

[richardandtracy]

I'm not totally convinced that the aliasing problem you highlight is a real one - but I will admit it could actually be the cause of my software model not working, and my not seeing it as a problem being the fatal flaw in my idea. I'll have a think on it.

The omni-directional receivers are just recorders of an analogue signal. Then signal processing is applied based on a knowledge of the supposed position of the recievers. The solution is easy if the signal emanates within the triangle of receivers - but it's a bit late to identify a target once it's inside your vessel..

I take the point about the expense of the receivers.

Regards

Richard.


<hr width=100% size=1>

 

[AndrewJ]

how about a omini xmitting antenna and an omni receiver. the transmitter, and receiver, would vary in frequency based on direction. (the variation in frequency could be comparatively slight and still be within the authorized bandwidth) i.e. a 161.005 ghz signal would be from 000, while a 161.0055 signal would be from 002. etc. (the freqs are not correct above but give you an idea of what I'm referring to)
I also understand that unless the antenna is very high, the max range for a sail boat radar is the visible horizon. Thus power requirements are reduced. I'm sure there are flaws in this concept.

<hr width=100% size=1>The only easy day was yesterday

 

[richardandtracy]

I see what you mean. I've seen this done on some static systems around airports. Having 30 odd receivers must rather add to the expence of the receiving hardware but also reduce the processing overhead.

Regards

Richard.


<hr width=100% size=1>

 

[Evadne]

I've been working on a very similar problem in acoustics for the last couple of years, fitting a homing array to a vehicle. We use an array of 3 omnidirectional transducers at about 1/3 wavelength spacing. I's quite easy (in something like Matlab) to show that this works. We are not interested in range so have gone for a different receiver algorithm to work out he angle of returns, but range is something you want so an amplitude-based algorithm such as a beamformer would be the way I'd go.

The aliasing problem is fundemental, it is really an extension of Nyquist's theorem if you think about it. If the receivers are further apart then they are not looking at the same wave and thus you get ambiguity. Weighting the receivers only works in the absence of aliasing. If you do a polar plot of an array with too great a spacing you get more than one beam formed for a given set of weights.

You could, if you're just looking at discrete returns, measure the time difference between returns from widely spaced antennae but you just need to look at, say, a radar screen picture of a coastline to realise that this is not going going to work. The <A target="_blank" HREF=http://www.iee.org.uk>IEE</A> has a great fund of knowledge on these subjects, if you've not looked there yet.



<hr width=100% size=1>

 

[richardandtracy]

You've actually identified the area I was most concerned about [apart from a generalised algorithm for working out positions] - the tiny time differences between signal arrival due to the limited distance apart that the recievers could be placed. This is a big potential stumbling block for the whole idea. It is not insuperable and with careful component selection I think it could be avoided, probably.

Massive computational resources..
Laptop. A 1.6Ghz + PC is actually more than adeqate to do the computation required in a second for a 640 x 640 pixel resolution map (got that far with my software model on a 1.6GHz desktop..)

Bearings..
How accurate are Radar bearings? I know the precision is good, but how accurate are they? Precision & accuracy are not the same.
This radar method would make taking bearings exceptionally sensitive to timing problems between the receivers - and is likely to make the bearing taken rather less reliable than with existing methods, something that users may baulk at.

Finally - positioning the receivers. Like a complete idiot I'd forgotten that the plan view of the vessel doesn't apply 40ft up! So, yes, finding somewhere to put the receivers is going to be difficult. Doh!

Regards

Richard.


<hr width=100% size=1>

 

[richardandtracy]

To be honest - it was to find out the stupid things I'd forgotten that I made the initial post. And forgetting the plan view was a prime example!

Regards

Richard.


<hr width=100% size=1>

 

[bedouin]

The bearing on traditional radar is pretty good - resolution depends on the beam width but roughly speaking you can guarantee that it is in the direction it appears on the screen.

Now for your approach, assuming a lateral separation of about 5 metres, you are looking at a 10cm path difference representing a 1 degree difference in bearing - or in time terms that is about 3e-10 seconds - not a lot. With that sort of time you will probably need to worry about things like the thermal stability of the wires and components. Even then
I doubt that the wave returned from a real-world target will be sharp enough to make that sort of resolution realistic.

Possibly a more feasible solution would be to use multiple transmitters and a single receiver

<hr width=100% size=1>

 

[Bergman]

What an interesting idea.

What I think you are proposing is really a DF system where you supply the signal to be measure by creating a reflection from a distant object.

I suspect your first problem will be the question of power. As I'm sure you know a conventional radar transmits several kilowatts ERP- effective radiated power. Most of this comes from aerial gain and is achieved by making the aerial directive. The actual RF power generated is quite low. By using an omni-directional aerial you will be unable to do this and will have to actually generate several kilowatts of power albeit only for a short duration.

To be able to measure direction you will have to measure the phase difference between three separate receivers, or more acurately three aerials. This is quite difficult to do accurately at the sort of frequency used by radars, particularly since the you are trying to do it with a very short duration pulse which will be subject to considerable distortion having been reflected from an unknown object. Again conventional radars use lots of pulses depending upon its design function a PRF of several thousand PPS may be used.

Following from this will be the probem of discrimination. Your single pulse will be reflected from all the things that affect a conventional radar, water droplets, waves, birds etc. but you will not have the information available to process the incoming signal to remove such "clutter" because of your very low PRF. This will be made worse by the fact that your receive aerials will also have to be omni-directional and will therefore have only very low gain compared to a conventional radar aerial. This will mean that the incoming signal will have a much worse signal to noise ratio than would be the case with a directional high gain aerial again making accurate phase measurement difficult and digital signal processing even more so. The receiver design would be extremely challenging to resolve such a short duration signal with a low gain omnidirectional aerial. It would mean that to produce a useable signal return without aerial gain in either transmit or receive path would demand an enormous amount of transmit power. I've not done any sums but certainly tens if not hundreds of kilowatts of RF signal would be needed. I suspect this problem of received signal to noise ratio would be the single biggest hurdle to overcome.

My last reservation is cost, To generate several kilowatts of signal at microwave frequencies will be far more expensive that a conventional radar, several hundred kilowatts would probably cost more than the boat.

I believe a similar technique to the one you describe is used for long range "over the horizon" radars but they use much lower frequencies and have incredibly large antennae.

Sorry if I have poured cold water on the idea, perhaps instead of replacing radar it may have other applications.

<hr width=100% size=1>

 

[halcyon]

First thought, reminds me of a 1930's / 40's early radar, they used a very low freqeuncy.


Brian

<hr width=100% size=1>

 

[Bergman]

Yes

Dead right frequency was around 30 MHz I think.

And I believe that the individual stations only produced ranging information, direction was found by triangulation.

Thing is if you start to develop along those lines you end up re-inventing Decca.

Maybe a better approach is to tap into what exists.

Most coastal waters, at least the busy bits, are covered by powerful coast based radars. Why not have a system that transmits the processed information from these radars and simply have a screen display on the boat. GPS could identify your position on the plan and would show everything in the area at much better resolution than a small boat radar could achieve.

A UHF channel would supply the necessary bandwidth or possibly use the G3 cellular network. If a screen update somewhere around 2 - 5 seconds was used the bandwidth would not be too great and would give an adaquate display for normal use.

Perhaps an opportunity for the Cellular people to make some more money.

<hr width=100% size=1>

 

[MarkV]

Nice idea, not sure about the timeing issues and the comparison with GPS, the GPS timeing differences are caused by the satalites being dotted around the world, where as the timeing differences in your system will be caused by position of the arial around the boat, this is a compleetly different scale.

The other thing I thought about is as the number of targets increeses (including all of the false signals) wouldn't there be a danger of inducing additional false returns, I've not thought about this too much as it makes my brain hurt. I was thinking of identifying a point on a grid. 1 point on the grid gives 2 readings (1 horizontal 1 vertical), 2 points on the grid gives 4 readings (2 vertical 2 horizontal), reversing this gives 4 posible points.

I had another idea, isn't there some kind of radar transponder system already. Normal Civil air trafic radars can not resolve altitude, they rely on the aircraft answering a request for its height (this is the secondary survailance radar with the antenna being the straight array mounted on top of the parabolic primary radar normally found around airports). Getting back to my original thought wouldn't it be posible to send and omni request and for any ship fitted with a similar system to return its gps position heading, speed, identity... which can then be plotted on a chart plotter. As I said earlier isn't this already available (arpa, marpa or somthing). You wouldn't be able to see the land but it might help avoid collisions in fog.

<hr width=100% size=1>

 

[richardandtracy]

Thanks everyone for your comments, they are appreciated.

I've obviously missed a few obvious points and some I didn't even know may be a problem. I will need to sit back and think it through. I may end up just developing the software side as an intellectual exercise with a view to finding an application for the idea one day.

Thanks again for your comments,

Regards

Richard.


<hr width=100% size=1>