Results 1 to 2 of 2

Thread: Countermeasures in DCS

  1. #1
    492d vFS Pilot Beardy's Avatar
    United Kingdom
       United Kingdom
    Join Date
    Jul 2016
    Posts
    198

    Awards Showcase

    Thanks Thanks Given 
    227
    Thanks Thanks Received 
    206
    Thanked in
    87 Posts

    Countermeasures in DCS

    Whilst working on the F-15E syllabus, I went down a rabbit hole of threat counter tactics and DCS modelling of seekers and countermeasures. This post is an attempt to collect the various bits of information I picked up and consolidate them into one place for the group to make use of in future.

    Existing literature:

    Starting with Scaley's hornet countermeasures post, and the linked studies by RocketmanAL, I was able to start getting some insight into the way the countermeasures are modelled. The Hornet study link was dead, but the A-10C study and the AV-8B study provided some initial key insights:

    1. For the threat system examined, flares per defeat is insensitive to the dispense program.
    2. Success rate increases with number of flares in a fashion consistent with repeated independent probability checks.
    3. Dispense rate over time has no effect on the probability of defeat, unless it becomes similar to the time of flight of the missile, at which point decoys leaving the seeker FOV will be a factor.
    4. The trends from the data don't match nicely from one aircraft to the other, suggesting other confounding variables not controlled across both studies (discussed below). They do however follow the same shape.


    Digging through the currently exposed lua files and work from Quaggles lead to determining that each system has a ccm_k0 value that specifies its countermeasure resistance and that IR weapons also have a seekerSensitivityDistance, which is the distance at which a target with an IR signature of 1 can start to be tracked. Each platform also has an RCS value, an IR value without AB and an IR value with AB.

    Further searching on this forum led me to ancient texts old extracts of lua files that are no longer exposed in the current version of DCS. This indicated the existence of a list of probability coefficients used in deciding if a weapon will accept a decoy. This led to further searches in the Russian parts of the ED forum.

    Other titbits of information include that the Hornet and Viper seem to use a smaller 1x1x8 flare that is given around 60% of the effectiveness of the 2x1x8 flares. There seems to be no effort to consider the chemistry/composition of flares in determining their baseline probability, just that more mass must be better.

    Chaff is supposedly treated by seekers differently from how it is rendered in the game. It would also seem that the cloud update means that weather has no impact on EO/IR seekers.

    The AIM-120, AIM-54 and AIM-7 seem to no longer use any of the modelling discussed in this post and we might even see improved radar and IR seeker models in the soon(tm) future.


    The DCS model:
    My take on the information available suggests to me that a probability of decoy calculation starts with determining what platforms and decoys are in the seeker's field of view. The independent probabilities indicated earlier suggest that each entity is then iterated over independently, with modifiers added or subtracted to the probability roll based on the circumstances of the case.

    The modifiers seem to include:
    ccm_k0 - Listed on a weapon-by-weapon basis, it may be the same as what k0 used to be, or an independent probability modifier/multiplier.

    k0 - A base probability of seeker seduction to a different target, or a modifier?
    k1 - Another variable that seems to be involved in the probability of seduction but remains unclear.
    k2 - Same again, but possibly for IR missiles?
    It is possible that only one value is selected depending on the seeker type.

    Then a series of coefficients seem to exist for Radar missiles:
    k3 - Probability of seduction when the weapon is approaching the target head-on.
    k4 - Probability of seduction when the weapon is approaching from the beam.
    Of note is that in the examples I found, the beaming value was 2000x higher than the nose-on value.

    k5 - Radial velocity of the target above which k3 is used.
    k6 - Radial velocity of the target below which k4 is used.
    Comments from the devs suggest that this is a step change rather than linear interpolation, and the k5/k6 is often as low as 10 or 20 m/s. What is unclear is what happens between k5 and k6 when there is a gap between the two values.

    Then a single IR modifier:
    k7 - This seems to simply subtract from the base IR signature in the forward quarter, and add to the base IR value in the rear quarter.

    Finally, a pair of modifiers exist for what Google translates as "target shielding" which seems to be the scenario when the decoy is between the seeker and target.
    k8 - Modifier for radar seekers.
    k9 - Modifier for thermal seekers.

    Comment from the devs suggests that the probability of defeat is increased by the modifier when you are opening with the seeker and reduced by the modifier when closing with the seeker, despite having the decoy still in front of you. This may be a mistranslation and instead be a case of reduced if you are in front of the decoy from the threat seeker perspective.

    In addition to this, the distance to each object seems to have a weighted modifier, with closer objects increasing the probability of defeat. It is unclear whether this is linear, r squared, or r to the 4, but seems to be an attempt at J/S ratio.

    The background scene is thought to matter, and is commented on occasionally in the old Russian forum threads. I think this is limited to ground clutter/lookdown for the RF threats and the Sun for EO/IR.

    It is unclear whether flares have a contrast lock defeat mechanism on automatic optical trackers. I couldn't find evidence either way and will need to test it in-game.

    Some testing and rough math:
    This led to some testing and rough math:
    An F-15E was made to orbit a threat site at a fixed speed and throttle setting (or as best as the pilot could manage), trying to maintain a constant DME and altitude with the threat on the beam to control for aspect.

    A variety of threats were used SA-18 (IGLA & IGLA-S which seem to both use the IGLA-1E database entry), Stinger, SA-9 and SA-13. The sample size was small with 12 firings for each threat (14 for one of the SA-18's). The pilot was briefed to drop 18 flares (3 packs of 6) in quick succession on seeing the launch. They were alerted to each launch and knew the location of the threat beforehand yet would often only see the threat in the second half of the time of flight.

    The SA-18s and Stinger have near identical database values of ccm_k0 = 0.5 and SeekerSensitivityDistance (SSD) of 10.5km and 10km respectively. This gave 92% probability of defeat across 38 launches. The SA-9 has a ccm_k0 of 1 and an SSD of 13km which suffered a >95% probability of defeat (all valid shots defeated, 12 launches tested). The SA-13 with ccm_k0=0.5 and SSD=15km was only defeated in 83% of 12 launches suggesting that the probability of defeat scales with range relative to SSD, although a larger sample size would be better.

    Assuming that all aspect-related variables (k7) are equal for each aircraft, I estimate that a rough but suitable function for threat analysis and countermeasure development for IR threats would go as follows:

    PDefeat = 1 - [1-(DecoyP * (10km/SSD) * (1/IRSignature) * ccm_k0)]^NumberFlares

    DecoyP for the 2x1x8 would be around 0.2, and 0.12 for the 1x1x8 flares.

    This isn't exact and will overestimate the threat by a bit, but would hopefully help guide the selection of countermeasure profiles for a given threat profile. Feel free to critique and improve.

    Further work is needed to develop an RF equivalent and for Air-to-Air IR seekers.

    So What:
    The following conclusions can be drawn out, most of which we have already figured out anyway:
    • Dump countermeasures as quickly as possible to maximise time to assess and re-lay more countermeasures if required. There is no benefit to fancy deployment patterns or delays.
    • Whilst pre-emptive countermeasures do nothing to seekers that haven't been fired. My thinking is that most weapons are not detected by the pilot until a while after launch so they feel like they work when in fact they are working in the interval between launch and detection. I'd suggest using them if you have lots of them and the threat is susceptible to small numbers of them. (e.g. early IR/RF threats). Don't waste your CMs on pre-emptive when in a high-threat environment as you'll have to dump an awful lot of them to achieve reliable probability of defeat.
    • For RF threats the strongest sensor defeat will be to lay chaff through a curving notch (very tight tolerance for some threats), followed by a nose-low slice to get down-range of your chaff and below the horizon in the threat seeker and re-lay some more with the benefit of ground clutter. This should stack the DCS modelling of notching, screening and clutter effects in your favour.
    • For IR threats, the best sensor defeat is lots of flares and turns to get the threat away from the tail.
    Last edited by Beardy; 09Oct23 at 14:23.

  2. The Following 16 Users Say Thank You to Beardy For This Useful Post:

    Ahmed (09Oct23), Barney (15Oct23), Ben (10Oct23), Cheezemad (10Oct23), Connellie (09Oct23), Jester (15Oct23), Kampf (10Oct23), Karma (11Oct23), Loco (09Oct23), Oliver (09Oct23), Puma (11Oct23), Radge (15Oct23), Roman (09Oct23), Scaley (15Oct23), Swift (09Oct23), Tihi (11Oct23)

  3. #2
    Senior Member Scaley's Avatar
    United Kingdom
       United Kingdom
    Join Date
    Jul 2015
    Posts
    1,190

    Awards Showcase

    Thanks Thanks Given 
    506
    Thanks Thanks Received 
    1,194
    Thanked in
    517 Posts
    Beardy, thanks for a great post. That essentially all tracks with the research I did when I wrote the Hornet profiles, and the (very non-exhaustive) testing I did then for the Hornet, and more recently for the Apache. You conclusions seem to be supported by what I got from Apache testing with IR threats where the same rules applies (many flares very quickly is the best tactic).

  4. The Following User Says Thank You to Scaley For This Useful Post:

    Beardy (15Oct23)

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •  


Like our website?

You can help us by donating to cover our costs.

Many sincere thanks!


Search

Follow us

Twitter Twitter youtube