EVE Search - [Statistics] ECM and ECCM: Facts instead of feelings.

Check My IP Information

All Channels

Ships and Modules

[Statistics] ECM and ECCM: Facts instead of feelings.

» Click here to find additional results for this topic using Google

Monitor this thread via RSS [?]


Author	Thread Statistics \| Show CCP posts - 0 post(s)
Multimorph Gallente NailorTech Industries Ethereal Dawn	Posted - 2008.09.11 07:48:00 - [1] Introduction In all the discussions here on the forum concerning ecm and especially falcons, two points are often made: one is being 'permajammed', i.e. being taken out of a fight completely by enemy ecm, the other is that eccm is a useless module. I am pretty new to EVE, playing for around 5 month now and have very limited pvp experience. So i cannot contribute to this discussion in terms of experience. However, there is pretty solid mathematical basis for ecm: the chance of being jammed is simply: jammer strength of jamming ship / sensorstrength of victim ship I thought that one should be able to derive some data about jamming propabilites out of this, providing some facts that might be useful in the discussion. When dealing with statistics and propabilites, intuition and personal experience are often very much misleading so a felt that some solid calculations might be helpful in this matter. Addressing the effect of ECM I adress the effectiveness of ECM by percent of time a ship is removed from a fight. In my opinion, this is a quantitiy that one can actually picture: if someone tells you, you will be jammed for 50% of the time of a fight, everybody knows what this means. The situation I look at is very much simplified: A single EWar ship is jamming a single victim over a certain amount of 20-second cycles. Parameters are the jamming strength of the EWar ship, the number of jammers used and the sensor strength of the victim. My idea was not only to see if something like "permajamming" or so is possible, but to compute what the possibilities are to remove a ship from a fight for a certain amount of time. For example i wanted to answer a question like: If a Falcon with jammer strength 13 is putting 2 jammers on my battleship with sensor strength 22 in a 200 second fight (=10 cycles), how likely is it i will be jammed for 50% (75%, 100% whatever) of the fight? How does this propability change if i fit an eccm? How does it change if the Falcon uses only 1 jammer or if it puts all 5 jammers on me? To compute these propabilities, i wrote a short and simple program in Matlab that can simulate such a situation and counts the total time the victim ship is jammed succesfully. The programs runs through a set number of fight cycles and checks by a random number the result of the jamming attempt for this cycle. If succesfull, the "jammed time" is increased by 20 seconds. The program also accounts for the time the target needs to re-lock: This re-lock time is set to a fixed number and if a ship is jammed for one cycle but the jam fails in the next, this re-lock time is also added to the total "jammed time" because the target, although no longer jammed, cant do anything while re-locking. By letting the program simulate a large amount of fights (i used a value of 10^6 = 1 million), you get a pretty solid estimate of the propabilites of different total jamming times. For those interested in it, the sourcecode of the program can be found below. Matlab is a pretty simple and generic language, so everybody with experience in programming will be able to understand the code.
Multimorph Gallente NailorTech Industries Ethereal Dawn	Posted - 2008.09.11 07:49:00 - [2] Edited by: Multimorph on 11/09/2008 07:50:24 Scenarios The fights simulated for the following figures lasted all 10 cycles or 200 seconds. The re-lock time was set to 10 seconds. The jamming strength of the Ewar ship is always 15, while the number of jammers varies. The base sensor strength of the victim is 22, modified by eccm in b) and d). a) A battleship with sensor strenght 22 and no eccm is jammed by a Ewar ship with jamming strength 15, using a single jammer So what can be seen here? The x-axis shows the percent of time the victim ship is removed from the fight by either being jammed or busy re-locking. So a value of 50 means the ship could only fire for 100 seconds while being incapacitated for the other 100 seconds. The y-axis is the propability that the victim is jammed this percent of time. Looking at the above figure, one can for example see that the propability of being jammed for 50% is around 0.02 or 2%. The vertical red line is the expected value( this is the average result if you compute more and more fights. For example, the expected value of rolling a dice is 3.5: if you roll it very often, every number will occur around 1/6 of the rolls. So the average result is (1/6)1 + (1/6)2 + (1/6)3 + (1/6)4 + (1/6)5 + (1/6)6 = 3.5 ). You can see that in almost all fights the ship will be jammed for more than half of the time, most likely somewhere between 70% and 90%. An actual permajam of 100% is possible but not very likely (around 2%).
Multimorph Gallente NailorTech Industries Ethereal Dawn	Posted - 2008.09.11 07:52:00 - [3] b) Now the battleship fits one eccm-II, increasing its sensor strength by 96%. The Ewar ship still uses only one jammer. (Note that as some values for jamming time can be obtained by more combinations of jamming and relocking than other, the graph shows "zig-zags". This is not important here and does not effect the discussion). The sensor strength has increased to 43.12 = 1.96*22 by using eccm. One can see, that the figure has significantly changed. The expected value dropped from almost 80% to around 47%. Permajamming can still happen but is now very, very unlikely. The highest propabilities are between 30% and 60%.
Multimorph Gallente NailorTech Industries Ethereal Dawn	Posted - 2008.09.11 07:53:00 - [4] Edited by: Multimorph on 11/09/2008 07:55:28 b) Now the battleship fits one eccm-II, increasing its sensor strength by 96%. The Ewar ship still uses only one jammer. (Note that as some values for jamming time can be obtained by more combinations of jamming and relocking than other, the graph shows "zig-zags". This is not important here and does not effect the discussion). The sensor strength has increased to 43.12 = 1.96*22 by using eccm. One can see, that the figure has significantly changed. The expected value dropped from almost 80% to around 47%. Permajamming can still happen but is now very, very unlikely. The highest propabilities are between 30% and 60%.
Multimorph Gallente NailorTech Industries Ethereal Dawn	Posted - 2008.09.11 07:53:00 - [5] Edited by: Multimorph on 11/09/2008 07:54:58 b) Now the battleship fits one eccm-II, increasing its sensor strength by 96%. The Ewar ship still uses only one jammer. (Note that as some values for jamming time can be obtained by more combinations of jamming and relocking than other, the graph shows "zig-zags". This is not important here and does not effect the discussion). The sensor strength has increased to 43.12 = 1.96*22 by using eccm. One can see, that the figure has significantly changed. The expected value dropped from almost 80% to around 47%. Permajamming can still happen but is now very, very unlikely. The highest propabilities are between 30% and 60%.
Multimorph Gallente NailorTech Industries Ethereal Dawn	Posted - 2008.09.11 07:54:00 - [6] reserved: I am still working at it here :)
Multimorph Gallente NailorTech Industries Ethereal Dawn	Posted - 2008.09.11 08:06:00 - [7] Edited by: Multimorph on 11/09/2008 08:13:45 Edited by: Multimorph on 11/09/2008 08:07:56 Originally by: Ethan Hunte so what you're saying is, ecm is over powered... SHOCKING. This is not what I wanted to say, actually I have not decided what this means. I think others here are much more able to judge these numbers. Originally by: Chienka I think a histogram would be more appropriate for displaying data by frequency, since you're looking at ranges and probability. Good work nonetheless. You are propably right, I just used the easiest way available for me to display the data. However, I hope it becomes pretty clear what the figures say. After all, connecting the points by a line is also not correct as one can not jam a ship for something like 49,7458568% of a fight so the propability should be zero there. However, I felt this makes the figures easier to read.
Multimorph Gallente NailorTech Industries Ethereal Dawn	Posted - 2008.09.11 12:08:00 - [8] Edited by: Multimorph on 11/09/2008 12:11:24 Originally by: Loree eccm isnt ment to make 1 ship immune to ewar. Its ment to ensure that a falcon is forced to use ALOT more jammers on you, ensuring that less people in your gang are subjected to ewar. Well, no one here claimed that eccm should make you invulnerable. But you made an interesting point here! If the victim fitted one eccm-II and the Ewar ship uses 3 instead of only one jammer, the result again becomes comparable to jamming a ship without eccm with a single jammer: Thats not entirely the same distribution as for the one jammer/no eccm case but it is quite similar. So to overcome the effect of eccm, the ewar pilot has to use three instead of one jammers. As pointed out correctly above, I assume that the ewar ship fitted the right racial jammers! I might try to extend to program a bit, so that i can account for multi and nonsuiting racial ecm, too. Shouldnt be very difficult. @The Tzar : Thank you very much for the kind words :) @The Tzar & Cpt Branko : This is really the kind of discussion I hoped to start with this post! Thank you both, keep on going!
Multimorph Gallente NailorTech Industries Ethereal Dawn	Posted - 2008.09.11 12:33:00 - [9] Edited by: Multimorph on 11/09/2008 12:35:14 Originally by: Nikunai Doesn't all this rely on the fact the falcon is left alone for the entire fight? Wouldn't the data completely change if one person in the gang MWD's towards it or makes any attempt to force it to cloak or leave the field? Oh wait, that would mean someone might lose out on a kill mail if they have to go after the falcon. It all makes sense now. Of course you are completely correct that this is a simplfified model. But what I wanted to compute was pure propabilities. How these translate into effects in a real EVE fight will then be a matter of discussion. Originally by: Hugh Ruka hmm ... I see 2 problems in your little analysis. 1. you are using a random number generator that's relevant for the system you are making your experiment on. this can differ from the one used on actual TQ hosts, also seeding is an issue. 2. 200 seconds is a very long fight atualy. Even if you get only 3 cycles off and successfull, people will scream "permajammed !!!111" So I would modify your experiment setup: 1. max 4 cycles 2. victim sensor strength 20 3. jammer strength 14.7, that's the max you get on a falcon without implants (are there any ecm strength implants???) and t2 rigs. still you have the problem of the RNG difference ... I am very convinced that differences in random number generators will not have any significant effect for these simulations. There are issues, but I very much doubt that they will come into play in such a simple computation. Basically, the RNG in Matlab and EVE has to compute random numbers between 0 and 1 that are sufficiently equally distributed. I doubt that in one of these two are errors that of the order of magnitude of 0.01 percent or so which is the error my simulations make by running only a finite number of examples. I will try the changes in parameters you suggest and come up with some more figures. Now that I have to program, this can be easily done.
Multimorph Gallente NailorTech Industries Ethereal Dawn	Posted - 2008.09.11 12:47:00 - [10] Edited by: Multimorph on 11/09/2008 12:48:17 @Hugh Ruka Here are the results for a 80 seconds fight (=4 cycles) with 1 jammer and no eccm (a) and one eccm (b) respectively. While the distributions change a bit, the expected values are comparable. Overall, the difference between a 4 cycle and a 10 cycle fight do not seem that big?! (a) (b)

Multimorph Gallente NailorTech Industries Ethereal Dawn	Posted - 2008.09.11 14:13:00 - [11] Edited by: Multimorph on 11/09/2008 14:15:14 Originally by: Ampoliros How much of the percentage out of the fight is relocking time, and how much is ECM'd time? While this could be computed during the simulation, I have no idea how to display this nicely in the figures. However, I tried a "lowchance" scenario as Cpt Brano suggested: One jammer with strength 10 against sensor strength 30. a) is with a locking time of 10 seconds, in b) the relocking is instant. a) Relocktime 10 seconds b) Instant relock The effect is indeed quite noticable. Hm, propably i should "bin" the times in a different way to simplify the comparison of these figures. I might go with bins of 0-10%. 10-20% etc. Will see...
Multimorph Gallente NailorTech Industries Ethereal Dawn	Posted - 2008.09.11 14:47:00 - [12] Originally by: Crazy Tasty Originally by: Multimorph Edited by: Multimorph on 11/09/2008 08:32:19 Edited by: Multimorph on 11/09/2008 08:12:44 d) Again the battleship fits an eccm-II, the ewar ship still uses 5 jammers with strenght 15 With 5 jammers, the effect of a single eccm is a lot weaker here. The expected value is still clearly over 90% and being jammed for less than 60% is unlikely. The chance for being permajammed dropped to 30%, but still the most likely results are jamming times over 90%. I fail to see the issue there. If I had fit my EW ship completely to deal with a single ship why should he not be perma-jammed? You can do basically the same thing with EW Recons when you fit 100% to disable 1 other ship. I agree with this completey! The figure was made to explore the more extreme scenarios possible, it is not meant as an argument that ECM is "overpowered". I dont share this opionion (yet?), I am just trying to get a more complete picture of what ECM can do and what it can't do.
Multimorph Gallente NailorTech Industries Ethereal Dawn	Posted - 2008.09.11 16:10:00 - [13] Originally by: Roemy Schneider i'd like to check my own stochastic results... could you plot me 14.5 jam 2 jammers (any more racial jammers isnt plausible) 4sec relock 17 and 33.3 sensor (scimitar and overloaded eccm) 200 sec battle and 1.5 jam (ec-600) 20 jammers 4 sec relock 17 & 33.3 sensor 200 sec battle I will glady do this - but tomorrow. I dont have Matlab installed at home, so I have to run it at work tomorrow. Originally by: Trevor Warps This Multimorph's thread gets the Trevor Warps stamp of approval. STAMP What ever that means - thank you very much ;)
Multimorph Gallente NailorTech Industries Ethereal Dawn	Posted - 2008.09.12 07:04:00 - [14] Edited by: Multimorph on 12/09/2008 07:10:40 Edited by: Multimorph on 12/09/2008 07:09:17 Originally by: Roemy Schneider i'd like to check my own stochastic results... could you plot me 14.5 jam 2 jammers (any more racial jammers isnt plausible) 4sec relock 17 and 33.3 sensor (scimitar and overloaded eccm) 200 sec battle and 1.5 jam (ec-600) 20 jammers 4 sec relock 17 & 33.3 sensor 200 sec battle Here we go (results binned into 0-10%, 10-20%, ... , 90-100% because this way they are much easier to compare). a) (no eccm) and (overloaded eccm) b) (ECM Drones) (no eccm) and Wow, the ECM drones do a nice job. But how do get 20 of these little things out there? @Discussion: I really do not see why it is a problem, if a ECM ship pointing all his jammers onto a single target can put this target out of fight for nearly 100% of the time?! In my opinion, the ECM pilot would be plain stupid to do this, so if i was the victim I would sit back happily and watch my fleet members melt the enemies faces while every one of our ECM ships takes out 3-4 ships for 50-60% of the fight. To be honest, if anyone wants to argue for the "overpowerdness" of ECM, I would follow the line of Cpt Branko and stick to the figures with a single jammer. Especially the point that a eccm-fitted battleship is still very likely jammed for around 50% of a fight with a single jammer is worth discussing I think.
Multimorph Gallente NailorTech Industries Ethereal Dawn	Posted - 2008.09.12 14:15:00 - [15] Edited by: Multimorph on 12/09/2008 14:16:09 Well, this going pretty fast here now:) Just want to make sure I understand what you guys have in mind: Your idea is to let the length of the jamming cycle depend on the jammerstrength/sensorstrength ratio? I.e. if I jam a ship with low sensorstrength it will be jammed for a longer time than a ship with high sensor strength? And now you try to find a formula for the dependance of jamming time on the jammer/sensor strength ratio that reproduces the effect of ecm as it is if applied to a non-eccm'ed ship?
Multimorph Gallente NailorTech Industries Ethereal Dawn	Posted - 2008.09.25 16:16:00 - [16] Standards in this thread seem to deteriorate quickly. Sad :(

COPYRIGHT NOTICE
EVE Online, the EVE logo, EVE and all associated logos and designs are the intellectual property of CCP hf. All artwork, screenshots, characters, vehicles, storylines, world facts or other recognizable features of the intellectual property relating to these trademarks are likewise the intellectual property of CCP hf. EVE Online and the EVE logo are the registered trademarks of CCP hf. All rights are reserved worldwide. All other trademarks are the property of their respective owners. CCP hf. has granted permission to EVE-Search.com to use EVE Online and all associated logos and designs for promotional and information purposes on its website but does not endorse, and is not in any way affiliated with, EVE-Search.com. CCP is in no way responsible for the content on or functioning of this website, nor can it be liable for any damage arising from the use of this website.