Mike Burr - log

[comp] Bombing them forward to the space age

We are very close to having the following become practical. I describe it in hokey-story form:


Knowing that delivery by airplane or missile is not going to be practical for numerous reasons, we decided on a giant slingshot! Slingshots surprisingly are not regulated by any international body. And we built a really really big one!

This was our 2nd successful insertion.

We sent 17,482 nodes, all identical. Why this number? It's a long story. It should be clear that we need many.

Node specs:

  • 18 month battery (23 g)
  • multi-mode antenna + soft/hardware radio.
  • endless clever software
  • small, durable packaging
  • cheap "af"
  • overall mass: 42 g

What happens is these many many nodes end up hundreds of meters away from one another and spread out over a large area, sometimes several kilometers in extent. It works out to being a giant ellipse; major axis ~1 km, minor axis ~0.5 km.

The distribution of nodes in this area is fairly uniform. The "ball" (as we call it; this being the projectile loaded into the slingshot) has a crafted shape and density profile and the nodes themselves have special aerodynamic properties that help ensure this.

Plans exist to have more control over the shape and distribution of the mesh, in a geographical sense. The benefits of this are endless but just to cite a couple of examples -- We might want to end up with the henemy's positions ending up in "holes". We might have a desire to control the balance of node longevity (which is correlated with mass) and the density of nodes in an area. Maybe we want just a "causeway" of long-lived nodes with two "blobs" on either end?

Consideration has also been made about the aerodynamic qualities of the ball itself: As it has some rigidity when packed tightly and as there is ultimate control of node design and their arrangement, we can have what we call "quasi-structural" areas that can control the ultimate distribution; maybe give the ball some initial lift, or perhaps spin. Its behavior can change throughout flight as areas "oblate" (nodes peel off or drift away). We are able to make a kind of arbitrary "metamaterial" with our node-packing.

Oh and it's not really a slingshot. It's more rails and rams and stuff. We used steam both times. Cheap, plentiful, predictable...

Insertion testing is cheap, easy and risk-free from a strategic point of view. We have several testing groups that can print and assemble dummies, set up their slingshot, launch, collect data and still be home for dinner.

The printed dummy nodes are printed in bright, highly-visible colors allowing a drone to measure their distribution with sufficient accuracy. With a good enough camera, 99% of the dummies' insertion points can be measured to an accuracy of centimeters.

It's big. 7 meters long and about 4 meters high, but it's stuff you could source from a junkyard and assemble in-place. They'll be looking out for them going forward, but only to buy them time as they well know.

No, we won't go overboard. They are cheap after all, and easy to insert. The threat (and growing practicality) of orbital insertion sends a strong message of futility to the henemy.

As for this insertion: Roughly speaking if one imagines shooting a shotgun charge with many small pellets, say "birdshot", at about a 45-degree rise and then one imagines the location of these pellets once they reach the ground, one would have a good idea of where all of these nodes ended up and the rough distribution of them.

Once manufactured, these nodes immediately begin communicating. They start out in a simple management mode where there is no attempted intercommunication and they are only awaiting instructions.

For this application we instruct the nodes to go into sleep mode and use a timer to wake them up at a designated time. This is done in this particular application because we don't want any radio signature to be present until after deployment. You can probably guess why!

The nodes are, given what is occurring, launched rather quietly and they all take their place awaiting their timed revelation. When the timer goes off these nodes immediately begin forming a "mesh network". Every node has two jobs: 1) form a network with its immediate peers for the purpose of propagating data from one part to the other of the mesh and 2) attempt to discover Wi-Fi networks which can be used as a means of communication with the Internet at large.

Individual users (that is locals) are able to use this mesh for general Internet access. Each node has an adaptive AP mode -- It advertises itself with low power and switches to high-power AP mode when a host joins. Software-defined-radio can be used to communicate, albeit slowly, with a nearby node using the proprietary published mesh-native protocol.

As we don't want this network to end up isolated, we arrange for high-power open access Wi-Fi points to be within range of the mesh at the start. It is however possible for this mesh to discover and use new Internet gateways which we expect to become available as time goes by.

The client problem solves itself. Any cellphone and a bit of Joe Texter fiddling will get one connected. Wish someone had bombed my hometown when I was a kid.

Indeed if the henemy becomes aware of this mesh, as they inevitably will, our starter gateways will become an obvious target. Naturally many Wi-Fi gateways appear, become available, and are used by the mesh in a short amount of time. And even if this does not become the case, evasion is easy and wastes the henemy's time. Organizational purpose-made gateways easily provide the mesh with more than enough bandwidth and reliability and will continue to do so.

As mentioned the lifetime of the mesh overall is approximately 18 months. During this time the geographical area serviced by the mesh is in a sense "hardened"; better communication enables better defense. This makes it much easier to deploy, broaden and "freshen" the mesh. The mesh grows with time.


Use-case: DPRK et al

- 1 toast