Getting a satellite into space is ~50,000$ for a 1U cube sat when I looked around. You might be able to get NASA or somebody to send it up for free if it's educational and they have room.
So it's getting cheaper for sure! While that's certainly not the cheapest, it's now within reach of small startups/passionate makers which is exciting!
More likely in 50-100 years is manufacturing capacity in space gets to the point that building a 1u satellite up there is way cheaper than sending from a deep gravity well like Earth's. I imagine at some point, rockets will only be used for sending people and other things that can't be transmitted or manufacturered.
Still going to pay transport costs on getting raw materials to space, although I imagine raw mats will be densers and the volume savings will reduce costs.
But on currently on earth we have many massive factories to collect, refine, and process just the raw materials. Those factories will require massive amounts of resources to boot, and we don't start with any resources in space....
Right, so it's a bootstrapping question. If this was your long term goal, you'd right now be working on reusable rockets that you'd soon be using at cost to send up large quantities of infrastructure (cough cough Blue Origin and Space X, Space X being further along and openly talking about Starlink). Next step is mining asteroids for stuff like platinum and dropping it down to earth. Step after that is manufacturing with those raw materials in micro G. Now you have a complete manufacturing chain without gravity wells, and ostensibly a multi trillion dollar company.
You'll either need a mesh network (higher latency) or you'll need a backbone (which is an ISP) so you'll still run into similar issues. Not to mention at the lower bands you have limited bandwidth, not just for yourself but for everyone, so bandwidth congestion will become a huge issue if everyone used it for internet.
That said I do hate my ISP (and most others) with a burning passion so...
Could a mesh network be created on existing WiFi hardware? For example with a firmware refresh / OpenWRT? Then it’s just a matter of mass adoption / critical mass in neighborhoods.
I'm going to use Java, the tricky parts will be to build the frequency hopping (for me 410-525MHz) + power lowering (-1-+14/+20dBm) parts so that densely populated areas can function properly.
Largely NodeJS with https://www.npmjs.com/package/serialport but for some of the more ethernet-centric integrations I've been using C++. I haven't gotten to dynamic frequency setups yet, I've actually been lazy and using 2 of these on each side to act as separated TX/RX in different frequency bands. Certainly doesn't scale past a dozen but it made things a lot easier to get started.
I'm curious on your choice of optimization algorithm to balance cell size/interference with relay congestion.
You have access to this, you just don't want it. I like the cushion of corporate incentive which holds up the vast majority of the internet infrastructure. It is very inexpensive to access it, and it remains relatively open. Just because you don't have experience operating or directing a corporation, doesn't mean they are all evil entities which ruin everything.
There's a lot to overcome (having everything be completely decentralized), but a person can dream...