open All Channels
seplocked Ships and Modules
blankseplocked [Statistics] ECM and ECCM: Facts instead of feelings.
 
This thread is older than 90 days and has been locked due to inactivity.


 
Pages: [1] 2 3 4 5

Author Topic

Multimorph
Gallente
Federal Defence Union
Posted - 2008.09.11 07:48:00 - [1]
 

Edited by: Multimorph on 11/09/2008 10:43:59
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 I 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 you tell someone, he 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
Federal Defence Union
Posted - 2008.09.11 07:49:00 - [2]
 

Edited by: Multimorph on 11/09/2008 12:14:51
If someone wants to use this in order to do some calculations on their own - feel absolutely free and encouraged to do so :)

Source code (feel free to skip it and go to the figures below)

function [time_jammed prop_t] = computeTpPropDistribution(fight_cycles, sensor_strength, jammer_strength, jammer_number, lock_time);

jamming_p = 1 - ( 1 - (jammer_strength/sensor_strength)).^jammer_number;

N_fights = 2e5; % number of fights to be simulated

time_jammed = [];

% This vector counts the absolute frequencies of the times stored in time_jammed. This simply means that if for an index of i.e. k = 3 we have time_jammed(k) = 40 seconds
% and abs_freq(k) = 5, then in all simulated fights there were 5 in which the ship was jammed for 20 seconds over the whole fight.
abs_freq = [];

lastcycle_jammed_bool = 0;

for kk=1:N_fights

time_jammed_thisfight = 0;

for jj=1:fight_cycles

x = rand;

% succesful jamming:
if x < jamming_p

time_jammed_thisfight = time_jammed_thisfight + 20;
lastcycle_jammed_bool = 1;

% failed jamming:
else

% if ship was jammed last cycle, it now has to re-lock
if lastcycle_jammed_bool

time_jammed_thisfight = time_jammed_thisfight + lock_time;

end

lastcycle_jammed_bool = 0;

end

end

% check if time_jammed has already occured in a previously simulated fight: this command retrieves all indices where time_jammed_thisfight is equal to the entry in the vector
% time_jammed
ind = find(time_jammed == time_jammed_thisfight);

% time_jammed_thisfight has not occured before, the set of indices is empty
if length(ind)==0

% find all indices for which the entry in time_jammed is lower than the jammed time from the simulated fight
ind_sort = find(time_jammed < time_jammed_thisfight);

% Technical thing here, max of an empty set is also an empty set in matlab but has to be zero here to work properly.
if length(ind_sort) == 0
max_ind_sort = 0;
else
max_ind_sort = max(ind_sort);
end

% add the new value into the vector time_jammed so that its entries are still increasing (vector has to remain sorted)
time_jammed = [time_jammed(ind_sort) time_jammed_thisfight time_jammed(max_ind_sort+1:end) ];

% the absolute frequency of the new value for time_jammed is one: it has occured the first time
abs_freq = [abs_freq(ind_sort) 1 abs_freq(max_ind_sort+1:end)];

% If the value of time_jammed_thisfight has occured before, add +1 to the corresponding absolute frequency
elseif length(ind)==1

abs_freq(ind) = abs_freq(ind)+1;

end

end

% Now we compute the relative frequency of the different realizations of time_jammed and use this as an approximation for the propability. For a large number
% of simulated fights, this approximation is very good.
prop_t = abs_freq./N_fights;

Multimorph
Gallente
Federal Defence Union
Posted - 2008.09.11 07:52:00 - [3]
 

Edited by: Multimorph on 11/09/2008 08:29:30
Edited by: Multimorph on 11/09/2008 07:58:31
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

Jam graph

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
Federal Defence Union
Posted - 2008.09.11 07:53:00 - [4]
 

Edited by: Multimorph on 11/09/2008 08:30:28
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.

Jam graph with ECCM

(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%.

Cpt Branko
Retired Pirate Club
Posted - 2008.09.11 07:53:00 - [5]
 

Edited by: Cpt Branko on 11/09/2008 08:00:40
Link the pictures properly.

Very interesting results otherwise.

Really rushed with the reply here Very Happy

Multimorph
Gallente
Federal Defence Union
Posted - 2008.09.11 07:53:00 - [6]
 

Edited by: Multimorph on 11/09/2008 08:31:29
Edited by: Multimorph on 11/09/2008 07:56:39
c) Back to a battleship with no eccm (sensor strength is 22 still). The Ewar ship now uses 5 jammers with strength 15

Jam graph, no ECCM v five jammers

Now the ewar ships puts his 5 jammers all onto the victim. This will almost certainly permajam the victim with a possibility of far over 90%. Accordingly, the
expected value is also almost 100%. At no occasion is the victim jammed for less than 85% of the fight.

Multimorph
Gallente
Federal Defence Union
Posted - 2008.09.11 07:54:00 - [7]
 

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

Jam graph, one ECCM T2 v five jammers

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%.

Ethan Hunte
Ninjas With Frikkin StarShips
Posted - 2008.09.11 08:03:00 - [8]
 

so what you're saying is, ecm is over powered...

SHOCKING.

Chienka
Di-Tron Heavy Industries
Atlas Alliance
Posted - 2008.09.11 08:03:00 - [9]
 

I think a histogram would be more appropriate for displaying data by frequency, since you're looking at ranges and probability.

Good work nonetheless.

Multimorph
Gallente
Federal Defence Union
Posted - 2008.09.11 08:06:00 - [10]
 

Edited by: Multimorph on 11/09/2008 08:33:21
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.

/E: After thinking a bit about your comment I changed all figures into bar diagrams. This might be more appropriate. Thanks for the advice.

Loree
Southern Cross Incorporated
Flying Dangerous
Posted - 2008.09.11 11:13:00 - [11]
 

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.

The Tzar
FinFleet
Raiden.
Posted - 2008.09.11 11:43:00 - [12]
 

Edited by: The Tzar on 11/09/2008 11:43:29
Excellent post, probably one of the best ECM posts. Nice one OP.

So, in reality a falcon pilot faced with a BS fleet will use only 1 jammer per ship, if it fails generally the next jammer will be used on a different ship as it's very rare to have a rack of only one race jammers.

What I read from the post is that when a falcon is using one jammer on one BS and the BS then fits ECCM, this is a very effective counter to ECM.

Please don't anyone run away with the part where the OP talks about 5 ECM put onto one BS. First of all this assumes 5 racials of the same flavour (does this ever happen, reallY?) and that the falcon's gang is attacking ONLY one BS (caught by yourself in a BS, with no support, you deserve to die).

This really is the only way to test a module that runs on probability.
Either this is get in a falcon yourself and see how often it wins / fails.

The accuracy of probability testing goes up exponentially as you tend towards infinity. Therefore all you whiners out there who have been 'permajammed' in a fleet context in a BS, can only make your claims after say a thousand instances or more...

If you are by yourself I'm afraid you have nothing to complain at. You should have either scouted, been aligned, travel fitted or had support with you. The fact you got caught by larger numbers is not the falcons fault.

Again great post Multimorph. I wish more would follow your example and use maths to reason their arguments as this is ENTIRELY what actually happens in the game.

The fact that people get so emotional about it is just another show that CCP have done a great job in making this game.

edit : spelling

Cpt Branko
Retired Pirate Club
Posted - 2008.09.11 11:44:00 - [13]
 

Edited by: Cpt Branko on 11/09/2008 12:17:48
Edited by: Cpt Branko on 11/09/2008 11:50:26
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.


Yes. However, we can clearly see that even Battleships with a preety solid sensor strenght get heavily neutralized by applying only one jammer, and in case the battleship in question is fitting a ECCM, it's still hit very hard by a single jammer.

This is particularly true in case of Minmatar/Amarr BS which get notably smaller sensor strenghts (for instance, 33.3 for a ECCM-ed Armageddon) and one jammer is therefore quite effective vs them even when they are fit with ECCM. A ECM ship should at least be forced to use two jammers, which it really isn't for a Minmatar/Amarr BS.

I'll plot some graphs for various battleships when I have a bit more time.

Originally by: The Tzar

So, in reality a falcon pilot faced with a BS fleet will use only 1 jammer per ship, if it fails generally the next jammer will be used on a different ship as it's very rare to have a rack of only one race jammers.

What I read from the post is that when a falcon is using one jammer on one BS and the BS then fits ECCM, this is a very effective counter to ECM.



Quite the opposite. The graphs show that fitting one ECCM does not really force a ECM ship to use two jammers to be really effective.

Being out nearly 50% of the fight in case of higher sensor str ships is a bit too effective for one jammer - and the effectiveness is higher if the target is a sub-BS ship or a Amarr/Minmatar ship. A Armageddon with 33.3 sensor str after one ECCM is still knocked out for a huge part of the fight with one jammer.

If using a ECCM forced a ECM pilot to use two jammers to be reasonably effective and would be only marginally effective with one (around 25-30%), it would be a much more valid counter.


Edit:
It appears that the best choice of action (for DPS reduction, against BS with 1 ECCM) is not bothering to use two jammers on each, but rather just one per ship.

If we take a 46% of being put out of the fight with 1 jammer on a single ECCM BS, then it's far better to just spread your jammers on a number of ships and supress 46% of their effectiveness (if permajamming is not essential but you're after DPS reduction instead) then use two jammers for ECCM-ed targets. I'm under impression this is the effect of relocking time, but I might be wrong.

Kagura Nikon
Minmatar
Emptiness.
Posted - 2008.09.11 12:08:00 - [14]
 

And is even worse if you use matari BS that have lower sensor strength. Try with 19 for example.

Multimorph
Gallente
Federal Defence Union
Posted - 2008.09.11 12:08:00 - [15]
 

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:

Jam graph, one ECCM v three jammers

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!

Nikunai
Posted - 2008.09.11 12:25:00 - [16]
 

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. Shocked

Hugh Ruka
Exploratio et Industria Morispatia
Posted - 2008.09.11 12:27:00 - [17]
 

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 ...

Multimorph
Gallente
Federal Defence Union
Posted - 2008.09.11 12:33:00 - [18]
 

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. Shocked


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.

Evanade
Rionnag Alba
Against ALL Authorities
Posted - 2008.09.11 12:42:00 - [19]
 

Great work, as i said before already. While the extremes are nice, (assuming max skills, 3+ racial jammers and a fitting wholly geared towards jam strength) if i were you i'd build a case where you assumed lvl 4 skills, max. 2 racial jammers and a more reasonable fit (like, 3x SDA but without rigs, or 2 SDA and rigs, prob. will yield about the same netto), because no-one will have max skills *AND* T2 rigs *AND* 2+ racials.


Multimorph
Gallente
Federal Defence Union
Posted - 2008.09.11 12:47:00 - [20]
 

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)

Jam graph, four cycle fight, no eccm one jammer

(b)

Jam graph four-cycle fight, one jammer, one eccm

Cpt Branko
Retired Pirate Club
Posted - 2008.09.11 12:48:00 - [21]
 

Edited by: Cpt Branko on 11/09/2008 12:49:47
Edited by: Cpt Branko on 11/09/2008 12:48:23
Originally by: Nikunai

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?



Doesn't that stay the same if they've also got a Huggin/Rapier? Doesn't it change if it's a 200 man fleet v a 200 man fleet and they just one-volley whatever support? Doesn't it change if you've got twice the amount of people and just undock 10 ECCM-ed snipers? Etc...

We're not discussing anti-Falcon tactics or situations you might find yourself in. Every ship is totally useless when you bring dedicated ships to shut it off, and everything is vulnerable and going to die easy vs a 200 man blob.

Also, for your information (even though you are really derailing), it doesn't always change. Let's say a Falcon menages to jam 3 BS initially for one cycle, and then a MWD-ing ship forces him off (while putting himself out of the fight for quite a bit too) and Falcon pilot decides not to come back.

Hey, 3 BS jammed initially in a, say, 5 BS + falcon on 5 BS + nano something fight means that assuming everything stays the same, 5 BS + Falcon still have the advantage. Just because of the initial three jams. It really depends on who is fighting what. Falcons definitely murder even slightly outnumbered gangs.

Originally by: Hugh Ruka

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.



It's a nice suggestion.

It's already done, interesting data.

Vim
Spiritus Draconis
Posted - 2008.09.11 12:54:00 - [22]
 

I've got one ship that got 5 racials... Its a falcon, SBII, ECCM, 5xjammers 2xSDA 1xBackupECCM 2xOptimal rigs

...the racials are all caldari, worked wonders O_O

Cpt Branko
Retired Pirate Club
Posted - 2008.09.11 13:03:00 - [23]
 

Edited by: Cpt Branko on 11/09/2008 13:03:09
In light of the data, I really like a suggestion someone posted about ECM mechanics: keep locks active, but prevent use of guns/RR/EW while jammed (thereby removing relock time).

Relocking time appears to increase the effects of jamming at lower chances (so something like every second/third jam happening) too much (while having less of a effect with multiple jammers).

Cautet
Celestial Apocalypse
Posted - 2008.09.11 13:31:00 - [24]
 

The assumptions of 10 seconds reapplying lock seem slightly harsh. Reapplying lock to a BS say would I think take less time that that.

The 15 ECM str one at the end is not much use. You will not face 15 ECM str.

Could be nice to see one that illustrates differing sensor strengths and for lock time to perhaps be tweaked.

I don't really know the math's behind this - but from a practicle point of view a chance of say 30% to jam for 20 seconds and 70% to miss for 20 seconds would mean that for every 20+x seconds jammed you will not be jammed for (20*2)-x [x = time to relock] so if x=10 (which is high in my view) for every 30 seconds jammed you are not jammed for 30 seconds [see how much a difference the relocking makes as much as the ECCM - that underlying assumption makes a huge difference]

if time to relock = 5 seconds then: jammed 25 seconds not jammed 35 seconds, so from 50% jammed down to 41% jammed.

Cpt Branko
Retired Pirate Club
Posted - 2008.09.11 13:40:00 - [25]
 

Originally by: Cautet

I don't really know the math's behind this - but from a practicle point of view a chance of say 30% to jam for 20 seconds and 70% to miss for 20 seconds would mean that for every 20+x seconds jammed you will not be jammed for (20*2)-x [x = time to relock] so if x=10 (which is high in my view) for every 30 seconds jammed you are not jammed for 30 seconds [see how much a difference the relocking makes as much as the ECCM - that underlying assumption makes a huge difference]

if time to relock = 5 seconds then: jammed 25 seconds not jammed 35 seconds, so from 50% jammed down to 41% jammed.



Yes, your calculations are correct. Relocking time boosts the effectiveness of jammers in a low success rate scenario more then in the high success rate scenario, which makes ECCM less effective as a counter-module.

10 seconds is not unrealistic (in actual TQ conditions, meaning a second of module activation lag and a part of a second at the end of the cycle before you can 'start' locking), but I'll have to check that. Generally, adding 1.5s to 'all L5' EFT lock time would yield a fairly realistic number for relocking time in combat.

Ampoliros
Aperture Harmonics
K162
Posted - 2008.09.11 14:04:00 - [26]
 

How much of the percentage out of the fight is relocking time, and how much is ECM'd time?

Multimorph
Gallente
Federal Defence Union
Posted - 2008.09.11 14:13:00 - [27]
 

Edited by: Multimorph on 11/09/2008 14:54:56
Edited by: Multimorph on 11/09/2008 14:53:42
Edited by: Multimorph on 11/09/2008 14:52:43
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?


/E: For better comparison, i modified to figures below. Now, the x-axis consists of 10 bins: 0%-10%, 10%-20%, .. 90%-100% and the y-axis
shows the sum of all possibilites of jam times belonging to the according bin


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

Jammer strength 10 v Sensor strength 30


b) Instant relock

Jammer strength 10 v Sensor strength 30



The effect is indeed quite noticable. The expected value drops by over 10 percentage points, the extreme near "permajammed" events are a lot less likely and the most likely events move from the 40-50% bin to the 30-40% bin.


Evanade
Rionnag Alba
Against ALL Authorities
Posted - 2008.09.11 14:22:00 - [28]
 

Originally by: Cpt Branko

In light of the data, I really like a suggestion someone posted about ECM mechanics: keep locks active, but prevent use of guns/RR/EW while jammed (thereby removing relock time).

Relocking time appears to increase the effects of jamming at lower chances (so something like every second/third jam happening) too much (while having less of a effect with multiple jammers).



Agreed, it would balance out ECM nicely if it would only inactivate all mods affecting others and not brak the lock. That, or ECCM could give a boost to scan resolution as well.

Hugh Ruka
Exploratio et Industria Morispatia
Posted - 2008.09.11 14:40:00 - [29]
 

Edited by: Hugh Ruka on 11/09/2008 14:45:08
Edited by: Hugh Ruka on 11/09/2008 14:43:40
Originally by: Cpt Branko
Edited by: Cpt Branko on 11/09/2008 13:03:09
In light of the data, I really like a suggestion someone posted about ECM mechanics: keep locks active, but prevent use of guns/RR/EW while jammed (thereby removing relock time).

Relocking time appears to increase the effects of jamming at lower chances (so something like every second/third jam happening) too much (while having less of a effect with multiple jammers).



This was proposed before ECM nerf in RMR (don't remember, it was long ago). People screamed that it won't have any effect on the resulting situation ...

A skilled ECM pilot is actuay living on the relock times in case of failed jams. You have to space out your jammers so that you are always getting a jammers to end of cycle shortly after each other ... 3-4 seconds is ideal when dealing with cruisers, about 6-7 for battleships (assuming you are flying a falcon). This way you have time to relocate jammers in case even multiple cycles fail.

Actualy the best scenario is a recalibration time dependant on the ratio of jammer/sensor strength.

If I black out a frig that has lower sensor strength than my jammer, he gets hit by a sensor cluster recalibration and then relock time. His relock time is short, so most of the penalty comes from the jamming. A BS on the other hand, most of the penalty comes from relocking as the sensors are much more resistant, so I actualy only may break his locks but he can start relocking instantly ... but this was also propsed long ago ... This even forces falcons to work side by side with gallente recons for better effect ...

Crazy Tasty
Beyond Divinity Inc
Beyond Virginity
Posted - 2008.09.11 14:40:00 - [30]
 

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.


Pages: [1] 2 3 4 5

This thread is older than 90 days and has been locked due to inactivity.


 


The new forums are live

Please adjust your bookmarks to https://forums.eveonline.com

These forums are archived and read-only