Towards distributed wilderness search using a reliable distributed storage device built from a swarm of miniature UAVs

Searching a wilderness environment for missing persons is a task that is well suited to a swarm of miniature UAVs, given the high availability of low cost drones and the large geographic area that is to be covered. The use of a swarm for detection provides an alternative to automated detection solutions that require high processing power to search all images encountered for all possible combinations of items that may be detected. Challenges faced in implementation include limited storage and processing capacity per node, the designation of data storage points when the swarm is far from any base station, and inhospitable wilderness environments that may destroy UAVs. The solution that we employ leverages on the use of distributed in-network storage that allows each UAV to behave as a sensor node which is specialized to the task of detecting certain features. Our data storage strategy self organizes in an efficient way using a randomized selection of backup addresses, and is robust to the loss of data even after events that cause the destruction of a large number of nodes. Analysis and simulations show the survivability of data, how the communication latency varies with the number of backup addresses and that a set of simple processors can be used to collaboratively perform distributed searches with relatively high accuracy.