Years ago I stumbled upon article describing geolocation logger for animals, allowing biologists to track their position. The device was self-sustained, logging into its memory, available to be downloaded after retrieval.
As usual, no details were released, I just remember method of determining position from time of dawn and dusk - it certainly piqued my interest. I decided to investigate more into this method... one day. And coin cell contest gave me good reason to start it.
This project ran in two phases, with two different hardware variants built:
1, At first I have no single clear idea how to turn light input data into geolocation output - though I had partial ideas that can be employed. In order to make the decision easier, I built simple and "disposable" loggers I sent around the world to see and log as many sunsets/sunrises as possible. Those loggers were designed to not make any geolocation output, just logging light levels; and were of no use after the mission is done.
2, Once the geolocation algorithm was clear, I made the final LLG hardware, providing geolocation output on display, also with logging capability.
The accuracy of geolocation seems to be in order of dozens of kilometers. It depends strongly on surroundings, though in really unsuitable conditions it can provide valid results within few hundred of kilometers.
The LLG has calculated run-time of nearly 10 years with LCD permanently on, or over 30 years with LCD off. EEPROM capacity is for more than 22 years worth of logging.