Cyber Security In Space

[Canis_Dirus]

Hello everyone, I am new to the board, and for my first thread I have a question for you all. I did use the search feature and didn't see an existing thread on it, so here we go:

I'm trying to design a proper warship computer system layout for an empire I am using in a scifi RP group. I'm not that knowledgeable about computer layout, so I'd like to ask everyone for their opinion.

So far my design is as follows:
-The ship has separate computer systems for weapons (including fire control and target acquisition sensors), navigation (engine systems, navigation sensors, etc), and communication (has separate subsystems for friendly and unknown/hostile channels). Each system has backup hardware.
-each of the above systems is managed by an officer on the bridge, who is assisted by a lesser AI (think 'dumb' AI from Halo)
-there is also a main computer system, along with a backup computer system. The ship's main AI (a 'smart' AI) runs on this. The main computer can be connected to any of the other systems via a physical connection, which can be engaged/disengaged by the bridge crew as needed.

Technology wise, the civilization that builds these ships uses quantum computers and AIs on a regular basis for their military hardware, and has a pretty large industrial base, so cost isn't as big of a deal.

[NecronLord]

This is how your warship will input directional controls into its guidance computer every time it wants to dock, if that guidance computer is isolated from all communications.

If you want to operate a fleet, and your comms computer is physically separate from your gunnery system, someone has to write down coordinates and vectors and physically carry those over to the gunnery panel every time the flagship orders you to engage a target.

By that time your enemy has zorched you.

You can run a faster computer, naturally, and slap a nice interface on it, but your problem of people putting in data via clip-boards while your enemies just push 'approve signed target package' is still there.

Might I suggest you try and trick rival players into abolishing networking, then you can keep its advantages for yourself.

[Purple]

The greatest way to ensure cyber security is NOT to isolate the computers from interacting but to isolate the people from the computers. Just keep everyone but vetted bridge officers from accessing anything vital AND make sure your ships are not networked with one another and you'll be fine.

[Starglider]

Total segregation of networks is a non-starter, because the benefits of fused sensory processing are so enormous. The direction of travel in software development is a highly redundant grid of sensor and compute nodes, with only the most latency-sensitive tasks restricted to local processors. Everything else is a compute job to be shared across the total grid of locally connected computing power. Certainly in a setting that has human-equivalent AI, you are far better off having active monitoring and counter-intrusion measures that bring all available intelligence to bear on shutting down any penetration, than you are trying to rely on passive blockades; an analogy could be made to obsolete passive battleship armour vs modern active defences (SAMs, lasers and point defence guns).

Furthermore if you have transhuman AI software will be far more reliable and practically impossible to remote-hack anyway (although sensors could still be spoofed). All current software is fairly awful due to the mismatch between human evolved mental capabilities and what you'd actually need to visualise the total interlocked state transition graph for significant software systems. Mature AI systems would not have this mismatch and nearly all the kinds of subtle flaws in software we currently have would be immediately obvious to an AI.

[Terralthra]

The brief period between omnipresent AGI and human-designed software will be rife with examples of AGIs taking over a ship's computer network by hacking it through passive sensors. Think this, but using it to obtain root access through the sort of hull sensors and cameras that any ship will have so it's not just a metal box floating about randomly.

[Jub]

I wonder if detecting the vector of the signal, vetting it against positions of known friendly ships, and then assessing the probability of it being broadcast by an enemy might help? If the signal is rejected the information isn't even read before being dumped. It's not foolproof, but it could prevent certain vectors of attack.

[Borgholio]

It sounds like you're trying to design a system that will protect against the kind of virus that the Cylons used in nBSG. That actually should not be difficult provided the network connection is limited. For instance, the gunnery computers don't need to have full access to the navigation systems...they just need to know the ship's current course and speed, as well as a basic feed from the external sensors. Likewise the gunnery officer on the bridge just needs inputs for ship's course and speed, enemy positions, and weapon status. Basic telemetry such as that would not be able to contain a virus, so if you hard code the systems to make it impossible to process any data other than the stuff that they actually need, they would be secure unless you had a saboteur on board to screw with the data streams directly. Plus, things like navigation and gunnery would have no business being directly connected to external communications...all incoming transmissions would be handled by a dedicated and isolated system that would act as a firewall / filter and prevent direct access to any critical systems. So even if someone managed to pass a virus through someone's email, it might infect their personal terminal in their quarters but would not be to get anywhere near the computers that control the ship's reactor, shields, weapons, or anything like that.

[Khaat]

Except in nBSG, it was a pushed software update before the Big Attack that laid dormant until it received a signal- the Colonials weren't "hacked", they were "updated".
The Galactica was later "hacked" (when networked to calculate jump coordinates), but that was possibly activation of old, original code "features".

Computers with "soft AIs" as mentioned in the OP suggest they would be designing your software architecture anyway. Provided they weren't corrupted, they'd keep the systems clean as part of routine operations.

[Borgholio]

True point, but the fact is what kind of software update could cause the weapons and targeting systems to shut down if they were isolated (aside from telemetry data feeds)? Even a back-door installed in a targeting computer would be cut off from the activation signal.

[Canis_Dirus]

Hmm it seems I should clarify what I was getting at better:

This system design is supposed to be developed in response to the fact that one of the people I'm at war with excels in hacking and electronic warfare, and used those to great advantage to kick my empire's ass in a battle. And yes as mentioned above part of this was done by hacking via sensors. My idea was to isolate the systems physically and connect only what needed to be connected, with switches to easily isolate systems if they are compromised to prevent the spread of malware.

Oh and Necron, the weapon system has its own sensors for fire control and such, and the navigation system has separate navigation sensors. The AI on each does most of the work, the officer there is mostly to oversee stuff.

[NecronLord]

Hmm it seems I should clarify what I was getting at better:

This system design is supposed to be developed in response to the fact that one of the people I'm at war with excels in hacking and electronic warfare, and used those to great advantage to kick my empire's ass in a battle. And yes as mentioned above part of this was done by hacking via sensors. My idea was to isolate the systems physically and connect only what needed to be connected, with switches to easily isolate systems if they are compromised to prevent the spread of malware.

Oh and Necron, the weapon system has its own sensors for fire control and such, and the navigation system has separate navigation sensors. The AI on each does most of the work, the officer there is mostly to oversee stuff.

Then this doesn't protect you as your fire control computer is still directly wired up to a port he can get in through. He can get in via your sensors and get to your weapon systems, arm your warheads in their cells or reverse you rail-guns programming so that they fire into their own breech, or whatever hilarity is possible once you get into the fire control computer.

Your ship is also carrying extra weight if it has separate sensors for navigation and fire control; what technology level is this roleplay world at - or rather how soft is your sci fi?

For that matter why bother having meatbags in the ship at all? Is it a setting rule or can you dispose of the delicate little painflowers and automate your ships?

More seriously, you're gimping your systems for a benefit that is marginal at best. Just follow IRL best practice for defending your network.

[bilateralrope]

True point, but the fact is what kind of software update could cause the weapons and targeting systems to shut down if they were isolated (aside from telemetry data feeds)? Even a back-door installed in a targeting computer would be cut off from the activation signal.

That depends on the nature of the activation signal. If it's a specific radio message, then you could stop it being received by the compromised code.

If the signal is something along the lines of ships flying in a specific formation, you can't stop the compromised targeting software receiving the signal without reducing its value as targeting software.

[Simon_Jester]

Hmm it seems I should clarify what I was getting at better:

This system design is supposed to be developed in response to the fact that one of the people I'm at war with excels in hacking and electronic warfare, and used those to great advantage to kick my empire's ass in a battle. And yes as mentioned above part of this was done by hacking via sensors. My idea was to isolate the systems physically and connect only what needed to be connected, with switches to easily isolate systems if they are compromised to prevent the spread of malware.

If I were trying to address this flaw, my priorities would be:

1) Figure out how to hack via sensors and stop it. This should not be inherently hard, especially if you have AI smart enough to grasp basic concepts like "literally no software changes come in through the data flow from the radar antenna. EVER. Those executable files? They're intruders. Get them." Building computers capable of differentiating between 'data' and 'software change' is not hard in principle, it's just a question of implementing systems that are capable of screening out the undesired input.

2) Disable ability to reprogram software during battle without some kind of root access that you can't get without being the captain of your ship (or possibly your ship's IT officer and electronic warfare officer). As far as the gunnery, navigation, sensor, communications, and other software are concerned, in battle mode all the ship's .exe files are read-only. If nothing can be reprogrammed, nothing can be seriously hacked.

3) Make the fastest possible "reset from backup" system for your computers. If the hacking doesn't manage to instantly cause an explosion or something*, and the problem is just that it's interfering with your ability to fight using the ship, that can be countered. You just need a backup computer that can be reset to a default state that is just like the pre-hacking software. The disadvantage is that it would be unaware of conditions during the battle because you do NOT want it running off the same hard drives as the main computer... but it's better than flying around with compromised fire control software.

4) But basically, if you're in the middle of a battle and you're forced to resort to manually shutting down the connections between computers and transferring information by hand between physically isolated computers... If you ever needed those computers to work together, if there was ever a reason you connected them, then you've already lost. Because your combat effectiveness will drop off so sharply that against a peer-competitor opponent, you're meat on the table.
______________________

*(which arguably should not be possible, machines should be designed to that a software-only glitch can't blow them up, with missiles having fuzes that won't arm until the missile has left the ship and so on)
______________________

Of course, if your problem is that you're dealing with a DM (or DM-equivalent) who thinks hacking is omnipotent and doesn't understand what its limitations are, you can make a straightforward series of suggestions to shut up the ignoramus. But that's a separate issue.

[Zixinus]

Everything on a warship, modern or futuristic, needs to be designed from a security standpoint.

One form of defense in specialism: you do not have regular updates like with commercial software. You have a fixed, very-well debugged and tested software for all necessary functions and only update rarely with security locks (for example, firmware updates should require physical switches being flipped or similar security measure). Any deviation detected in that software? Shut it down and restore from read-only memory.
Stuff like this should be built into operating system to begin with. Actually, stuff like this would make use of human systems rather than computers doing all the work: certain operations are hardware-made to only happen with human commands (firing weapons, retargeting). Humans should have overrides for everything.
There should be watchdog programs on every system that monitor and independently assess activity. You can have a security loop that continuously encrypts and decrypts status information and anyone braking that loop would be suspicious. There should also be something like this regarding sensor feeds. If you can gain access to a camera's computer network by using lasers carrying code, something has gone seriously wrong with that camera.

Another idea is creating hierarchical layers of systems that have fixed information updates and do not exchange code.
For example, the most deepest network on a warship would deal with life-support, controlling the reactor, doors and anything directly vital to the crew. Other systems can read data but not make orders or give code to this layer. This layer can make overrides to higher layers, such as forcing a comprimised system to shut down.
Another layer for external sensors and navigation. Maybe squeeze communications somewhere.
Another layer for weapons and anti-weapon devices.
This way a hack on one system can only spread so far and cause only so much damage. Yes, one layer comprimised can still reduce the effectiveness of the ship in battle but this limits how much damage can be done.

The greatest way to ensure cyber security is NOT to isolate the computers from interacting but to isolate the people from the computers. Just keep everyone but vetted bridge officers from accessing anything vital AND make sure your ships are not networked with one another and you'll be fine.

This is a warship, not smartphones. Physical access control, compartmaninalisation, secure access is a given. Somebody that hasn't been vetted should not even be on the warship.

[Purple]

This is a warship, not smartphones. Physical access control, compartmaninalisation, secure access is a given. Somebody that hasn't been vetted should not even be on the warship.

And yet it is worth pointing out. If for no other reason than because traitors and infiltrators make for very good characters and plot lines to explore. And RP is all about those.

[Zixinus]

And yet it is worth pointing out.

Military computers work in a very different paradigms of problems. Think in terms of securing your house versus securing a military base. Securing your house will make burglars less likely. For a military house housing military things, burglars have to be treated as a certainty.

For example, for a warship to be effective, it needs to have some sort of networking with other warships. Data, intelligence, orders, etc. needs to be shared live. It's not like a civilian workstation doing tasks that just require that workstation. One warship cannot see everything, so they need to share their sight. Warships firing weapons and using anti-weapon countermeasures working together are more effective than a bunch of warships working independently (your mileage may wary with tech of course). That's what makes a fleet a fleet, as opposed to a group of ships. A warship that is forced to operate in complete isolation on a battlefield is simply a moving target that can shoot back and is just waiting to be destroyed.
So what can you do? You make several layers of security that make enemy hacking as costly and ineffective as possible. You encrypt, you encrypt what you encrypt on a different level, you run chains of authorizations, verifications, etc. to prevent weak security links, etc.

As for physical access, spaceships are very controlled environments to begin with and that is a very different sort of problems. A traitor or infiltrator installing bad code is relatively smaller compared to, oh, gunning down the crew from the inside or blowing shit up. But this is a very different security problem, in a different direction. In a warship on a battlefield, you want to stop bad code from getting in. Stopping bad code from getting out is another matter. Stopping bad code from getting inside is a different problem again.

[Purple]

I know all this. I happen to be a programer and all that. Which is why I focused on the character angle because frankly I know how to competently design such a system (in theory anyway) and I know describing it would be boring in writing. It all always just boils down to:
- Make changing software impossible other than by physically exchanging the hard drives containing it.
- Limit networking to only carry needed information.
- Use proprietary hardware and software compared to foreign civilizations whose concepts of computing evolved in a completely different environment so much so that NOTHING is going to be compatible on any level anyway without you even trying to make it so.
- Completely ignore the concept of jamming and ECM in general so as to even allow networking between starships to function.

[Lord Revan]

Tbh you simply cannot make system that's 100% traitor-proof while still being useble to loyalist (since the very defination of a traitor is someone who used to be a loyalist) at most you can make a system that's secured against known traitors.

EDIT:What you want to do in the end depends on the needs of the narative. That should be the driving force, it doesn't matter how "realistic" you are if the narative falls flat or seems contrived.

[Zixinus]

I know all this. I happen to be a programer and all that. Which is why I focused on the character angle because frankly I know how to competently design such a system (in theory anyway) and I know describing it would be boring in writing.

Yes, but in practice the ideal way of writing that software is never quite going to work in practice because there are other goals in making the software. Design compromises have to be made and compensated for.

[qoute]- Use proprietary hardware and software compared to foreign civilizations whose concepts of computing evolved in a completely different environment so much so that NOTHING is going to be compatible on any level anyway without you even trying to make it so.[/quote]

This assumes an alien civilization as opponents. That may not be the case. It easily could be opponents that even have the same tech base (civil war).

- Completely ignore the concept of jamming and ECM in general so as to even allow networking between starships to function.

Erm, what? How would this make things more secure?

EDIT:What you want to do in the end depends on the needs of the narative. That should be the driving force, it doesn't matter how "realistic" you are if the narative falls flat or seems contrived.

Yes, but a good place for a writer to start writing a fiction is to see how something like this should have happened and deviate from there.

[Simon_Jester]

- Use proprietary hardware and software compared to foreign civilizations whose concepts of computing evolved in a completely different environment so much so that NOTHING is going to be compatible on any level anyway without you even trying to make it so.

This assumes an alien civilization as opponents. That may not be the case. It easily could be opponents that even have the same tech base (civil war).

It's still a good idea, though.

Because it DOES help if you're dealing with aliens, or with a foreign country whose technical base hasn't been shared with your own for decades. For instance, Soviet and American computer operating systems in the 1980s were not compatible, except insofar as some Soviet systems were copies of American systems. And military computer systems were simply not thus copied, period. The Soviets weren't aliens- but their computer technology had been evolving mostly independently for a long time.

Unless you're the All-Spanning Galactic Empire and your military is the ONLY military, there will be foreign powers that use different and incompatible hardware and software. Encouraging this is to your advantage if you're worried about being on the losing side of a hacking war.

[Purple]

Yes, but in practice the ideal way of writing that software is never quite going to work in practice because there are other goals in making the software. Design compromises have to be made and compensated for.

True, but nobody likes a RP where the posts look like something out of a technical manual.

This assumes an alien civilization as opponents. That may not be the case. It easily could be opponents that even have the same tech base (civil war).

If it has come to a civil war than your biggest worry will be personnel though. Traitors being easier to find than hackers.
And I don't mean the kind of traitor that walks up to your starship PC with a floppy disk. I am talking about the kind that will give out the exact details of your sensor equipment and targeting software to an enemy who can than find and exploit loopholes in the logic.

Erm, what? How would this make things more secure?

Not more secure but possible. Simply put if we were to take modern levels of jamming in say air combat and extrapolate onward into the AI age we'd end up with every physical battle being accompanied by a virtual one fought by AI who jam and counter jam one another in a continual dance that frankly is going to make networking separate starships together impossible or at least very difficult.

So basically the situation is that you can have realistic hacking, realistic jamming, pick one.

Yes, but a good place for a writer to start writing a fiction is to see how something like this should have happened and deviate from there.

To an extent, but for that we would need a LOT more data about his setting.

[Zixinus]

If it has come to a civil war than your biggest worry will be personnel though. Traitors being easier to find than hackers.

Yes, but both have to be treated as an eventuality. If the cost of hacking is low then it will happen.

It's still a good idea, though.

True, but it's still security trough obscurity. Can work but never solely rely on it. Aliens can capture and study your technology.

Simply put if we were to take modern levels of jamming in say air combat and extrapolate onward into the AI age we'd end up with every physical battle being accompanied by a virtual one fought by AI who jam and counter jam one another in a continual dance that frankly is going to make networking separate starships together impossible or at least very difficult.
So basically the situation is that you can have realistic hacking, realistic jamming, pick one.

That's a very lazy approach to the problem.

[Purple]

That's a very lazy approach to the problem.

You'll find that realism often is. The real world is, sadly, always far more boring and mundane than what we'd like. That is why we have fiction to brighten it up.

[Simon_Jester]

I will observe, grudgingly, that Purple has yet another point here. I mean, it's not like people in modern fighter jets fly around trying to hack each other's computers by way of the radar or anything.

[Sea Skimmer]

Not more secure but possible. Simply put if we were to take modern levels of jamming in say air combat and extrapolate onward into the AI age we'd end up with every physical battle being accompanied by a virtual one fought by AI who jam and counter jam one another in a continual dance that frankly is going to make networking separate starships together impossible or at least very difficult.

No, with laser communications in space directed by steering, shrouded laser mounts its the jamming that will become nearly impossible to make effective. The only way it will be possible at all is to place the jammer almost directly between the two platforms trying to communicate, and since a network could involve more then two nodes such a jammed link could be bypassed. Recognition of such jamming will be very easy too, making any kind of spoofing attack all the harder. Plus with large platforms like spaceships implementing one time pad encoding of real time communications is actually feasible.

I will observe, grudgingly, that Purple has yet another point here. I mean, it's not like people in modern fighter jets fly around trying to hack each other's computers by way of the radar or anything.

Yeah, actually we kinda do these days, and the US fully exploited a number of a Soviet era air defense communication networks. But omni directional and skywave radio signals are much easier to mess around with then optical systems in the first place, and there is generally a considerable lag time in this sort of process. Aka US exploited stuff used in the Yom Kippur War by the end of the 1980s. This is also why War Emergency Only modes are very important for secure communications and sensors.