Apollo Redundancy

Somtaaw wrote:If they can't pull FTL data from their RDs, and can't FTL update Apollo, the GA does lose many of it's advantages. They'd be down to better compensators, and relying on Apollo's jacked-up borderline AI computers being in autonomous mode.

tlb wrote:The only place we have seen that happen was at Beowulf, after the destruction of Mycroft; where the autonomous missiles still destroyed over sixty percent of the attacking Solarian ships. I am not sure how that would compare to attacking a fleet with better defenses, but it still might be almost as good as a pre-Alliance Havenite fleet attacking ships of the RMN.

Somtaaw wrote:Yes they do have other advantages, but if Manticore were stripped of it's FTL advantage while facing a near tech-parity foe? Take pre-Alliance Haven, they put up quite a good fight for years, and if they could knock FTL out of the equation, their point defense of mass over accuracy could blunt even a lot of Apollo.

Because knocking out FTL not only impacts Apollo fire control, it also impacts Ghost Rider drones seeing EW changes and decoy launches in real-time. It also removes the advantage of having GR drones track missiles that go ballistic and sending that intel back in time to be useful, which impacts on Manticore's ability to perform long-range countermissile hits.

Overall Manticore becomes less accurate, falls for decoys and ECM more often, and they'll take more hits in return because they're half-blind compared to what they've gotten used to.

Taking all FTL communication out, the Apollo control missile and its brood are still better situated than the same missiles would be without the control missile; as you said the much improved computer capability is the reason for that increase. Here is the text from chapter 12 of Storm from the Shadows:

. There were some potential drawbacks to that, but it allowed us not only to retain the full range of the MDM, but actually required very few modifications to the existing Mark 23. And, somewhat to the surprise of several members of our team, using a dedicated control missile actually increased tactical flexibility enormously. It let us put in a significantly more capable—and longer-ranged—transciever, and we were also able to fit in a much more capable data processing and AI node. The Mark 23s are slaved to the control bird—the real 'Apollo' missile—using their standard light-speed systems, reconfigured for maximum bandwidth rather than maximum sensitivity, and the Apollo's internal AI manages its slaved attack birds while simultaneously collecting and analyzing the data from all of their on-board sensors. It transmits the consolidated output from all of its slaved missiles to the firing vessel, which gives the ship's tactical department a real-time, close-up and personal view of the tactical environment.
"It works the same way on the command side, as well. The firing vessel tells the Apollo what to do, based on the sensor data coming in from it, and the on-board AI decides how to tell its Mark 23s how to do it. That's the real reason our effective bandwidth's gone up so significantly; we're not trying to individually micromanage hundreds or even thousands of missiles. Instead, we're relying on a dispersed network of control nodes, each of which is far more capable of thinking for itself than any previous missile has been. In fact, if we lose the FTL link for any reason, the Apollo drops into autonomous mode, based on the prelaunch attack profiles loaded to it and the most recent commands it's received. It's actually capable of generating entirely new targeting and penetration commands on its own. They're not going to be as good as the ones a waller's tac department could generate for it if the link were still up, but we're estimating something like a forty-two percent increase in terminal performance at extreme range as compared to any previous missile or, for that matter, our own Mark 23s with purely sub-light telemetry links, even if the Apollo bird is operating entirely on its own."

ThinksMarkedly wrote:

Somtaaw wrote:Your assumption is flawed that a stealthed ship somewhere between the two ships is only using CMs for anti-missile. During the ballistic phase, attack missiles (if you can localize them) are HIGHLY vulnerable to beam weapons. Mk23s are only crowding the 0.3-0.6 range, but beams are twice as fast and there's no missile wedges to block the laser/graser.

If you had a ship that was entirely, or mostly beam-based, they'd be able to put a pretty large hurting on a ballistic Apollo salvo. And being at the mid-point, it would have much better eyes on where Apollo's stage 2 drive burnt out and the salvo went ballistic; than the ultimate target would.

That's an interesting concept: using anti-shipping beam weapons (which have a range of 2 to 3 light-seconds) instead of PDLCs. That gives you much better ranging against the missiles and allows the ship that is doing the attacking to be further out. If I were that ship, however good my stealth was, I wouldn't want to be within 100,000 km of the missile salvo -- there's a non-negligible chance that the networked sensors would see me!

However, anti-shipping beams don't have a high cycle rate. At 0.54c, the missiles would be out of range within 4 seconds, so it seems like the ship would be able to fire each emitter once only. Plus, anti-shipping isn't designed to hit such small targets, even though they aren't evading.

This attack is also geometrically difficult: the interception platform must have been placed directly on the flight path of the missiles before they were themselves launched. So it can't be done if the source of the missiles isn't known well in advance (e.g., they've transitioned from hyper in the last 10 minutes). What's more, this means this technique can only be used once: after the interception has happened, the next salvo(s) will curve away and add an orthogonal separation of a few million km. That decreases the powered range and increases the ballistic one, but it makes interception negligible again.

This would delay the missile strike though, which is almost as good as being able to kill some or most of a salvo. To have a functional curve around a heavily-stealthed ship, or ships, near the midpoint would require adjusting your MDM drive activations and actually pushing the minimum to 3, DDMs like the Mk16 would be unable to curve.

You'd have your initial gravity acceleration from the launcher itself, followed by a traditional 60s max thrust (or 180s lower thrust) burn curving the missile around where the stealth ship just wiped out your previous salvo. Your 2nd drive has to activate at what used to be the mid-point to redirect the missiles back toward the final target with another 60s or 180s burn followed by a second ballistic segment. Then finally somewhere around 2-4 million km from the target your third drive will finally kick in for terminal maneuvers.


System Defense missiles, having 4 drives could still pull a 2 drive burn at the start for a better initial kick to get around any stealth interceptors, use third drive to redirect after passing and then 4th drive for final maneuvers.



And overall using main energy weapons like grasers for missile killing is already semi-doctrine for many navies, so it's not exactly new. Haven does/did it a lot, Manticore used to before they developed off-bore launching and probably still use it to supplement their PDLC's if the salvo is large enough.

Jonathan_S wrote:Yeah, we don't get a hard comparison of that.
UH gives us two bits of information on autonomous Apollo, but both seem to compare it to FTL controlled Apollo -- not to pre-Apollo Mk23s.

The first tells up that the (then?) current Apollo FTL range was between 3 and 4 light minutes; and that by losing connection before reaching Chin the 23Es having to go autonomous were less effective than if they'd still been in FTL range.
The second quantifies that and says that autonomous accuracy is 30% of its FTL accuracy (though I'd assume it would be a sliding scale down to 30%, based on how long they'd had to be autonomous -- not an instant step change where cutting off FTL instantly degrades them 70% )

Still, while we're not given a hard number, it's quite clear than 30% as effective as FTL Apollo is still vastly better than non-Apollo MK23s would do at their extended ranges.

But it'd be nice to know where the crossover point would be between effectiveness of lightspeed controlled Mk23s and fully autonomous Apollo -- are autonomous 23Es as effective as controlled non-Apollo 23s at 30 million km, 50, 15?

With respect, Honor launching against Chin, she could still FTL update those Apollo birds until they finally exited her range for real-time updates, and that was with the pre-seeded Hermes buoys around Manticore system.


My notional scenario of wiping out all FTL would mean Apollo gets ZERO in-flight updates via FTL. At that point in time, you're really just launching standard Mk23's that are marginally better controlled because Apollo can think better than a Mk23 can.


And no FTL would also mean as I observed, for the Ghost Rider drones to be unable to send information quickly enough about EW changes; your target could suddenly change course and drop off 6 Lorelei-type decoys that take up station between your target and your missiles... no FTL update from Ghost Rider means you cannot know those decoys got dropped, what EW they're using, or in any way try to exclude them.


We'd be returning to, quite literally, Basilisk Station levels of combat, just at MDM ranges rather than SDM. Even at near SDM range, EW could throw off many missiles, so even Apollo is going to have a lot of work in trying to sort out what's a decoy, how to ignore jamming, and whats a genuine target. Particularly when the information it has to calculate from came from sublight RD's transmitting say 6 million km to the launch ship and then another 5 million km back to Apollo, with another 1million left to go? That's a lotta time for EW & decoys to play merry havoc on what could have been a perfect salvo.


Apollo is good, there's no denying that, but it's still built very heavily around benefitting from FTL Ghost Rider drones giving detailed accurate information that is usually less than 60 seconds old. Thus every Solarian pretty much crapping their drawers when they'd speak to a Manty, lean back expecting to wait 20 minutes for a reply and 20 seconds later "Oh fuck!"

cthia wrote:You are not considering that if the attack birds cannot detect a spider drive, they don't see a thing. Sending back a bigger collective picture of nothing, is still nothing. A collage of nothing is nothing.

I am.

Sensors don't report a step function of "this pixel here has nothing, the next one over has something." They're reporting a series of levels, if nothing else because of noise. Noise in the sensors, noise in the missile body, noise from background radiation, solar wind, other emissions, etc. The stealth hides the ship by causing their contribution to the signal to be low enough that's indistinguishable from noise: it doesn't cross the threshold where the AI or the human behind it says "a-ha! here's a ship!"

But random noise averages out. If you add the images of two sensors that are diverse, then a portion of the noise from one will cancel the noise of the other. Not everything, of course: if the two sensors are looking at the same source of noise like background radiation (like being bathed by the same solar wind), they'll be mostly identical and thus add up (constructive interference). But some of it cancels out, so with the appropriate software, adding two images reduces the noise level and increases the signal-to-noise ratio. Do this multiple times with multiple sources and your SNR goes considerably up. Add to this the ability to integrate over time, and the fact that they're closer to the target and getting closer every second (which also means that the background is changing). As a result, more eyes looking at the same particular spot in space for some time have a much better chance of seeing something.

How much that is, of course, we can't say. This is all dependent on the Plot Needs. All I am saying is that there's a technical & scientific foundation for missiles being able to detect what the ships can't. I can't make RFC use it.

Somtaaw wrote:
This would delay the missile strike though, which is almost as good as being able to kill some or most of a salvo. To have a functional curve around a heavily-stealthed ship, or ships, near the midpoint would require adjusting your MDM drive activations and actually pushing the minimum to 3, DDMs like the Mk16 would be unable to curve.

You'd have your initial gravity acceleration from the launcher itself, followed by a traditional 60s max thrust (or 180s lower thrust) burn curving the missile around where the stealth ship just wiped out your previous salvo. Your 2nd drive has to activate at what used to be the mid-point to redirect the missiles back toward the final target with another 60s or 180s burn followed by a second ballistic segment. Then finally somewhere around 2-4 million km from the target your third drive will finally kick in for terminal maneuvers.

Why on earth would you use the entire 1st drive to build your lateral vector and the entire 2nd drives cancel it out?!? (Rather than just turning about half-way through the 'burn' of the 2nd drive?)


Missiles can change heading while under acceleration. And over a 50 million km flight to divert by even as much as a million km by the midpoint they only need to alter heading by about 2.3 degrees at launch (and then turn back 4.6 degrees at the midpoint), extending a 50 million km flight path to 50.04 million km.

Sure, if you're making a close in attack you'd have to alter your heading more and generate a broader curve to get that same lateral displacement, but you'd also have more time on your drives to compensate. If you needed to dodge a million km off of a 20 million km path you need to change your heading by 5.71 degrees, and lengthen your 20 million km flight path to 20.1 million km.

Even a Mk-16 that had to accept a long ballistic segment should be able to recover. On that 50 million km flight, a 2.3 angle off to one side, even if you never turned back, only imparts a total of 2 million km lateral displacement at the target. The Mk-16 would have to accept a marginally longer ballistic segment, and then cut back a bit more sharply, but it shouldn't have any trouble dodging a million km out of the way, coasting along, and still making it to the target.


It simply isn't a big deal to adopt a slightly less than direct routing. And heck, it'd usually make some sense to have each salvo spread out on slightly different bearings anyway. Once they turn back it gives you more widely separated views of the target, it makes it even harder to thin out the salvo by using MDM proximity kills, and it would tend to thwart attempts to have anti-missile pickets lying doggo along the flight path.

ThinksMarkedly wrote:

cthia wrote:You are not considering that if the attack birds cannot detect a spider drive, they don't see a thing. Sending back a bigger collective picture of nothing, is still nothing. A collage of nothing is nothing.

I am.

Sensors don't report a step function of "this pixel here has nothing, the next one over has something." They're reporting a series of levels, if nothing else because of noise. Noise in the sensors, noise in the missile body, noise from background radiation, solar wind, other emissions, etc. The stealth hides the ship by causing their contribution to the signal to be low enough that's indistinguishable from noise: it doesn't cross the threshold where the AI or the human behind it says "a-ha! here's a ship!"

But random noise averages out. If you add the images of two sensors that are diverse, then a portion of the noise from one will cancel the noise of the other. Not everything, of course: if the two sensors are looking at the same source of noise like background radiation (like being bathed by the same solar wind), they'll be mostly identical and thus add up (constructive interference). But some of it cancels out, so with the appropriate software, adding two images reduces the noise level and increases the signal-to-noise ratio. Do this multiple times with multiple sources and your SNR goes considerably up. Add to this the ability to integrate over time, and the fact that they're closer to the target and getting closer every second (which also means that the background is changing). As a result, more eyes looking at the same particular spot in space for some time have a much better chance of seeing something.

How much that is, of course, we can't say. This is all dependent on the Plot Needs. All I am saying is that there's a technical & scientific foundation for missiles being able to detect what the ships can't. I can't make RFC use it.

Adding - we know that spider ships have to use their stealth skin to emit heat - they just use it to emit the heat AWAY from known sensor sources. Essentially, if you have sufficient sensors in a 360 degree circle looking towards the spider ship, one of them will see the heat radiation.

Now, how narrow a window the heat emittor is, and how long they may be able to close up before they have to emit (if they can) are unknowns. But essentially, enough sensors looking long enough, will detect a spider ship's heat.

And supposedly the Spider's drive CAN be detected - it's just not something anyone is looking for.

Somtaaw wrote:This would delay the missile strike though, which is almost as good as being able to kill some or most of a salvo. To have a functional curve around a heavily-stealthed ship, or ships, near the midpoint would require adjusting your MDM drive activations and actually pushing the minimum to 3, DDMs like the Mk16 would be unable to curve.

We're not talking about DDMs or plain MDMs. We're talking about Apollo MDMs or 4DMs with a ballistic phase. Missiles with active wedges aren't easy to target, because a minute change in angle could interpose the wedge. And because they're accelerating, they could also accelerate sideways by a little, thus creating an evasive pattern.

The point is that the ballistic phase should start a couple million km off the direct path. With 2 stages of acceleration, a Mk23 has a range from rest of 29.3 million km. In order to add a lateral separation of 2 million km, the missiles have to simply fly towards a point 4° off from straight. Of course, they'll have carried that lateral velocity all through the ballistic phase, adding about a light-second per minute, so they'll have to activate the 3rd stage later to compensate. But all of this within the capabilities of the missiles. In fact, it may be beneficial to not let the enemy know what exact angle the attack is going to come from until less than 3 minutes remain.

Or the missiles can accelerate 10° away in the first stage and compensate that in the second stage, so they have their ballistic phase parallel to the straight course.

You'd have your initial gravity acceleration from the launcher itself, followed by a traditional 60s max thrust (or 180s lower thrust) burn curving the missile around where the stealth ship just wiped out your previous salvo. Your 2nd drive has to activate at what used to be the mid-point to redirect the missiles back toward the final target with another 60s or 180s burn followed by a second ballistic segment. Then finally somewhere around 2-4 million km from the target your third drive will finally kick in for terminal maneuvers.

No one does two ballistic segments.

You're also placing the interception ship around the range of one drive, which is 7.3 million km or less of the launching ships. That's dangerously close because the launching ships can vector a dozen Ghost Riders on you very quickly, especially after they got a lock on you because you've fired at missiles. GRs can make a high-speed pass at you in 3.7 minutes or a zero-zero in 5.2. Or the launching ships can send a salvo of Mk23 in sprint mode, arriving in 127 seconds. In that time, the stealth ship has moved 20,000 km.

Somtaaw wrote:My notional scenario of wiping out all FTL would mean Apollo gets ZERO in-flight updates via FTL. At that point in time, you're really just launching standard Mk23's that are marginally better controlled because Apollo can think better than a Mk23 can.

And no FTL would also mean as I observed, for the Ghost Rider drones to be unable to send information quickly enough about EW changes; your target could suddenly change course and drop off 6 Lorelei-type decoys that take up station between your target and your missiles... no FTL update from Ghost Rider means you cannot know those decoys got dropped, what EW they're using, or in any way try to exclude them.

We'd be returning to, quite literally, Basilisk Station levels of combat, just at MDM ranges rather than SDM. Even at near SDM range, EW could throw off many missiles, so even Apollo is going to have a lot of work in trying to sort out what's a decoy, how to ignore jamming, and whats a genuine target. Particularly when the information it has to calculate from came from sublight RD's transmitting say 6 million km to the launch ship and then another 5 million km back to Apollo, with another 1million left to go? That's a lotta time for EW & decoys to play merry havoc on what could have been a perfect salvo.

Apollo is good, there's no denying that, but it's still built very heavily around benefitting from FTL Ghost Rider drones giving detailed accurate information that is usually less than 60 seconds old.

Where you said "marginally better controlled", the text said there was a 42% improvement without FTL. I get that the FTL communication has a major effect; but compared to the pre-Apollo missiles, so does the computing power in the command missile. Unless you want to insist that the text is wrong:

It's actually capable of generating entirely new targeting and penetration commands on its own. They're not going to be as good as the ones a waller's tac department could generate for it if the link were still up, but we're estimating something like a forty-two percent increase in terminal performance at extreme range as compared to any previous missile or, for that matter, our own Mark 23s with purely sub-light telemetry links, even if the Apollo bird is operating entirely on its own.

tlb wrote:[
Where you said "marginally better controlled", the text said there was a 42% improvement without FTL. I get that the FTL communication has a major effect; but compared to the pre-Apollo missiles, so does the computing power in the command missile. Unless you want to insist that the text is wrong:

It's actually capable of generating entirely new targeting and penetration commands on its own. They're not going to be as good as the ones a waller's tac department could generate for it if the link were still up, but we're estimating something like a forty-two percent increase in terminal performance at extreme range as compared to any previous missile or, for that matter, our own Mark 23s with purely sub-light telemetry links, even if the Apollo bird is operating entirely on its own.

And you've still missed the point. Knock out FTL entirely.... let's take Henke's simulation from Storm From the Shadows (ch 13). Your target is ~82 million km away, that's roughly about 4.5 light-minutes away. With Ghost Rider you have intel only ~4 seconds old, without grav-pulse that intel at time of launch is a minimum of five minutes old.

The missile salvo was sixty-eight million kilometers from Artemis, speeding steadily onward at 150,029 KPS. Its birds had been ballistic for four and a half minutes, ever since the second drive system had burned out, and they were still ninety-three seconds—almost fourteen million kilometers—from their target, even at half the speed of light.

Your intelligence at time of launch is what the Apollo missiles have, by the time any change in target information is transmitted from the RDs to the launching ship, and from launch ship to Apollo is going to be too long. 5 minute sublight is enough time for the info from the RD's to reach one way, but not go from RD to ship back to missiles, that would require somewhere between 6 and 9 minutes total loop, depending on how far the missiles were and when the RD transmitted relative to the launch.


So Apollo has been effectively in autonomous from the second it launched, certainly by the time 2nd drive cut out and it went ballistic because the target is simply too far for sublight commands to be effective. If the target launched decoys during the ballistic phase, Apollo cannot possibly know how many, it doesn't know those decoy signatures or locations, and it can't even see them at all due to the shrouds covering all sensors for the ballistic portion. In Henke's simulation they jettisoned shrouds around 10 million km from target (72 million km from launch ship), thanks to FTL that's a 4s one-way loop, without FTL thats 4 minutes.


THAT is the point I'm making, that 42% increase was assuming they had Ghost Rider giving the launch ship accurate information in real-time, and that the launch ship could update Apollo, even with sublight links until it got so far downrange you couldn't update it at all. Remove Ghost Rider from that equation and it is physically impossible for Apollo to give a 42% increase, it'd be more like 10%, maybe 15% at best. Apollo computers would have to be more powerful than the computers used in planetary HQs for it to be able to take information 4+ minutes out of date and somehow achieve a 42% accuracy increase. And if they had computers that good... why aren't they using them on the ships, rather than one-and-done missiles?

Somtaaw wrote:THAT is the point I'm making, that 42% increase was assuming they had Ghost Rider giving the launch ship accurate information in real-time, and that the launch ship could update Apollo, even with sublight links until it got so far downrange you couldn't update it at all. Remove Ghost Rider from that equation and it is physically impossible for Apollo to give a 42% increase, it'd be more like 10%, maybe 15% at best. Apollo computers would have to be more powerful than the computers used in planetary HQs for it to be able to take information 4+ minutes out of date and somehow achieve a 42% accuracy increase. And if they had computers that good... why aren't they using them on the ships, rather than one-and-done missiles?

That is not what the text says, it clearly states "even if the Apollo bird is operating entirely on its own"; which means without FTL or Ghostrider or Hermes buoys.

As the missiles approach the target, the information being received from the brood's sensors is considerably more recent than 4 minutes old. The text specifically says the computers are not as powerful as that on a ship, but the fact that the computer is there means that the missiles are more effective than anything that is undirected.