already reported AIM-120 Midcource Guidance Desync

[dundun92]

AIM-120 midcource guidance is often not being synced between clients, reuslting in the missile position being quite different between the shooter and target. In this example, the AIM-120, shot by Sizzle, on my end as defender had no midcourse updates, lofted to 60kft and quite abruptly "snapped" active at around 10nm, appearing to pass quite far behind me (the RWR indicated this as well). This is also what was shown on the server track/tacview:

Meanwhile, on the shooter (Sizzles) end, the AIM-120 appears to guide normally with midcourse, only lofting to around 53kft because of me diving, conserving energy, with no abrupt snap upon active. It passes in the end quite close to me (this is what is shown on sizzle and his wingman's track/tacview):

tracks (from me, sizzle, his wingman, and server) and tacviews (just me and sizzle for illustrative purposes) attached

server.acmi sizzle_shooter_client.acmi dundun_defender_client.acmi moldy_wingman_client.trk sizzle_shooter_client.trk server.trk dundun_defender_client.trk

[Mareno]

bump

[comie1]

Yeah BUMP!

I've pretty much quit playing DCS for now due to this paticular bug and the lack of improvement in AIM 120s over the past few years. 

[Mareno]

@BIGNEWY@NineLine ping because it seems like this was completely overlooked by accident

[Маэстро]

Looks like this desync was fixed a few months ago, but for some reason was not included into public DCS version. Will be available in the next update.

[okopanja]

18 minutes ago, Маэстро said:

Looks like this desync was fixed a few months ago, but for some reason was not included into public DCS version. Will be available in the next update.

Was this amraam only issue, or did it apply to all missiles?

I would think this is a common network code, since I observed difference in the past when firing 2 missiles (e.t. ET, followed at 77). Server tacview showed me ET hit the first while 77 missed, while tacview generated by replying my own track, showed that 77 did in fact hit and kill the target. Just to be clear: not submitting a bug report or derailing the topic, but likely you might know the answer, which would tell us if it makes sense to continue watching for issue in future.

Edited February 6, 2023 by okopanja

[Маэстро]

21 hours ago, okopanja said:

Was this amraam only issue, or did it apply to all missiles?

I would think this is a common network code, since I observed difference in the past when firing 2 missiles (e.t. ET, followed at 77). Server tacview showed me ET hit the first while 77 missed, while tacview generated by replying my own track, showed that 77 did in fact hit and kill the target. Just to be clear: not submitting a bug report or derailing the topic, but likely you might know the answer, which would tell us if it makes sense to continue watching for issue in future.

 

That was particular amraam issue. R-77 and R-27 use different missile API and can not be affected by this desync.

[okopanja]

2 hours ago, Маэстро said:

That was particular amraam issue. R-77 and R-27 use different missile API and can not be affected by this desync.

Thanks, this is interesting, I will keep closer look for future events with other missiles and report if I manage to get the reasonably small trk file.

[firimar]

it happens with all missiles, including surface to air

[_SteelFalcon_]

i'll just leave this here. DTT launch, 2 targets, 2 amraams, 1000000000 questions.

 

amraam_DTT_drunk.zip.acmi

[skywalker22]

Damn, just shot 6 of them from close range (hot 5-10nm), and none hit the target... waiting for server`s tacview, but something is definitely not ok here.

This is getting worse and worse, lost all the hype of flying DCS.

[Ghosty141]

Am 7.2.2023 um 18:14 schrieb _SteelFalcon_:

i'll just leave this here. DTT launch, 2 targets, 2 amraams, 1000000000 questions.

 

amraam_DTT_drunk.zip.acmi 1.6 MB · 6 Downloads

I've seen this multiple times already, the missile somehow doesn't try to intercept but rather catch up which ends up with it going parallel to the target. Not sure if this is intended or not.

[skywalker22]

Yesterday on GS server - What do you guys think of my amraams.

Red F-16 (Skywalker22)

Tacview-20230209-145926-GS-server.txt.acmi

PS: I would dare to say, that this is also the price we have to pay on overcrowded servers.

Tried amraams on 3 different servers with 10-20 people, and all was fine, as expected.

--

As @Ghosty141 said, it really seem that the missile tries to fly parallel with it's target, and not trying to intercept it - in most cases. You can see the proof of that on my example here as well.

Edited February 10, 2023 by skywalker22

[_SteelFalcon_]

58 minutes ago, Ghosty141 said:

I've seen this multiple times already, the missile somehow doesn't try to intercept but rather catch up which ends up with it going parallel to the target. Not sure if this is intended or not.

the focus in that tacview is more on "what in the world is the 2nd missile doing?" it turns around 180 for no freaking reason.

[Ghosty141]

My understanding: The missile checks the course vector of the target and extends it as if it will keep flying in a straight line. If the target now keeps turning, the course line moves with it making the missile adjust to the point where it thinks it will intersect with the target (which it assumes keeps flying straight) which is far infront of it. That constant adjusting bleeds a lot of energy depending how the turn by the target is timed.

You can see this VERY well if you turn on the "course" vector in Tacview and go to the cockpit view of the missile. (see attachment)

It's very understandable that it works this way, how would the missile know its target is going to keep turning. Although I believe the real one might have some smart heuristics up it's sleeve. Some that I can quickly come up with: If a plane turns with 5g -> it will most likely keep doing that turn for a certian amount of time, or if the target is in a steep dive it can only do it for a certain amount of time until it hits the ground so it only makes sense to aim down a certain degree.

It would be amazing if a dev could clarify this since I personally haven't found any posts about how the guidance works.

Edit: I also added an attachment that shows me evading a aim120c shot from 34k ft/M1.5 at me (6.5k ft/M0.95). I only start my defensive maneuver once I hear the missile warning tone at ~8nm distance from the missile. Basically: turn to get the missile on your 6 while keeping it looking far down the course vector, once it its on your 6 turn again in the opposite direction.

aim120c_defense.acmi

Edited February 10, 2023 by Ghosty141
add trackfile

[_SteelFalcon_]

12 hours ago, Ghosty141 said:

My understanding: The missile checks the course vector of the target and extends it as if it will keep flying in a straight line. If the target now keeps turning, the course line moves with it making the missile adjust to the point where it thinks it will intersect with the target (which it assumes keeps flying straight) which is far infront of it. That constant adjusting bleeds a lot of energy depending how the turn by the target is timed.

You can see this VERY well if you turn on the "course" vector in Tacview and go to the cockpit view of the missile. (see attachment)

It's very understandable that it works this way, how would the missile know its target is going to keep turning. Although I believe the real one might have some smart heuristics up it's sleeve. Some that I can quickly come up with: If a plane turns with 5g -> it will most likely keep doing that turn for a certian amount of time, or if the target is in a steep dive it can only do it for a certain amount of time until it hits the ground so it only makes sense to aim down a certain degree.

It would be amazing if a dev could clarify this since I personally haven't found any posts about how the guidance works.

Edit: I also added an attachment that shows me evading a aim120c shot from 34k ft/M1.5 at me (6.5k ft/M0.95). I only start my defensive maneuver once I hear the missile warning tone at ~8nm distance from the missile. Basically: turn to get the missile on your 6 while keeping it looking far down the course vector, once it its on your 6 turn again in the opposite direction.

aim120c_defense.acmi 101.36 kB · 2 downloads

 

So explain this then: We'll focus on those 2 amraams. both migs move relatively similar, no crazy maneuvering at all.
1 amraam (marked green) flies a what i'd consider logical and very understandable course. if it hits or not doesnt matter.

null

but the other amraam (marked red) for some reason starts pulling 29.2G on a target that is not maneuvering at that point at all, according to the targets future course line.

you can see this very well with the LOS & Course lines enabled (which I always enable anyway).

this to me seems very odd and not that understandable. The missile ends up turning 180° and basically pointing away from the target before the target even starts doing any maneuvering.

Edited February 11, 2023 by _SteelFalcon_

[okopanja]

15 hours ago, skywalker22 said:

Yesterday on GS server - What do you guys think of my amraams.

Red F-16 (Skywalker22)

Tacview-20230209-145926-GS-server.txt.acmi 4.55 MB · 4 downloads

PS: I would dare to say, that this is also the price we have to pay on overcrowded servers.

Tried amraams on 3 different servers with 10-20 people, and all was fine, as expected.

--

As @Ghosty141 said, it really seem that the missile tries to fly parallel with it's target, and not trying to intercept it - in most cases. You can see the proof of that on my example here as well.

 

I took a look at your recording. From what I can gather it is not your amraams that are affected but also those fired by your opponents at you. At the same time I did notice that your aircraft exhibits a step like behavior, indicating that from the point of server it does not report often enough. Just to eliminate the possibility lag on server, you will notice that player Moltar has perfectly smooth trajectories. Given the status of sort of a celebrity, you can expect him to have a high end configuration, and he is also in USA having much lower ping than you who are likely in Europe.

In your case: either your network has issues or your PC is to weak. Consider reducing the level of details to achieve higher frame rates.

 

[Ghosty141]

vor 3 Stunden schrieb _SteelFalcon_:

So explain this then: We'll focus on those 2 amraams. both migs move relatively similar, no crazy maneuvering at all.
1 amraam (marked green) flies a what i'd consider logical and very understandable course. if it hits or not doesnt matter.

null

but the other amraam (marked red) for some reason starts pulling 29.2G on a target that is not maneuvering at that point at all, according to the targets future course line.

you can see this very well with the LOS & Course lines enabled (which I always enable anyway).

this to me seems very odd and not that understandable. The missile ends up turning 180° and basically pointing away from the target before the target even starts doing any maneuvering.

 

I agree with the other poster, don't forget that missiles ONLY rely on the shooters connection, meaning if you experience some kind of lag for a short time it may affect your missile. After the lag it may have lost lock and "gave up" (iirc once it loses lock and doesn't get new information it's basically dead, if it does circles or whatever doesnt matter at that point).

[_SteelFalcon_]

51 minutes ago, Ghosty141 said:

I agree with the other poster, don't forget that missiles ONLY rely on the shooters connection, meaning if you experience some kind of lag for a short time it may affect your missile. After the lag it may have lost lock and "gave up" (iirc once it loses lock and doesn't get new information it's basically dead, if it does circles or whatever doesnt matter at that point).

the other poster is not talking about the same tacview. if you watch skywalkers tacview, it is pretty obvious he's lagging.

I wasnt lagging at all in my case nor did it affect the first amraam fired at the same time

the excuse of everything being a connection issue is really getting old

[okopanja]

44 minutes ago, _SteelFalcon_ said:

the other poster is not talking about the same tacview. if you watch skywalkers tacview, it is pretty obvious he's lagging.

I wasnt lagging at all in my case nor did it affect the first amraam fired at the same time

the excuse of everything being a connection issue is really getting old

That is correct, and just to add on top: all missiles are affected by these class of problems. The missile can guide as accurately as it is the known position of the target. If the target reports less frequently the guidance will be less accurate and lock may be lost.  In this case root cause is the at shooter's end. Therefore @skywalker22 's example should not be considered within this topic.

The original issue, however, is actually caused by the game itself. In this case there is no lag, but we still witness 3 different bubble universes.

As I stated in my original post about R-77 and R-27ET having a race condition, one can only conclude that this might be a deeper problem in the way the DCS multiplayer works.

The totally synchronous behavior outside of the LAN network is very difficult to achieve, since by the nature not all of the clients report with the same "tact". Therefore all sides have their own universes, which get simulated based on the last known parameters, with server probably acting as the mediator, by sending the update. In such situation is totally possible that 3 different parties calculate totally different trajectories.

One thing to consider is also having updates at 20ms vs 200ms, given the speeds involved in calculation can cause a rather large calculation differences.

Maestro correct me if I am wrong here.

[skywalker22]

6 hours ago, okopanja said:

I took a look at your recording. From what I can gather it is not your amraams that are affected but also those fired by your opponents at you. At the same time I did notice that your aircraft exhibits a step like behavior, indicating that from the point of server it does not report often enough. Just to eliminate the possibility lag on server, you will notice that player Moltar has perfectly smooth trajectories. Given the status of sort of a celebrity, you can expect him to have a high end configuration, and he is also in USA having much lower ping than you who are likely in Europe.

In your case: either your network has issues or your PC is to weak. Consider reducing the level of details to achieve higher frame rates.

 

How can you tell the difference of the smooth/not smooth trajectories from tacview? By just observing it? I wouldn't take that as granted, as to be 100% accurate at all times.

I am from EU, and I have ping cca usually 100-200ms. I surely cannot have same as those from the states/canada. And I agree on your last statement about calculations, but still, missile should still try to go into intercept trajectory (sooner or later (we talk in ms or seconds) and not just follow up it's target (which happens in many occasions - from me and to me). But again, latency can be a bi*tch, specially at this very high speeds, I won't deny it.

Edited February 11, 2023 by skywalker22

[okopanja]

23 minutes ago, skywalker22 said:

How can you tell the difference of the smooth/not smooth trajectories from tacview?

Select your aircraft  zoom in. You will notice there are straight lines. Replay slowly and you will notice these micro jumps.

Toward end even Moltar had this to much smaller degree.

In contrast dundun92 had smooth expirience, but still 3 different universes.

[skywalker22]

Well, since you were saying about this celebrity Moltar guy and his smooth trajectories, can you then explain AMRAAM's behavior from the video bellow. I went watching him, and I discovered this weird thing, just 2 minutes after we "met" in the sky - and btw, he should be dead by my missile, but miraculously my missile has gone stupid, despite perfect parameters (he wasn't doing anything to it that it should went stupid - but ok, lets leave that).

 

Never ever seen that something before.

Edited February 11, 2023 by skywalker22

[okopanja]

2 hours ago, skywalker22 said:

Well, since you were saying about this celebrity Moltar guy and his smooth trajectories, can you then explain AMRAAM's behavior from the video bellow. I went watching him, and I discovered this weird thing, just 2 minutes after we "met" in the sky - and btw, he should be dead by my missile, but miraculously my missile has gone stupid, despite perfect parameters (he wasn't doing anything to it that it should went stupid - but ok, lets leave that).

 

Never ever seen that something before.

 

You were already stuttering yourself at that moment. You see it's not only about the data at which you send updates on your own movement, its also how fast and how often you receive.

nullDo you remember if you blacked out at this moment? You had the load of 20G, plus were teleporting through space.

 

 

[skywalker22]

No, I didn't pass out. This must be something wrong with the telemetry data, 20Gs (also on my as well). How is that possible? On the HUD max I ever saw, was 9,8 or something. 2x more? This has to be somehow connected with the latency. At one time there is 0G, next second is 20G (or some way higher number then 9).

Example: 

And I was pulling Gs all the way, despite tacview draws a straight line, and 0Gs. What I really pulled was a circle shape and max 9.2G at max - according to the HUD.

So here is the proof that DCS struggles in multiplayer, specially on crowded servers like GS is.