PDA

View Full Version : Understand Detection Curves.



RedDevil
12-26-2005, 01:20 PM
I'm taking a look at the 1.03 object files to try and get a handle on the operating characteristics of the sonar systems for the subs and other platforms. I am having a little bit of trouble trying to figure out theoretical maximum detection range though. My best guess is that it has to be the combination between the frequencies detected, and the detection curve values.

Can anyone explain to me how the detection curve values work? (I am assuming bigger is better, but I can't seam to find a constant that converts them to metres or any other meaningful value that I am familiar with from gameplay).

RedDevil
12-26-2005, 01:32 PM
I have seen from OneShot's sonar post that you are detected or not from the formula Passive Sonar Signal Excess = Radiated Noise - Propagation Loss - Interfering Noise - Recognition Differential.

Let's assume that Propagation Loss and Interfering Noise are constants, and remove them from this equation.

Now we have:

Passive Sonar Signal Excess = Radiated Noise - Recognition Differential.

The size of Passive Sonar Signal Excess determines your SNR ratio, I am assuming. So great, SNR>1 then you have detected the target. However, there must be a hard maximum range, just as there is a hard maximum speed at which sensors cease to funtion.

So further questions:

1) What is the Recognition Differential, and how do I find it (Nrd maybe?).
2) How are Detection Curves factored into this, or if they are not, what do they reprisent?
And most importantly:
3) How do I figure out a rough comparable range for sensors, assuming that Signal loss and Background noise are constant in all this case?

I hope this helps to make the answer to my first post easier to do without posting a total manual on how the flags in the object database work.

Cheers,
RedDevil.

LuftWolf
12-27-2005, 08:20 PM
You are asking all the hard questions right off the bat. :smile:

The detection curves work in one of two ways, depending on the sensor modality. The sonar (active and passive) work one way and all other sensors work another way.

The active and passive sonars have a whole game engine dedicated just to processing sound propagation and sonar performance, so the detection curves in the database only have one significant value. That is the absolute, hardcapped, max range of the sensor. Even if the acoustics engine would have a contact theoretically appear on the sensor beyond that range, based strictly on the sound propagation model (which takes the Passive Sound Level or Active SL and calculates the propagation loss through the acoustic conditions and then combines it with the sensor NRD, which is, simply put, the ability of the sensor to reduce background noise to distinguish signal from noise, or "how sensitive the sensor is," in more basic terms... but you also have to look at the frequency range, which is just as if not more important, to which the sensor is sensitive to get its actual in game sensitivity), no contact will appear on active or passive sonar that is farther than that hardcapped value. That is why in stock DW you get loud contacts suddenly appearing out of nowhere as very strong contacts. It was a top priority in the LWAMI Mod to set these hardcaps at ranges beyond where even the loudest contacts would appear. I honestly don't know why SCS didn't do this as well, as sensor hardcaps for sonars aren't really necessary at all.

In terms of detection curves for other sensors, they don't have their own engine like the acoustics system, so the curves determine the actual in game ranges in all cases, depending on modifiers. Basically, the double or triple digit value corresponds to the detectable object's detectability for the given sensor (like MAD, visual, radar, etc) and the large number associates that level of detectibility to a range, in meters, for which an object with that level of detectability would be detected. So, for an object with a radar signature of 80, being searched for by a radar with curve 100,000-100, 80,000-80, 45,000-50, 10,000-25, 5,000-0, the object would be detected at 80,000m. For values that fall between the stated values, like 90, the curves are linear, so an object of 90 radar signature would be detected by this radar at 90,000m.

In terms of getting a sense of how the sonars work, my advice is to compare the Sphere and Towed Arrays, since these are, in a way, the most different. Start to mess around with the Nrd and the Frequency ranges for an American platform (so you can see the effects clearly on the waterfall). You'll notice that even if you set the Nrd for the Sphere much lower (negative values are more sensitive) than the TA, the TA will still out perform the Sphere in terms of range, but the BB contacts will show up more clearly on the Sphere. It's a bit of a complicated starting point to look at this stuff first, but you might want to compare the values and in game performance for stock DW, and how the Sphere performs in the LWAMI 3.00b Mod.

BTW, Amizaur and I are looking for someone to do the detection curves for radars, if you decide that you have gained proficiency in the editor and understand thoroughly how database changes influence game behavior (won't be sooner than a couple weeks or months...) and don't mind spending hours and hours of your time doing detail oriented tasks, strictly "for the glory." :wink:

RedDevil
12-28-2005, 02:22 AM
There are already some great guides out there that explain how sonar works, and techniques to use with different subs to avoid detection, or evade torpedo's etc. etc. Unfortunately all these guides are for the real world, which seams to be modeled fairly accuratly in DW. However, not enough hard data is supplied to you as a captain of any of the playable platforms to be sure of what you are actually capable of "in game". I think any compatent Captain of any submarine in the real world, would know how to ring as much performance out of his ship as possible, including exaclty how your speed affects both your profile, and how you can detect. Whether he actually does this or not is another story, but I think he would at least know! For some reason we are not told this though.

So, seeing that it hasn't been done yet, I decided to figure it out on my own.

I was going to write a detailed guide on how the sensor systems in general work in game, including everything from maximum range, max speed to still get a signal, the SNR (which I already assumed was used as variable to artificially simulate descrimination through background noise). As part of this, I would also explain how the frequency range that each system detected mattered, and more specifically how it is that both the Sphere and Towed arrays had the same max range, yet the towed array would still detect sound signals farther away. This is due to the game engine modelling fairly well how lower frequency signals propogate farther in water. I would then make a quick sheet, which showed max speeds for each sensor, and a comparison graph to show how much noise you would make at which speed you were traveling, in comparison with all the other subs. (IE a 688(i) at 10kts is as noisy as a SSN21 at 15kts for example...).

The other sensors, such as MAD, radar, and even visual I would address after. I would then followup with effects of countermeasures, bouys, missiles and other misc items.

The only things that I didn't really know for sure was what the numbers in the Detection Curves actually stood for, and it was holding up the whole project!
Was it actually a graphical "curve" or a hardcap? And what was the second number for? I had assumed a hard cap, but the second number threw me for a loop thinking that maybe it was an X-Y 2D graph of some sort (hence the "curve" part), and I couldn't be certain if the hardnumber was in metres or not, as it produced a rather large Nm range when I converted it.

You now cleared up how the Detection Curves work, so I think I pretty much have the entire system nailed now.

RedDevil
12-28-2005, 02:27 AM
I'd be happy to help out. Check your PM inbox when you get a chance.

Cheers,
RedDevil.