What statistical distribution did you draw the noise levels from?

None! My complex math is so rusty it's pathetic. I used the C++ rand() function and limited the maximum noise level to 7 pixels above the bottom of the display.

Man, i think my demo isn't passing muster with the hardcore :oops:

TG

The Mk 23 TAS was also distinctive... i remember hearing it in the audio of a CNN broadcast once after i transferred to CIVLANT. One of their reporters was on the flight deck of a carrier yapping about something or other, and some of the video/broadcast equipment must have somehow demodulated the PRF, and the pulse train bled into the audio feed. I recognized it instantly... some of those signals used to follow me into my rack and haunt my dreams :o

It did indeed have its benefits! We were so short on qualified ESM operators that I could spend all my time below PD qualifying... didn't take me long! And I NEVER cranked.

TG

I guess it's no problem. With sonar it's a bit more complicated, I for example use different averaging windows for slow and fast broadband .. which should affect the 'look' of the noise. I too use 'flat' pseudo-randoms for the fast display, and I average few to get more Gauss looking noise for the slow displays.

Hey SeaQueen .. since you mind .. you know something more about it ? Some links ? I haven't found any. Also about the importance of this in sonar (& ESM) ?

rand() would imply the uniform distribution. I suspected that from looking at it. God... you know you're a physicist when you can tell the distribution of random noise by looking at it. That scares me.

It would be a better simulation of real life if you used the Gaussian distribution. You can generate those from the uniform distribution using the Box-Muller method. You should be able to find a good discussion of it. The idea is that it's hard to generate one Gaussian distributed random number but it's easy to make two and throw one away (or use it for something else). I actually wrote a short program to generate Gaussian distribute random numbers. I use it for various projects of mine. I could send you the source if you'd like.

I really enjoyed seeing this. It's a great start. It would be nice if there were menus where you could put in more parameters and play with things like wave forms and what not. Like... on the top display I assume I'm looking at frequency versus the received level (in dB?). The other two displays I'm not sure I understand entirely.

One of my gripes with naval simulations is that it seems like game designers focus too much on the button-ology without paying enough attention to the substance which drives decision making. This is techno-weenieness, but it's good techno-weenieness. You can learn something from it. If something like this could be improved upon and integrated into something like Dr. Sid's subsim, that'd be absolutely fabulous. The interface doesn't necessarily have to look slick, it just has to be usable and updatable so that eventually it could be slick.

The distribution statistical distribution of the noise? Well... for one it contributes to your false contact rate, although that's just a part of it. In a LOFAR gram, say, just by random chance, you might see a target that's not there due to the noise in the system. That's why you adjust your integration times. If you use long integration times, the noise averages out more and you can see a signal better with fewer false contacts, but at the expense of not getting information as quickly. At shorter integration times you can't see as far, but you can react more quickly albeit at the expense of more false contacts.

In radar it's less important because the rate of sampling is much higher, so things average out much more quickly, but it does have an effect. Most signal processing is designed around Gaussian noise, though. You can generate pairs of Gaussian distributed random numbers with the Box-Muller method.

http://www.taygeta.com/random/gaussian.html

The noise level is a big driver in sonar particularly, but radar as well. A lot of the physics of radar and sonar are the same. The jargon is just different. The basic ideas in both cases is that the atmosphere and the ocean is a waveguide. They just rephrase the jargon. In radar, for example, they plot the refractive index of the atmosphere versus altitude, while in sonar they plot the speed of sound versus depth. Regardless, it's just the speed the wave propagates at. In the end, it's all just Snell's Law.

It's all good stuff.

Sure, I'd be interested in seeing that!

We never actually measured incoming signals in dB, dBm, or any specific units (unless the ESM system was being groomed). We used the crude, but effective, signal strength scale I outlined in the readme:
-SS-1, signal is barely discernible from background noise. No side lobes detectable.
-SS-2, signal is clearly discernible from background noise, but still relatively weak.
No side lobes detectable.
-SS-3, signal is clearly discernible and of moderate strength. There may be some weak
side lobes present.
-SS-4, signal is strong, with several side lobes clearly discernible/audible.
Depending on other parameters, this represents a possible detection or collision threat.
-SS-5, signal is very strong – the receiver is saturated, i.e. little or no distinction
can be made between the side lobes and the main lobe. Depending on other parameters, this
represents a detection or collision threat.
This worked great if you kept in mind the characteristics of the radar signals and how they affect the capabilities of the radar (range, resolution, etc).


The other two displays are the demodulated radar pulses. The first pulse analysis display (the middle display) shows multiple pulses so that the Pulse Repetition Interval can be measured. The second display shows a single pulse so that the pulse width can be measured.

Here's the ESM Quick-tutorial part of the readme:
ESM Quick-tutorial – ESM stands for Electronic Surveillance Measures, and is basically the passive
(receiving or listening) side of ECM, or Electronic Countermeasures (or EW, Electronic Warfare).
ESM performs several functions in a military context:
-identification of platforms in a tactical situation, to identify both friendly and enemy forces
-evaluation of signal strength and emitter type, to determine whether or not the emitter’s platform
is a detection threat (i.e. will be able to detect ownship using the emitter detected by ownship)
or a collision threat (i.e. the signal strength is so high that ownship is within the radiation
pattern of the emitter, close enough to be a hazard to ownship)
-reconnaissance of sea-based, airborne or landbased platforms for the purpose of gathering intelligence,
either in a tactical or non-tactical situation

All intelligence gathered using ESM is Signals Intelligence, or SIGINT. SIGINT includes intelligence
about various RF signals, which includes radar and communications signals. This simulator only depicts
intercepted radar signals.

Signal Parameters – the basic parameters of concern that are typically measured by an ESM system are:
-frequency in MHz (or GHz, 1GHz = 1000MHz). In general, the lower the frequency, the longer
the detection range. Typical marine navigation radars operate in the 8000-10000MHz range, while a
long-range early warning air search radar might operate at about 300MHz.
-Pulse repetition interval in microseconds (us); the duration between the leading edge of one pulse
to the leading edge of the next. In general, the longer the PRI, the longer the detection range.
-Pulse repetition frequency in pulses per second (pps), which is the reciprocal of the PRI; the number
of pulses transmitted by the emitter per second. In general, the lower the PRF, the longer the
detection range, which corresponds to a longer PRI.
-Pulse width in microseconds (us); in general, the longer the pulse width, the longer the detection range,
because of higher average power output. However, a longer pulse width contributes to a lower
range/target size resolution. A radar with a short pulse width is better able to distinguish between
multiple targets that are close together.
-Scan rate in seconds; this usually only applies to circular or sector scans. Scan rate is not measured
by this simulator.
-Signal strength – depends on who’s doing the measuring. Back when I was on the boat, we had a 5-level
signal strength system:
-SS-1, signal is barely discernible from background noise. No side lobes detectable.
-SS-2, signal is clearly discernible from background noise, but still relatively weak.
No side lobes detectable.
-SS-3, signal is clearly discernible and of moderate strength. There may be some weak
side lobes present.
-SS-4, signal is strong, with several side lobes clearly discernible/audible.
Depending on other parameters, this represents a possible detection or collision threat.
-SS-5, signal is very strong – the receiver is saturated, i.e. little or no distinction
can be made between the side lobes and the main lobe. Depending on other parameters, this
represents a detection or collision threat.

What is the typical polarisation of a marine/search radar? Vertical or horizontal? Or rotating?

No idea... I'm sure a Google search would help find that information.

TG

I think maybe I should clarify something about this program - it's just a simulation of how an ESM receiver might appear if it were to be included in a submarine simulation. It's a demo, at best!

The signals themselves are extremely rudimentary simulations. You will see the basic parameters of a signal's frequency, PRI/PRF, and pulse width, but you will not see the results of environmental effects. As an example of how rudimentary the signal simulations are, here's how I decided to program them: I started with how I wanted the display to look. Then I decided how often I wanted to update the display. I took those numbers and used them to program a low-frequency oscillating loop which would generate amplitude values to send to the display, based on signal parameters set at the program's initialization. To sum up - I started from the display and worked backwards!

I'll post more specifics later.

TG

IIRC most surface search radars are essentially navigation radars and randomly polarized. I would bet that is not universally true however.

It would be cool if this got into Sid's CSS. The radar/ESM model in DW is rather disappointing and this level of detail would be very nice to have.

If I could steal the attention of the ESM/radar buffs for a moment, I'm looking for some related information, if anyone would care to lend a hand. Thanks!

I have added a Word doc describing the "inner workings" of the program to the CADC Downloads center, along with a separate file containing the source code for the signal generation functions.

TG