How would you make "Sensors" for a starship?

[Wyrm]

Dr. John Schilling discusses why the exhaust plume of a decoy will have to have the same thrust as a real ship:

Problem is, the rate (i.e. velocity) at which the plasma is coming out, manifests itself as a doppler shift in the characteristic emission lines of the plasma. As soon as a dedicated tracking sensor focuses on the target for a second or two, the game is up. If the plasma is coming out fast, it can't help but produce thrust proportional to mass flow rate (manifested as luminosity) times velocity (doppler). If the plasma is coming out slow (or fast but in opposing directions), it will be seen to be coming out slow and thus be recognized as not a real engine.

This assumes that you can gather enough light in a few seconds to get a spectrograph with enough accuracy to tell if there is a dopplar shift. But if detecting the ship against a noisy background is hard, getting a spectrograph is harder. In detection, you're focusing all the light into a small number of pixels to give it the best chance to trigger some threshold. But in a spectrograph, you have to spread that line out so you can see the absorption lines — each emitted frequency has to build up enough photons to be distinct from the absent frequencies. For stars and things of similar apparent brightness, this takes hours.

Only then can you get the dopplar shift and figure out that you're looking at a decoy, but by then the enemy has gotten a few hours' advantage, and you still have yet to find the little bugger. If burns last only minutes, you are shit out of luck — you don't even know whether that thing you just saw as a decoy or the real mccoy.

The obvious solution would seem to be to have the decoy shine redder light from the front, transitioning to bluer light in the back. Problem is,

...that spectroscopy doesn't work that way. The atoms themselves are moving at some relative velocity, so their absorption lines are shifted. It's not a matter of redder or bluer, but the spectrum of a luminous object has absorption line patterns that resemble sodium or nitrogen or what-have-you, but shifted down a ways.

[Darth Wong]

Just for fun, let's examine the idea of sprinkling sensor platforms around an entire solar system, as many people are assuming the defender has already done. Remember that, at 50 million km, you aren't likely to see much more than a dot from a vessel of reasonable size, unless you have a huge array. The mean orbital radius of Saturn is nearly 1.5 billion km; if we treat the solar system as essentially two-dimensional and ignore the possibility of the enemy launching out-of-plane attacks (for some reason), then the two-dimensional plane area of the disc from Earth orbit out to 1.5 billion km solar orbit is roughly 7E18 km². Now let's say that we figure we have a tactically useful identification limit of 10 million km for enemy starships: if you were to divide the 2D space between 1AU and 10AU orbital radius up into 20 million km wide squares, you'd need approximately 17500 squares. Of course, that wouldn't work because the detection area of each scope is circular, not square, so you need somewhat overlapping circles. So let's multiply that figure by √2, which means you need around twenty five thousand sensor platforms scattered throughout the system. And you still haven't got coverage past Saturn, or outside the 2D plane.

Could the enemy have the ability to deploy hundreds of thousands of sensor platforms throughout his star system? Sure, I suppose. It's all a matter of writer's fiat, after all. But it does highlight the difficulty of setting up this hypothetical system. Frankly, if you wanted to attack fixed targets like orbiting platforms or planetary installations, I don't see why all this talk of detecting exhaust plumes even matters. You'd probably accelerate the thing from really far away, and just let it coast until it gets close enough to attack. Especially if you're using drones rather than manned warships, which (since we're all apparently assuming they're stuck with hard sci-fi propulsion systems) they'd probably be using for space warfare anyway.

Obviously, if you assume that you can quickly identify ships from 1 billion km away with a scope, this becomes much easier.

[Drooling Iguana]

If we assume that the enemy can come out of whatever FTL mechanism they use (if they have them at all in this sci-fi universe, though it'd be difficult to conduct interstellar war without them) at any arbitrary point in the system then yes, we'd need a huge number of sensor platforms scattered around, but if we assume that they have to enter normal space somewhere far from the central star (either due to limitations with the drive technology or to avoid detection) then wouldn't we really only need to concentrate the sensor platforms around the planets they're likely to use for gravity-boosts? Especially if they can't fire their engines close to your homeworld it would be pretty difficult to work up any significant amount of speed using engine thrust alone.

[Darth Wong]

If we assume that the enemy can come out of whatever FTL mechanism they use (if they have them at all in this sci-fi universe, though it'd be difficult to conduct interstellar war without them) at any arbitrary point in the system then yes, we'd need a huge number of sensor platforms scattered around, but if we assume that they have to enter normal space somewhere far from the central star (either due to limitations with the drive technology or to avoid detection) then wouldn't we really only need to concentrate the sensor platforms around the planets they're likely to use for gravity-boosts? Especially if they can't fire their engines close to your homeworld it would be pretty difficult to work up any significant amount of speed using engine thrust alone.

Who says they need to use gravity boosts?

[Junghalli]

Yeah, you guess. It can be a guessing game. In the center of the solar system, you'd have to show decoy light pretty much everywhere because its relatively easy to get 360 degrees around, say, Mars from Earth. If you're attacking from Neptune? Not so much. If you're attacking from Neptune, depending on the level of development of this solar system, the probability of the enemy having scanners behind you is probably low.

I wouldn't be so sure. If you can get warships out to Neptune you can get observation platforms out there more easily, and the enemy presumably is not stupid; he'd know when he had exploitable blind spots like that and he'd try to fix them wherever possible. Plus transNeptunian orbits would actually be a really good place to put the observation platforms, first because if you want to hide them: the farther away they are from everything else the less likely the emissions of their low power drive systems will be picked up as they make their plane or orbit changes, and second because that's a position that gives you a really great commanding view of the entire solar system.

Same if this is an extrasolar war, and you're attacking from 90 degrees out of the plane of the ecliptic.

If this is interstellar war and you plan on doing any actual space fighting (as opposed to just throwing interstellar missiles at the enemy at some high fraction of c, which makes a lot more sense) I think you'd probably best give up on the idea of stealth. Unless you're planning on spending tens of thousands of years on the journey your ships will have to do some significant fraction of c, and that means a very high energy drive system that would be pretty much impossible to hide or make any sort of remotely practical decoy for. Oh, and while you're decellerating to enter the enemy system your rocket will have to be pointed straight at him, so the cool umbrella idea isn't going to work either.

Yes yes, but they would only be significantly doppler-shifted along the axis of acceleration. You would shine normal exhaust-plume light to your sides, not just blue to the front and red to the back.

Except (IIRC - this isn't my field) it would be more complicated than normal light from the sides, shifted light from the back and front. The amount of blue/redshifting you'd see (again, IIRC) would depend on the exact angle the sensor was looking at the exhaust plume from. So you'd need a continuously shifting spectrum of emission, not red front - normal side - blue back.

This assumes that you can gather enough light in a few seconds to get a spectrograph with enough accuracy to tell if there is a dopplar shift. But if detecting the ship against a noisy background is hard, getting a spectrograph is harder. In detection, you're focusing all the light into a small number of pixels to give it the best chance to trigger some threshold. But in a spectrograph, you have to spread that line out so you can see the absorption lines — each emitted frequency has to build up enough photons to be distinct from the absent frequencies. For stars and things of similar apparent brightness, this takes hours.

Ah, thank you Wyrm. What advances/techniques would could be used to reduce the image processing time? Would it be a "simple" matter of more computing power to process the image, or is it more complex than that?

Only then can you get the dopplar shift and figure out that you're looking at a decoy, but by then the enemy has gotten a few hours' advantage, and you still have yet to find the little bugger. If burns last only minutes, you are shit out of luck — you don't even know whether that thing you just saw as a decoy or the real mccoy.

Ah, well, in this respect it is helpful that the sort of engines for which these decoys would be practical would be low thrust. The burns most likely would have to take hours or longer. And a few hours isn't much time at all if you're crossing AUs (unless you're moving really, really fast).

...that spectroscopy doesn't work that way. The atoms themselves are moving at some relative velocity, so their absorption lines are shifted. It's not a matter of redder or bluer, but the spectrum of a luminous object has absorption line patterns that resemble sodium or nitrogen or what-have-you, but shifted down a ways.

Ah, again, thank you.

As somebody who obviously knows more about this than me, what sort of methods could be used to get around this problem if you want to use decoys (assuming there are any)?

[Drooling Iguana]

Who says they need to use gravity boosts?

Well it all depends on the in-universe level of propulsion technology. Regardless, with any kind of reaction-based drive they'd be putting themselves at a disadvantage by not using gravity boosts, though they may do so anyway to avoid detection.

[Ariphaos]

Just for fun, let's examine the idea of sprinkling sensor platforms around an entire solar system, as many people are assuming the defender has already done. Remember that, at 50 million km, you aren't likely to see much more than a dot from a vessel of reasonable size, unless you have a huge array. The mean orbital radius of Saturn is nearly 1.5 billion km; if we treat the solar system as essentially two-dimensional and ignore the possibility of the enemy launching out-of-plane attacks (for some reason), then the two-dimensional plane area of the disc from Earth orbit out to 1.5 billion km solar orbit is roughly 7E18 km². Now let's say that we figure we have a tactically useful identification limit of 10 million km for enemy starships: if you were to divide the 2D space between 1AU and 10AU orbital radius up into 20 million km wide squares, you'd need approximately 17500 squares. Of course, that wouldn't work because the detection area of each scope is circular, not square, so you need somewhat overlapping circles. So let's multiply that figure by √2, which means you need around twenty five thousand sensor platforms scattered throughout the system. And you still haven't got coverage past Saturn, or outside the 2D plane.

I'll do this two billion better.

Sensor array, ~one light year radius, one light-day thick for the main portion. We want one light-second accuracy or better, so 4 * pi * 32,000,000^2 * 86,400 units = ~35 trillion stations. Assuming each station is made from ten cubic meters of material, that's 350 trillion cubic meters or a planetoid with a radius of ~44 kilometers.

Make the sensor net a light year thick and centered around 1.5 light-years, if you like. You are not scratching the surface of the Solar System's resources. Multiply resolution by sixty and you would need to use a Venus worth of matter or so.

Edit: Sixty is me thinking stupid. Try ten >_>

This would be pointless in any sort of FTL scenario, of course, and most STL scenarios. I imagine the most efficient means would be to have a thinner network of large sensory planetoids that would fire interceptors at anything that interested them.

My own personal solution to the starship sensor problem is: FTL sensing requires psychics. At some point I needed to come up with a reason to have human(oid)s, life support, et all on board, magic is a good a reason as any >_>

[Wyrm]

Ah, thank you Wyrm. What advances/techniques would could be used to reduce the image processing time? Would it be a "simple" matter of more computing power to process the image, or is it more complex than that?

No. This is a physical limit. At this point, you're waiting for the photons themselves to arrive. Impatience can only gain you false readings up the wazoo. You can't squeeze blood from a stone.

Ah, well, in this respect it is helpful that the sort of engines for which these decoys would be practical would be low thrust. The burns most likely would have to take hours or longer. And a few hours isn't much time at all if you're crossing AUs (unless you're moving really, really fast).

Of course, for an attacker whose mission depends on stealth is going to enter slowly and going to burn carefully and for short periods. It's not going to be surprise buttsecks attack.

As somebody who obviously knows more about this than me, what sort of methods could be used to get around this problem if you want to use decoys (assuming there are any)?

The point of decoys is to keep your opponent's eyes misdirected at them instead of at the really juicy stuff. In that case, you want it to act kind of like a ship that looks like it's trying to hide.

What I would do is have small thrusters and a large lamp. The lamp would make the plume appear bigger, and the small thrusters would do the actual pushing — their weak signal would be swamped by the lamp.

Except for the absorption lines, which is why I specified many thrusters. They surround the lamp and eject gas at the right velocity all around the lamp so that their absorption lines are doplar shifted by a good degree. The lamp spoofs a large mass flow, while the thrusters provide the right acceleration and the right exhaust velocity. Result: interesting object!

[Junghalli]

No. This is a physical limit. At this point, you're waiting for the photons themselves to arrive. Impatience can only gain you false readings up the wazoo. You can't squeeze blood from a stone.

Ah, thank you for the information.

Of course, for an attacker whose mission depends on stealth is going to enter slowly and going to burn carefully and for short periods.

Of course, ultimately they're still going to need to burn for just as long a time, and tracking the object's trajectory between burns should be a matter of simple math, so you should still be able to keep an eye on the predicted paths and do a cumulative profile, right?

What I would do is have small thrusters and a large lamp. The lamp would make the plume appear bigger, and the small thrusters would do the actual pushing — their weak signal would be swamped by the lamp.

Except for the absorption lines, which is why I specified many thrusters. They surround the lamp and eject gas at the right velocity all around the lamp so that their absorption lines are doplar shifted by a good degree. The lamp spoofs a large mass flow, while the thrusters provide the right acceleration and the right exhaust velocity. Result: interesting object!

How convincing would this illusion be? I take it it's something that could still be distinguished from the real thing, but it'd take a lot of observation so it would force the enemy to waste a lot of time on it. Or am I jumping to conclusions there?

[Kuroneko]

I'm not sure why we're doing coverings. And efficient circle covering will involve a factor 1.21 rather than 1.41, but it's not necessary. All we need is to require than externally-originating trajectory going below some radius is covered by a sensor--or more than one, for redundancy. Thus we could have rings (or spheres, in 3D) of overlapping sensors, but with no requirement that these rings themselves be overlapping.

--

How sensitive can we expect image sensors to be in the future? Can we expect them to be able to register one photon at a time, a la photomultipliers? (And what are the limits of current technology in this regard?)

Since equilibrium temperatures are already very low in space at large distances from the star, with a contraption like an 'umbrella' with heat pumped from one side to another that obscures the ship's heat signature, it's quite possible to basically require that the sensor counts individual photons in order to get anything at all--forget resolving any details; even a single tiny dot would be hard to tell part from background.

For example, at 10AU from the nearest detector with a collector area of 100m² (>20 Hubbles), a 40K umbrella of front cross-section of 1.0E4 m² would give less than 3.5 photons per second of exposure time (not counting possibility of being scattered/absorbed before detection/noninteracting). Against some noise, reconstructing even a dot would be extremely impressive.

[Darth Wong]

That's a good point; I didn't bother with an optimized layout, and simply squashed the circles together in the same basic square layout, hence I got 1.41 instead of 1.2.

As for the photon count, that's a lower limit that's pretty hard to argue around. And it's worth noting that any scheme requiring analysis of absorption lines implicitly requires a lot of light: you need enough photons impacting on your detector to be able to identify clear spikes and valleys in the data.

[starslayer]

The big problem for most astronomical CCDs right now is noise. They will tell you if you've received a single photon, but good luck distinguishing it from the inherent CCD noise without a very long exposure time. For example, the faintest galaxy Hubble has ever observed (and confirmed), in the Hubble Ultra Deep Field, it got a grand total of three photons from. That same field literally took months to properly expose.

However, even with single photon accuracy, you really aren't going to be able to tell that a particular photon came from something you're interested in. There are simply too many possible sources, even at high resolution, to get a good detection like that. It could be a background star, nebula, cluster, galaxy, asteroid, one of your own ships, one of theirs, or glare from something else, a diffraction artifact from the optical train, etc., etc; this is why the Hubble identification took months. Those three photons came in over a period of months, and they weren't removed by any of the noise reduction and sampling algorithms that were used. What was left was a very dim, red smudge - a galaxy.

The upshot of this is that, currently, I believe, a hundred photons in a short exposure would be a pretty good indicator, especially if the object didn't move very much (a hundred photons on the same pixel would be very compelling). Potentially, this could go all the way down to something like ten photons, and maybe five, but that would be heavily pushing it.

[Wyrm]

Of course, ultimately they're still going to need to burn for just as long a time, and tracking the object's trajectory between burns should be a matter of simple math, so you should still be able to keep an eye on the predicted paths and do a cumulative profile, right?

And how do you predict that path? Psychic powers? You don't have the path. You may have the starting point, but you don't know anything about their velocity. All you have is the velocity of the plume. Actually, you only have one component of the same: the velocity along the line of sight. You don't know the velocity crosswise. You'd have to track the object for some time, enough time for it to displace significantly in your telescope. It also has to be identified as the same object in many telescope so you can triangulate.

But in order to do that, you have to tell the other posts to watch the same object. How do you know the little blip you see in one scope is the same as the other? If there is even a modest number of other objects that could be misidentified as the same object, the number of wrong combinations explodes.

Of course, this only works while the thruster is thrusting. While the engine is off, you're essentially invisible. How do you connect strange object A with object B?

So you know neither the position nor the velocity with any significant accuracy. It takes some time to work these things out in astronomy, and stars have an advantage because you can build up an image over time, since the proper motion of stars is so small in comparison to their brightness. You can't build up an image unless you know the position and velocity, but you can't establish either until you build up an image. Catch 22.

(Also, watch how you're attributing quotes.)

[Ariphaos]

I'm not sure why we're doing coverings. And efficient circle covering will involve a factor 1.21 rather than 1.41, but it's not necessary. All we need is to require than externally-originating trajectory going below some radius is covered by a sensor--or more than one, for redundancy. Thus we could have rings (or spheres, in 3D) of overlapping sensors, but with no requirement that these rings themselves be overlapping.

--

How sensitive can we expect image sensors to be in the future? Can we expect them to be able to register one photon at a time, a la photomultipliers? (And what are the limits of current technology in this regard?)

Since equilibrium temperatures are already very low in space at large distances from the star, with a contraption like an 'umbrella' with heat pumped from one side to another that obscures the ship's heat signature, it's quite possible to basically require that the sensor counts individual photons in order to get anything at all--forget resolving any details; even a single tiny dot would be hard to tell part from background.

For example, at 10AU from the nearest detector with a collector area of 100m² (>20 Hubbles), a 40K umbrella of front cross-section of 1.0E4 m² would give less than 3.5 photons per second of exposure time (not counting possibility of being scattered/absorbed before detection/noninteracting). Against some noise, reconstructing even a dot would be extremely impressive.

If we are actually talking about slower-than-light interstellar conflict, there is absolutely nothing wrong with detecting small projectiles after they have already passed your sensor network (at two light-years out or however far), seeing them from behind the umbrella, which is going to be glowing a lot more - above and beyond how trivially little material it takes to create a 2ly radius sensor net.

You could probably get single-shot rails, rocket and coil combinations up to a small percentage of c, but eventually the problem becomes intractable - either you need to use light-years of coil, or you don't have a projectile. A target will have years or even decades to react.

That leaves attacking with light sails. Even assuming you are building an umbrella contraption with a thousand kilometer-radius sheet of tissue paper, it's still a thousand kilometers wide - big enough that a feasibly dense sensor network may actually have a satellite or two get atomized in its passage, which won't speak well for any attempts at stealth.

I've said it many times before, but really, the only feasible way to do damage to another star system without invoking magitech is just to point sunshine and happiness at it. Everything else is either slow or bulky.

[Darth Wong]

One simple countermeasure against detection, given the limitations of light-gathering, is to fire your engine only in short bursts. They need time to pick up anything useful, so don't give them that time. At random intervals, fire an engine for less than a second. Depending on what kind of engine technology this hypothetical sci-fi civilization has, they might still be able to achieve decent acceleration this way.

[Junghalli]

On the question of detecting decoys, what about radar? The decoy is a lamp, not a ship, so it should look rather different if you got a good look at it. That suggest one possibility is to ping all your "possibles" with a very high energy radar burst to pick out which ones are decoys and which ones are ships. I suppose you could design the decoys to have the same shape as your ships. They'd probably still have to be made of different materials though (high energy emitting light surface vs spacecraft hull), and that might show up. Plus, would the drive plume show up on radar at all? Because it if would, the weaker drive plumes of the decoys could probably be used to pick them out.

You'd need one hellaciously powerful radar though, to get any kind of detail at the distances we're talking about. We're probably talking something to put the Arecibo telescope to shame. Fortunately, unlike with your passive sensors you'd probably only need a few of them, and could keep them in well defended areas.

And how do you predict that path? Psychic powers? You don't have the path. You may have the starting point, but you don't know anything about their velocity. All you have is the velocity of the plume. Actually, you only have one component of the same: the velocity along the line of sight. You don't know the velocity crosswise. You'd have to track the object for some time, enough time for it to displace significantly in your telescope. It also has to be identified as the same object in many telescope so you can triangulate.

You may not be able to tell the path initially. But a light that just appears out of nowhere is exactly the sort of anomaly that a competently programmed sensor grid should pay attention to (as it's exactly what you'd expect of a rocket burn in deep space). If you have more than one sensor (and you should) you should be able to get the distance via parallax. Your sensor grid would note the existence, sky position, and distance of this weird light that winked on and off, keep a record of it, and be on the lookout for any more.

Let's assume your ship is accelerating by 20 km/s and has an acceleration of 1 m/s^2. It'll need 20,000 seconds or 5.5 hours to get up to change its velocity by 20 km/s. That's lots and lots of short bursts. The next short burst the sensor net should, again, note the existence, sky position, and distance. Now assuming each burst is 5 seconds you need 4000 bursts. That's plenty of time to work out that these bursts are happening in a straight line of decreasing distance, and you should be able to use the parallax measurements to calculate velocity and acceleration.

You can suggest a higher acceleration drive, of course, but there's going to be limits without magitech. Even at a very generous 10 G your ship is going to need more than three minutes, or 40 five second burns. There's also the factor that a high thrust drive is much more likely to draw your enemy's attention than a low thrust one. A big warship that can do 10 G will be putting out one bright exhaust plume, which will at the very least definitely clue your enemy in that you're up to something. A high thrust drive also means decoys are out; good luck designing a lamp that convincingly imitates the chain of exploding nukes behind an Orion and doesn't melt itself, and doesn't cost more than an actual ship.

Also something else to note: warship staging areas are probably going to be relatively busy places, with all the construction you need to set them up and everything, and it will probably be rather difficult to keep the enemy from at least knowing you have some sort of presence there. So it's rather likely that you'd have a pretty good idea where any enemy military expedition is likely to set out from, and you can have your sensors watch those areas and the known trajectories of the ships that set out from them, instead of watching the entire sky for anomalies. This should make the sensor grid's job quite a bit easier.

Note: I will admit I have no idea how long it takes to acquire a decent parallax reading of a distant astronomical object, so I may be underestimating the difficulties there.

[erik_t]

A few thoughts after discussion with Phong:

• The enemy can be expected to use nuclear missiles proximity-detonated to fry your CCDs. You will certainly have shutters on these, so that they can blink when a nuclear device initiates. However, you're probably going to need to blink these ahead of an initiation, and a blinked CCD is blind for the duration of the blink. This means that, if you have a possibly-nuclear missile incoming, you will need to blink any CCDs not needed for antimissile fire. This is a strong argument in favor of a variety of sensors of multiple ranges; probably at least long-range and short range detectors and trackers.
• You just can't get away with active detection (eg radar) at astronomical ranges. That R4 is a killer for anything but an impossibly narrow pencil beam. We're almost certainly reduced to passive detection at any range O(104) km. Note that with modern technology, SPS-48 uses about 2MW to scan a 45deg slice of sky out to 500km, with a data rate of about 0.125hz. Minimum RCS at 500km isn't obvious, but I'd bet O(1m2). This implies that you could scan the whole sky at a data rate of about two scans per minute. This seems pretty good until you realize that a 16 km/s object (500km per 30sec) isn't implausible in this environment.
  Of course, SPS-48 is really not that "modern". We'd be using some form of phased arrays, and could probably get away with (say) a 45deg-half-angle cone as a threat axis. Lower PRFs might allow us to get substantially larger ranges, and allow us to view the whole sky. Clever Doppler business will also allow us to discern target motion. Radar seems to be the likely short-range (1000km-ish) detection and tracking system in a high-threat environment, when the missiles have already started flying and you know your location is known.
• Note that dumping 2.2MW is a serious issue for a spacecraft.
• It's going to be challenging to operate with radiators in excess of about 1800K. We're probably going to be using some mechanical heat pump (peltiers will fry way too early), and with modern materials, turbine engines are having great difficulty exceeding 1700K. Meanwhile, we're going to be lucky to get an emissivity in excess of 0.85. Since thermal radiation goes P = AσeT4, taking an emissivity of 0.85 and 1800K surface temperature, you're going to be able to dump about 506kW/m^2 from your radiators. This is unlikely to be achievable in practice, but it's something to work off of.
  Actually moving this much heat to a radiator is left as an exercise to the reader. Consider, though, that if you were using water and 1200psi 1000K superheated steam, you'd need to move 0.2kg/s of this steam. Of course, you will need secondary loops to move heat from 1000K to 1700K; Carnot would suggest that this will operate at no more than 44% efficiency, and things continue to get worse...
  Note that due to the quartic scaling of radiative heat transfer with radiator temperature, operating at that 1000K will mean you can only dump about 5kW/m2! This is unlikely to be acceptable; to run your SPS-48, you need a radiator larger than 440m2, almost two tennis courts!
• Assuming we're using lasers as our missile-defense systems, we only need the current bearing of the offending missile, not the speed and range. Doppler radar, for instance, is not strictly necessary. Lasers seem unlikely to be used beyond light-ms ranges due to the difficulties in pointing, not to mention beam dispersion problems.
• At astronomical ranges, of course passive data will be more up-to-date than active data, by a factor of two. However I think this is unlikely to matter much; at long range, second-to-second information is not important, since nobody can do anything at short timescales anyway.
• Getting back to the blinking issue, long-range systems will necessarily be slower to blink and unblink than short-range ones. To blink, you're either going to turn the instrument away from the incoming missile, or you're going to have some internal shutter. Either way, a long-range instrument will be slower to blink because it is more massive (for turning) or more subject to internal vibration (a quickly-moving internal shutter). This implies that you can keep the enemy from watching your actions by keeping him blinking at incoming missiles. Whoever fires first may be the only one who ever gets to; this is a strong argument for longer-range tracking and target acquisition.
• However, radar systems (and radio-wavelength systems in general) may be less subject to danger from nearby nuclear blasts. Higher-energy waves are shorter wavelength, and a L-band radar (with a mattress-looking antenna with 20cm between elements) will be relatively unaffected by hard radiation. However it would likewise be unable to detect objects smaller than about 20cm; also, longer-wave systems have correspondingly larger and more massive antennas. Still, the implication is that there may be a place for long-wavelength systems to maintain some situational awareness even when nukes are flying and other systems are blinked.
• As mentioned elsewhere, long-range detection essentially has to be passive, presumably a combination of visual and infrared telescopes and ESM. Long-range target acquisition is likely to be at least partially passive, in the event that you yourself have not yet been detected. However very narrow pencil beams will allow specialized radars to be effective even at very long ranges. They would have the disadvantage of broadcasting your position and giving you older data, although the data rate would probably be higher than a passive system.
• It's unclear to me how effective umbrellas would be in shielding you from IR detection. Given thermal equilibrium of the umbrella, their effectiveness would seem to scale with the square of the distance from ship to umbrella, normalized by the cross-section of the ship, with some additional emissivity factor. However, maintaining an umbrella very far from a ship would be challenging for a few reasons. First, you can't maneuver without picking it up and moving it (having it rigidly attached is probably a non-starter). Second, you can't see through it; you'll need to have all sensor systems be able to look above and around it.
• Umbrella use becomes more challenging as the umbrella gets large compared to the cross section of the ship. And it must be large compared to the cross-section of the ship if it is to be of any use... a small umbrella presupposes you already know who is looking for you and where they are. I think all of this implies a variable-size umbrella, with only long-range detection systems able to see around the full-spread diameter. Once you detect a possible threat, you orient towards it, shrink the umbrella, and then observe it more closely with your (less-extensible) long-range acquisition systems.
• You can't use a small hole in a big umbrella to look through for the previously-mentioned reason that you don't know a priori where you need to be looking, and slewing the whole ship around will send out a bunch of engine plumes.
• The necessary variable-size umbrella is going to be mechanically complex and much more so for an actively-cooled umbrella, which is why I neglected the possibility. I suppose, however, that you could have a small actively-cooled sort and a large passive sort.

That's all I've got for now. This is mostly an excellent thread, at the highest level of discourse I've ever seen here. I hope it is preserved.

[Darth Wong]

Thanks for the input, Erik. I would also like to add an extra wrinkle: unless we are assuming magic-tech space drives of arbitrary power, it is entirely possible that you could attack a far-range enemy observation post and the enemy could see you coming for weeks, yet be unable to do anything about it. A network of distant observation posts seems like it would provide an impenetrable defense until you ask the question: who is defending these observation posts? Is each one going to be heavily armed?

[erik_t]

That is certainly the case. What they do allow is for defensive spacecraft to have much lower total delta-v than offensive ones. This savings in weight and/or cost can be applied to improved weapons, or greater numbers, or whatever. If detection outposts further out allow you to put a defensive craft in position to intercept for 1/4 the fuel and/or drive power, that may be a very worthwhile thing. Unless the enemy intends to attack frequently over the same course (which seems suicidal), applying significant effort to destroying those outposts seems wasteful.

[Sea Skimmer]

The post does its job by forcing the enemy to destroy it. If the enemy does so not only do you know the enemy is coming, you know he categorically has hostile intentions and isn’t just trying to test your reaction time.

This is not really any different then many real life sensor systems like SOSUS or the DEW line which provided critical strategic surveillance and yet had no defenses either. Heck many DEW line radar stations were not only undefended, they were completely unmanned except for maintenance crews that dropped by every couple months to do a week of work.

[Darth Wong]

That depends on just how far away these observation posts are. It is entirely possible that the defender would need more delta-v than the attacker in order to scramble a spacecraft to the observation post in time, depending on the distance to the nearest staging area.

For example, just for the sake of argument, let's say each post is capable of identifying attackers at a range of ½AU, but you have several layers of posts out to a range of 4AU, and you only have a network of staging areas out to 1 AU.

[Junghalli]

Actually trying to defend the observation posts seems like it'd probably be a waste of resources for the most part. It seems a better idea to invest resources in:

1) Trying to hide their exact orbits from the enemy, so actually dismantling your observation grid would require a prolonged, tedious search.

2) Trying to find a way to build them cheaply so you can hopefully throw up new ones faster than the enemy can find and blow up all the old ones. They'd be expendable assets, like ammunition.

[erik_t]

That depends on just how far away these observation posts are. It is entirely possible that the defender would need more delta-v than the attacker in order to scramble a spacecraft to the observation post in time, depending on the distance to the nearest staging area.

For example, just for the sake of argument, let's say each post is capable of identifying attackers at a range of ½AU, but you have several layers of posts out to a range of 4AU, and you only have a network of staging areas out to 1 AU.

Oh, no, I am unclear. They don't need delta-v to protect the outpost; as SS says, the outpost is boned anyway. They need delta-v to intercept at the final target of the bogeys.

Beowulf raised an interesting point on AIM that trying to mix active and passive long-range systems could be troublesome. At Pluto's orbit (5.5 light-hours from Sol), your passive kinematic fix must be extrapolated five hours into the future to use active systems. However, I'm not sure that this is a huge problem. You just extrapolate into the future and look where the object should be based on Newtonian motion. If the object is not there, it is thrusting and ergo is an unidentified spacecraft, at least neutral if not hostile. If the object is there, then you can look it over more closely to determine what it is, be it spacecraft or natural object.

You still reduce the uncertainty from (neutral/hostile/natural) to (neutral/hostile) or you active-query the thing directly. The latter is preferable, but the former is a serious improvement.

[Ariphaos]

That depends on just how far away these observation posts are. It is entirely possible that the defender would need more delta-v than the attacker in order to scramble a spacecraft to the observation post in time, depending on the distance to the nearest staging area.

For example, just for the sake of argument, let's say each post is capable of identifying attackers at a range of ½AU, but you have several layers of posts out to a range of 4AU, and you only have a network of staging areas out to 1 AU.

Why would you?

Point your star at the idiot and be done with them. No delta-v required, just their location.

[Darth Wong]

Oh, no, I am unclear. They don't need delta-v to protect the outpost; as SS says, the outpost is boned anyway. They need delta-v to intercept at the final target of the bogeys.

These observation posts would take a long time to replace; it's not as if you can just drive a couple of transport trucks out and repair it. The enemy would presumably take out a few posts in order to effectively blind a sector before launching an attack, so there's no reason to assume that you would have gotten a fix on their attack force before you lost the outposts. Sure, you'd see the drone they launched at the outpost, for all the good that would do you.

There's always the idea that they won't know where your forward posts are, but that's one of those offense/defense things, and depends on how vigilantly they are watching your activity at the edges of the territory under your control. And they can potentially tell whether you have any undiscovered observation posts along their line of attack by launching probing attacks and seeing if you respond. I suppose you could conceal your knowledge of their approach by delaying your response, but that's a gamble (and it would also make for an interesting sci-fi plot).

Beowulf raised an interesting point on AIM that trying to mix active and passive long-range systems could be troublesome. At Pluto's orbit (5.5 light-hours from Sol), your passive kinematic fix must be extrapolated five hours into the future to use active systems. However, I'm not sure that this is a huge problem. You just extrapolate into the future and look where the object should be based on Newtonian motion. If the object is not there, it is thrusting and ergo is an unidentified spacecraft, at least neutral if not hostile. If the object is there, then you can look it over more closely to determine what it is, be it spacecraft or natural object.

You still reduce the uncertainty from (neutral/hostile/natural) to (neutral/hostile) or you active-query the thing directly. The latter is preferable, but the former is a serious improvement.

There's still the question of the accuracy of the passive fix, especially if you need to get it from a very distant observer platform because your forward outposts were destroyed and it will take you many months to replace them.