CBTC is more complicated than "a wifi network and servers" unfortunately. And it's not a great business.

Firstly, no, WiFi is not going to work. WiFi runs in the junk bands like Bluetooth does. It's designed to run in unregulated spectrum, not be reliable, which is why WiFi is a synonym for "why does my internet not work today". Just think of interference from the riders phones, for example.

You also need to know the location of the trains to a much higher degree of accuracy than what WiFi would allow. The trains use transponders on the track that antennae pass over. It's closer to NFC than WiFi. When the train knows where it is, it can communicate that info to the control rooms but trains spend a lot of time inside tunnels which are not exactly famous for their excellent radio propagation properties. There's a lot of testing and planning work required to figure out where to install antennas and cable uplinks. Remember that track maintenance hours are severely constrained on any live transit system so reducing the amount of equipment to install is important.

Then you have redundancy. When your home wifi router goes on the fritz it's annoying but not that big a deal. If a part of the train monitoring network goes black, done naively that would shut down an entire line, which is catastrophic. So you want redundancy, you want spare parts inventories, you need to train staff how to do the replacements etc. Every train needs to be brought back to shop and refitted, usually using custom equipment and designs because trains are not standardised across the world.

That's just the hardware.

The software needs to be developed and tested too. Again, no standardisation in how transit systems work, so each deployment is a custom job requiring the engineers to have a good understanding of each railway with all its quirks and complexities, especially in exotic signalling and switching arrangements.

Still, in the end, $19 billion is too much. The NY premium.

I don't know why you got downvoted.

I've toyed with the idea of doing a popular subway improvement project where we'd make battery-powered devices (arduino, rpi, whatever -- the biggest cost irl would be labor) to plant on subway cars and stations for tracking trains. A proof of concept could be done with just a few tens of such devices. They'd use the subway wifi to transmit train locations.

Of course, there are complications. There's no such thing as a head train car -- these things get mixed and matched. So ensuring that every train has such a device might mean ensuring that every car has one or that every train gets a car that has one, and this seems difficult. But even so, this could cost just in the millions done right.

Obviously the price tag for the NY system is crazy, but honestly so is what you guys are proposing... It's almost completely impossible to certify a safety critical system with WiFi and Arduino/RPi. This kind of engineering is a whole different ball game, and for good reason (train crashes and derailments are really not pretty).

For an informational system (i.e. telling passengers how long trains are away or for statistics gathering), sure. You could absolutely do that like you say. But as soon as it involves controlling the train or anything to do with signalling, it just can't be done like that.