Hardware: Decide upon chosen hardware; microcontroller vs. device on shuttle

[doriangc]

This one should definitely be a group decision, so anyone who wants to contribute to the hardware, feel free to assign to yourself.

Addressing LoRA vs. LTE-M This will depend on whether we are able to achieve the desired range with LoRa. At the moment, it seems LoRa may not be able to deliver on required range due to terrain blockage on the route.

This is a radio path study between Brown building and Jeffco Government Center (terrain-only, no buildings). It's looking a bit tight, especially considering the distance.

Antenna would need to be placed at a height of 145m above ground level (at location of Brown) to guarantee line-of-sight. We don't necessarily need line-of-sight, but terrain does definitely affect range.

LTE-M is an LTE standard designed specifically for IoT devices. LTE-M data plans are fairly cheap (much cheaper than normal LTE plans), so we may just want to go with that. Example from AT&T, unlimited plan.

There a few options I can think of (feel free to add if you can think of more viable ones). We could also start with either a prebuilt or semi-prebuilt dev kit and then progress from there.

A fully pre-built LTE+GPS device (like a cheap smartphone or other widely available hardware):

Advantages:

• Much easier to implement, no HW design
• HW is proven to work well in advance
• Easier development, just use standard Android/iOS SDK
• Easier integration with the rest of the system (ie: lots of shared code).

Disadvantages:

• Potentially higher cost (even cheaper phones are >$100)
• Some lack of flexibility/fine-grained HW control.
  • Example: every time the phone reboots, the driver would need to re-open the app.
  • We also wouldn't be able to easily add external hardware, like buttons, lights, etc.
• In the case of phones, they are not necessarily designed to be plugged and in potentially warm/cold environment for years at a time.
• If using a phone, LTE plans are much more expensive than LTE-M plans (~$200 vs ~$30 a year).

Potential candidates:

• Motorola Moto G Play
• Nordic Thingy 91 (this could also be used as a dev kit if building custom hardware)

A pre-built microcontroller boards with "expansion" boards

Advantages

• Much more flexibility than fully prebuilt, we can add external buttons, etc.
• Don't have to deal with Android/iOS SDKs and their limitations
• Potentially lower cost

Disadvantages

• Less shared code
• More work than fully prebuilt (we would need to design an enclosure, pick correct parts).
• May end up as mess of wires shoved in a box.

Potential candidates:

• Adafruit Featherwing + Adafruit Feather
• Pi Pico + LoRa/SIM7080G module + GPS module
• ESP32 board + LoRa/SIM7080G module + GPS module

A fully-custom circuit board

Advantages

• Probably best learning experience (we design everything from the ground up)
• End result looks clean
• Every part can be tailored towards the application
• Could potentially be less expensive
• Smaller form factor
• We get to use the hot skillet to reflow!!
• Best power efficiency (not super important for a bus that constantly outputs 12V, but still)

Disadvantages

• Lots(!!!) of work (Impedance matching, RF can be hard but not impossible), it will take longer than other options.
• Will requite dev kits/temporary breadboard to develop on first
• Everything is custom (no shared code)

Potential candidates:

• Nordic 9160-based (it has an ARM Cortex-M33 and basically everything we need in one SiP!!)
• RP2040/ESP32/Atmega + SIM7080G/Quectel B660K/u-blox SARA R510s chip

[doriangc]

The hardware has been finalized.

An RP2040 will be used along with an SIM7080G.

For development, a Raspberry Pi Pico WH will be used along with the Waveshare SIM7080G dev board (https://www.waveshare.com/pico-sim7080g-cat-m-nb-iot.htm).