TWS as smooth as STT - ED Forums
 


Notices

Reply
 
Thread Tools Display Modes
Old 10-07-2019, 08:30 AM   #1
viper2097
Senior Member
 
viper2097's Avatar
 
Join Date: Jan 2019
Location: Austria - LOXZ
Posts: 1,717
Default TWS as smooth as STT

Since ever, the TWS lock is tracking a target as good as a STT lock.
Altough the TWS does not trigger a CL warning.

I brought this up first time two months ago, here is the discussion.

To explain it again:
In STT, the radar is all the time looking at the target.
Thats why, you get a real time, absolute smooth lock of the target on the HUD, and also on the radar screen.
We all know that, because that is the mode where you fire Fox 1 missiles.

In TWS, or LTWS, or SAM mode, a target is tracked, but the radar still sweeps up and down, and to the left and right as it does in RWS.
(not sure, but I guess the M2K radar then only sweeps left to right, so it goes to a 1bar scan)
So you do not get a real time designation of the target on the HUD and the radar screen.
The designation only gets updated when the radar sweeps over the target. So the track on the HUD and also the radar screen will be very stuttering and, mostly never be on spot.
There may be different mode of TWS, where the movement of the tracked target is also calculated and predicted, but if the target is manouvering, it must always be off.

I made a video for better understanding:


I know these are hard words, but I would already consider that behavior cheating. Especially in a PvP enviroment. As DCS is a simulation, please aknowledge that bug and fix as soon as possible. Thanks.
__________________
Steam user - Youtube

Last edited by viper2097; 10-07-2019 at 08:32 AM.
viper2097 is offline   Reply With Quote
Old 10-07-2019, 12:58 PM   #2
Ramsay
Senior Member
 
Join Date: Aug 2012
Location: UK
Posts: 1,935
Default

Quote:
Originally Posted by viper2097 View Post
So the track on the HUD and also the radar screen will be very stuttering and, mostly never be on spot.
No it won't, AFAIK most TWS radars will interpolate the contacts position between sweeps as the track file contains - position in 3d space, speed, direction of travel and altitude of the target.

Although it smooths out the movement of the HUD target box, etc. the main reason such interpolation is done is so the radar 'knows' where to scan in 3D space on it's next sweep and which contacts belong with which track file.

Quote:
There may be different mode of TWS, where the movement of the tracked target is also calculated and predicted, but if the target is manouvering, it must always be off.
You are correct, AFAIK most/all DCS modules do not yet model radar track files.

A real life technique is to fly in a smooth predictable way so an enemy's TWS radar has a good "solution" and then suddenly change your altitude and direction so the TWS radar looks in the wrong place on it's next sweep.

I don't know how difficult it is to implement the behaviour you ask for but it has yet to be done by ED themselves.

Last edited by Ramsay; 10-07-2019 at 01:20 PM.
Ramsay is offline   Reply With Quote
Old 10-07-2019, 02:02 PM   #3
Ramsay
Senior Member
 
Join Date: Aug 2012
Location: UK
Posts: 1,935
Default

Quote:
Originally Posted by viper2097 View Post
I made a video for better understanding:
DCS F-16C https://youtu.be/tHWc4og0DmI?t=58

AFAIK the DCS F-16C TWS behaviour you say is correct, is in fact incorrect.

• For a non-maneuvering target the F-16C HUD target box motion should be smooth.

• A hard maneuvering target that changes it's vector so it's not where TWS predicted, breaks the soft lock (the old 'track' file times out) and the pilot must make a new soft lock - TWS has no way of knowing if the 'new' contact is from the previous track or another enemy/friendly.

IRL you would be unlikely to use TWS during BFM due to it's low update rate and the ease of breaking it's 'soft lock', unless attacking an unaware bandit.

IMHO you should use IRL HUD tapes to argue for changes.
Ramsay is offline   Reply With Quote
Old 10-07-2019, 03:30 PM   #4
viper2097
Senior Member
 
viper2097's Avatar
 
Join Date: Jan 2019
Location: Austria - LOXZ
Posts: 1,717
Default

Whatever which TWS mode may have its details, the video is good to show what I'm talking about and that the actual TWS behavor in the Mirage just can't be correct.

If you find any HUD tapes of a TWS'ed contact of a M2K, feel free to share with us.
__________________
Steam user - Youtube
viper2097 is offline   Reply With Quote
Old 10-07-2019, 07:54 PM   #5
Ramsay
Senior Member
 
Join Date: Aug 2012
Location: UK
Posts: 1,935
Default

Quote:
Originally Posted by viper2097 View Post
If you find any HUD tapes of a TWS'ed contact of a M2K, feel free to share with us.
Here is a F-16 TWS AIM-120 shot, unfortunately the HUD footage is pretty bad but you can see the Radar screen contacts have 'jitter' as the predicted position is updated,

https://youtu.be/UKZjWT-i_Hk?t=89

... it doesn't do the large jumps seen in DCS's EA Hornet and Viper.

In part this is because ED's WIP HUD TWS target box is tied to the cockpit/aircraft datum, rather than a position/azimuth in 3D space.

The same issue applies to the Hornet's RWR and HARM sensor.

Hornet Bug Report: https://forums.eagle.ru/showthread.p...48#post3895048

Most HUD tapes from the M2K, F-16, etc. are BFM, by which point the pilots have switched to STT or Dog Fight mode.

CNN description of the F-16 AIM-120 shot: https://youtu.be/UrYcxXIg5fI

Last edited by Ramsay; 10-07-2019 at 08:05 PM.
Ramsay is offline   Reply With Quote
Old 10-07-2019, 11:53 PM   #6
viper2097
Senior Member
 
viper2097's Avatar
 
Join Date: Jan 2019
Location: Austria - LOXZ
Posts: 1,717
Default

Quote:
Originally Posted by Ramsay View Post
In part this is because ED's WIP HUD TWS target box is tied to the cockpit/aircraft datum, rather than a position/azimuth in 3D space.
I read that discussion already, understood it and hope they will fix it.

Quote:
Originally Posted by Ramsay View Post
Most HUD tapes from the M2K, F-16, etc. are BFM, by which point the pilots have switched to STT or Dog Fight mode.
Yeah HUD tapes with a good quality are hard to find, especially if you are searching for something special.

Luckily Elwood has replied on another thread that this issue will be adressed as Nr. 14 on his list. Thats good to hear that he is aware of it. Lets hope that this list also includes IFF and we will be seeing results soon(tm).
__________________
Steam user - Youtube
viper2097 is offline   Reply With Quote
Old 10-30-2019, 01:51 PM   #7
Elwood
3rd party developer
 
Elwood's Avatar
 
Join Date: Oct 2015
Location: Florence,Italy
Posts: 279
Default

Hello, I finally found time to have a look to this, so let me try to explain quickly, hope to give you and any other user that want to understand and not just play, more valuable information.

The RDI in TWS goes in single bar scan and follow the target with elevation.
So, it's true that raw data is limited by the number of swipes.
But let me explain a bit more how radar works and what has been simulated on the raw data elaboration.

Any raw data is processed, part of it, is a matching algorithm that will try to "old" contact with latest swipe contacts. If a match is found the target is confirmed to be the same and this happen any time if the contact is in TWS or not.
There are many others implications related to target signal, position, attitude, ground, interferences, position in the antenna cone, identification, Mirage position... some of them are classified some has been understood and reproduced.


But two of them are something you should consider in your analysis and video.
-One has been found, predicting target next position to facilitate the matching process, bear in mind that this prediction could also result in target out of radar frustum.
-Another one, the most important was not considered here, it's that every contact is ground stabilised, and that's I beleive is the results of the smoothness you see.

So everytime we draw a contact on VTB or HUD, this simple 2d coordinate, is the result of the computation of latest contact data plust Mirage position, antenna position, target attitude, prediction and ground stabilisation, this is why if the contact is lost or is out of cone it's still drawed with smoothness.

I personally think that Thales, few years back... did a miracle using analog and few digital components and here we are doing not bad into reproducing that with our possibilities (limited software computation and classified informations).

Let me add that without all of these you will probably have a WW2 radar.
If you want to know more on the subject just contact me on Discord.
Bye!
Elwood is offline   Reply With Quote
Old 10-31-2019, 08:35 AM   #8
viper2097
Senior Member
 
viper2097's Avatar
 
Join Date: Jan 2019
Location: Austria - LOXZ
Posts: 1,717
Default

Hi Elwood and thank you for the reply.

I understand what you said and it sounds logical for me.
While the radar is sweeping in 1 bar, it is still sweeping. Even it is predicting.
So the HUD box may still be smooth, but it can only be spot on when the radar sweeps over the target.
So lets come back to my video:
From 0:29 to 0:39, it may be correct that the box is smooth, as the target is nit maneuvering and exactley there where the computer predicts it.
But whats about 0:39 to 0:49? As the radar is stil sweeping, the target is maneuvering and the computer can't predict that, the box must cleary move away from the tracked aircraft.
Or please explain me, how the target can be tracked 100% smooth and correct, while the radar is not permanently looking at it?
__________________
Steam user - Youtube
viper2097 is offline   Reply With Quote
Old 10-31-2019, 10:26 AM   #9
Fri13
Veteran
 
Join Date: Aug 2013
Posts: 3,605
Default

It would be for many easier to understand a radar operation if they would go or when it is dark and take a flashlight with them. Start sweeping front of them in a120/60/30 degree arc at steady pace and think whole what they see if there would be something moving there.

Everyone would notice that 120 degree sweeping would be slowest, like if it would take 4 seconds to go from left to right (120°/4s = 30°/s) so every object in the scenery is visible only once every four seconds (for argument sake, let's think the flashlight beam is the 1.5° wide, and to give a context for that, if you extend your arm front of you and put thumb up then your thumbnail area is about 2-2.5° of your field of view). So going to 60° arc means you are now seeing everything every two seconds. Going for 30° is once a second.
Now, how many has been in disco and remembers those strobe lights flickering like 5 times a second?

If you would be in fist fight in that, you need to do a lot of predictions where the other person is so you can hit them. And if other does slight movement, it will throw off your capability to predict where they are.

Now, a radar that is scanning while tracking, needs to build a track. Track that is updated Everytime the radar completes its scanning pattern (left to right + X times as set by bar value. So example 120° arc in 4 seconds and 2 bars means it takes 8 seconds to create update) in that time the update is done the target can move where ever wanted and radar doesn't know anything where it is between that update time. Now if the same target is detected between two bars, it is marked as new contact so it becomes a ghost. And there are own ways to avoid that, so just to keep it simple. But between the updates radar tracking system generates two "gates" for each contact. One gate is the prediction of future position where the target will be in next time when radar "lights" it. And this gate starts big, as the target velocity, vector and distance are not accurate enough with one update. Typically it takes 3-4 updates to get target vector and speed accurately enough, but range is faster to get with 1-2 updates (rarely two). That gate prediction takes so on lots of time 4s * 3-4 = 12-16 seconds. And that is when target moves straight and staying in same speed. Now you have very good prediction where target will be in next X times of updates. And lets say you want to shoot a bullet (not missile) at it that takes 25 seconds to fly there, it requires to know very well where target is so it can hit it. A very complex thing to do at long range, but easier if it would, lets say 2 seconds to hit it.

That targeting gate is the intercept point where you need to aim to intercept the target. And it goes smaller and smaller (more accurate) on each update when target doesn't alter vector or speed. But if it does, that gate never gets smaller, it is never accurate as the position where it is heading on each update is changing and now your prediction of intercept point is huge, a total sum of all the previous calculated intercept points.

That is why there is a second gate, a "possibility gate". That is the gate that is calculated based all the information there is about the target, and that is used to generate gate that what are all the possibilities where the target can go between updates. So that is huge gate. And that is used when target lock is lost, as that gate area is then quickly scanned to find it again. It is like if you would be waving flashlight and person walks in park and suddenly he vanish, you know that person can not have gone far from last know location, so you start to wave flashlight in the general area of last seen position and where it might have gone in that time, like behind the tree next to him or the bush just little forward of last known location.
But you will give up that you have lost him and you were tracking him as you are waving light around him now, even if you don't see him.

The radar scope like the TWS smooth tracking is just a interpolation based prediction that where the target is between updates. It should get accurate if target velocity and vector doesn't change, but any slight changes should cause inaccuracy and only be accurate when radar updates target again, after it is again lost in accuracy. So someone doing a spiral toward you and you can't get anything from it really. Or two targets snakes at you in close formation (crossing each others flight path) and you can't build a track nor predict where target will be between updates. So if you try to shoot at them, or even show a target box on it, it will miss a lot, or you can't even track them.

Targets merging and separating causes radars lose the track. And that is easy to do when two targeting gates merger or they operate inside the same probability gate. Your Targeting data is invalid. And radars has problem to detect close formation flying targets as radar resolution ain't enough to detect them. Low bandwidth radar (L band) can have hundreds meters to thousands, and high bandwidth can have tens of meters to few. So if you can fly in close formation like 5 meters from each others in distance, you become invisible to radar or you become too look like one big target. And 3-4 fighters can be merged as one against X-band radar in nice close formation.

Now make that formation to change vector, speed and you can't get a valid taken data. Make the flight separate and merger, and track is lost. A flight of 3-4 fighters can approach a other fighter and deny then getting a weapon grade lock.
Trying to guide a missile to such target means it will fly at intercept point to where there are no one.

Add there all ECM functions that are changing the distance and position and speed of the target (group) and you will see in HUD how the targeting box will jump and lock is lost etc. And sooner than later, you are merged with multiple fighters and maybe get out one or two but rest gets you.

If the radar prediction box is smooth constantly in DCS, it is limitation in the radar operation in the DCS. As right now there is big problems in the radars as all seems to operate like a STT lock, even with a TWS.
__________________
i7-8700k, 32GB 2666Mhz DDR4, 2x 2080S SLI 8GB, Oculus Rift S.
i7-8700k, 16GB 2666Mhz DDR4, 1080Ti 11GB, 27" 4K, 65" HDR 4K.
Fri13 is offline   Reply With Quote
Old 10-31-2019, 11:29 AM   #10
jojo
Veteran
 
jojo's Avatar
 
Join Date: Oct 2008
Location: France
Posts: 3,669
Default

Please acknowledge the good values:
- radar beam is 3° wide.
- scan rate is 100°/s
- TWS is 1 line, 120° max so 1.2s refresh rate maximum. If you lower the scan width to 60° you guessed it, 0.6s.

Let's imagine the target is traveling at 300m/s.
If the target moves from 360m, at 18500m would it be obvious in the HUD target box ???

For the full 4 lines search pattern we can round up to 5s to scan the full volume. But no TWS here, no velocity vector, just contacts with closing velocity.
__________________
Mirage fanatic !
I7 7700K/ MSI GTX 1080Ti Gaming X/ RAM 32 Go 2400 Hz/ SSD Samsung 850 EVO/ Saitek X-55 + MFG Crosswind + Rift S
Flickr gallery:
https://www.flickr.com/gp/71068385@N02/728Hbi

Last edited by jojo; 10-31-2019 at 11:31 AM.
jojo is offline   Reply With Quote
Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump

All times are GMT. The time now is 12:16 AM. vBulletin Skin by ForumMonkeys. Powered by vBulletin®.
Copyright ©2000 - 2019, Jelsoft Enterprises Ltd.