Low-power LoRaWAN soil-moisture node with a battery-life budget

Krishna Valley's growers irrigate by habit, often twice as much as the vines need, and the co-operative's agronomist believes they could save a third of their pumping water and electricity. To prove it, she wants 60 soil-moisture nodes spread across 20 farms in a pilot, each reporting every 30 minutes to two LoRaWAN gateways already on the co-operative's office and cold store, 3 to 5 km away.

Her constraints are practical: - each node runs on two AA lithium cells and must last at least two years without a visit, because nobody will walk the rows changing batteries; - the node cost, without the enclosure, must stay under ₹2,500; - readings must be within ±3% volumetric water content after calibration for the region's black cotton soil; - the nodes must use the IN865 LoRaWAN band and stay within the network's fair-use airtime.

You will write the requirements, calibrate a capacitive soil-moisture sensor against soil samples you prepare at home, write the firmware in Wokwi or on any Arduino-compatible board you own, design a compact payload, calculate airtime for each spreading factor, and build a battery-life model from datasheet currents. You do not need LoRa hardware. The radio is represented by a stub in the simulator, and its current and airtime come from the datasheet and the LoRa airtime formula.

Done looks like a node design the co-operative's electronics vendor can build: requirements, calibration curve with error, firmware that spends almost all its time asleep, a payload of a few bytes with a decoder, an airtime and duty-cycle table, a battery-life estimate with its assumptions and a pilot plan. If two years is not reachable at the worst spreading factor, say what has to give.