Leaving aside the weapons and sensor capabilities for now, I'm going to take a look at what it takes to hide this thing.
I'll repeat a few parts of the specification (including this comment) here:
- This system consists of ... interconnected units deployed around a planet
- Provide universal coverage at any single point in time
- Undetectable until armed and in use by any society with a level of technology less than what it took to build the grid
So we need a lot of tiny things whizzing all over the planet. Sounds to me like a satellite network.
Satellites are visible to the naked eye. Big ones like the ISS are brighter than the brightest stars, but even small ones down to sixth magnitude are easily visible on dark nights due to their rapid movement. Typical ELINT and Earth observation satellites are mangitude 4.5 or higher with good viewing geometry.
The dense satellite constellation needed would easily be visible in areas with low light pollution, say, anywhere in the world before 1500. The Greeks, Arabs, and Chinese all were avid astronomers in their times, and would not doubt notice the satellite constellation. It would probably take until the early 1600s for the nature of the satellites to be determined.
Even if the satellites were the size of cubesats (10 cm), they would be easily detected by Cold War-era radar. Nowadays we can detect centimeter-sized
space debris microsats.
And if they're smaller? Good idea, but we can't. You'll want to know what's happening on the surface, so you'll need sensors. Take a look at the WFIRST telescope. The bus and optics are an obsolete (that's right) NRO Earth-observing satellite, and therefore somewhat representative of current 'spy satellites.'
Even this huge, 6.5 ton satellite has a resolution worse than 20 centimeters at the Earth's surface (or much worse, depending on orbit). If you try to make it any smaller, diffraction will blur your image until it's useless. ELINT satellites are even bigger, even those using SAR to reduce antenna size. This is a physical limit, not a technological one.
Paint it Black
Unfortunately even if we somehow manage to absorb all the sunlight incident on our satellites, we will still emit blackbody radiation, which will be visible in longwave IR. This will only delay discovery until infrared astronomy is invented, or until radar is invented. (Even a perfect blackbody will scatter EM wavelengths much larger than the particle size.)
Ok, so we can't do this in space. Let's try going the opposite way, making our sensor grid so small it can't be detected.
I have a few problems with smart dust in general, mainly that everybody just handwaves the issues of energy consumption (no physically viable energy source at that scale), mobility (all the 'motes' will be overwhelmed by Brownian motion), and computational power (critical density of motes required is too high).
Even ignoring all these issues, there are two important facts that spell doom for this scenario:
- The motes have to be everywhere for the system to work
- The electron microscope was invented before nanotechnology
Image scientists' surprise when they find every object they examine coated in a fine layer of dust!
I won't try to speculate on what sort of space-warping technology future civilizations might have under their belt, but consider this line of reasoning:
- To use something (a technology or physical principle) you must first learn how
- Learning requires observation
- Therefore, civilizations will always have the technology to study (and detect) a particular technology before they have the technology to utilize it.
Undetectable vs. Undetected
I may have painted a grim picture for any would-be secret planetary overlords. However, don't despair! Just because we can be seen doesn't mean that we will be seen! We just need to hide somewhere the humans won't look. (For the rest of this answer I'll assume a somewhat more advanced culture hiding from present-day humans.)
We don't want to be too far from the surface of the planet so that the system can be deployed quickly. Our options are therefore limited to low Earth orbit. Unfortunately humans monitor the space immediately around their planet very closely, both optially and by radar.
They also do the reverse: observe the surface from space. With the huge variety of instruments providing constant coverage, plus the above-mentioned defense imaging satellites, the surface is also well-covered.
However, the humans do not carefully search below their planet's surface (in either the lithosphere or the hydropshere). Although they use advanced seismic techniques to search for buried hydrocarbon deposits and monitor each other's use of nuclear devices, most of the lithosphere is unmapped. Their hydrosphere is similarly poorly mapped.
Therefore, to avoid detection, we can bury small elements of our system near (but not too close) to locations of interest, while hiding the larger elements deep in the oceans.
Obviously with this method we lose the ability to closely monitor surface activities with our own instruments. However, humans have always been obsessed with monitoring each other, and now provide an easily-accessible global network over which all their information can be accessed.
In order for the grid to be an effective tool for C4ISR, we must be able to communicate with it. Humans monitor almost all of electromagnetic spectrum, making this difficult. However, they do not closely monitor the range between microwave and submillimeter, around 100 to 500 GHz.
A narrow beam, high bandwidth burst with just enough power to reach high orbit should be almost undetectable. A single relay satellite in a highly inclined geosynchronous orbit will be almost unnoticeable. It should relay data to us over an x-ray link, which has the advantage of being so narrowbeam (about 1000 times narrower than optical communication, and 100 million times narrower than typical satellite X-band uplinks/downlinks) that the signal cannot be received by planetbound observers. (We can't use x-ray directly, however, since they are block by the atmosphere.)
Now that we're well hidden from the humans, we can start thinking about the weapons we should use on this weapons system.
For the smallest tasks (say, an individual human), aerodynamically propelled vehicles are probably the best option. Launched from our land-based locations, they will not have to travel far and can therefore be small and stealthy. Using a high-energy fuel can serve a dual purpose, incinerating or otherwise destroying the vehicle to prevent discovery.
At the larger levels of destruction, we no longer need to be concerned with stealth. Simply float the underwater emplacements toward the surface and from them launch cruise missiles, ballistic missiles, and satellites.
Chemical explosives are generally the most effective against personnel and most structures and vehicles. However, military structures and vehicles are designed to withstand this type of attack. Instead, move to shaped charges and hypervelocity kinetic impactors, propelled chemically or electrically. A combination system can penetrate hundreds of feet into rock to destroy even the most hardened installations. Finally, deep targets can be destroyed by orbital bombardment once your satellites are in place. This is the slowest but most powerful method, so save it for high-value targets.
You may also consider nuclear weapons for the highest levels of destruction. Their radioactivity is totally shielded from detection by miles of water when hidden, and they are incredibly effective, making them useful tools for wiping the planet clean. However, if you want to move in soon after the dust clears, consider instead a targeted bioweapon. It can remove as much or as little of the flora and fauna as you wish, depending on design, and you can dismantle and demolish any inconvenient infrastructure remaining at your own leisure.
Just make sure you have a variety of systems and sizes, to match the threat to the target, and to make the system more difficult to defend against. Happy hunting!