Ripple-2: a non-collaborative, asynchronous, and open architecture for highly-scalable and low duty-cycle WSNs

The design of Ripple-2, a wireless in-situ soil moisture sensing system is presented in this paper. The main objective of such system is to collect high fidelity and fine grained data both spatially and temporally compared to radar remote sensing, which is the more traditional way of capturing soil moisture, and to use the former to validate and calibrate the latter. To do so, the in-site sensor network must cover a sufficiently large area, on the order of at least a few square kilometers. At the same time, cost constraints (both in deployment and in maintenance) puts a limit on the total number of sensor nodes, resulting in a very sparse (on average) network. The main challenge in designing the system lies in achieving reliability and energy efficiency in such a sparse network. For instance, in our pilot deployment, a 200mx400m area is covered by 22 nodes (average inter-node distance > 50m). Traditional WSN technology typically calls for many more nodes to be deployed in such an area. Ripple-2 is introduced as a non-traditional WSN architecture where (1) the network is physically and logically segmented into isolated clusters, (2) a regular node (or end device, ED) only communicates with the cluster head (CH) of its segment, and (3) the ED-CH communication is distinct from the CH-sink (or CH-Data Server) and both links can use virtually any kind of point-to-point wireless technology. We use both simulated and empirical results to demonstrate the effectiveness of Ripple-2; it proves to be ideal for low duty-cycle data collection applications due to its exceptional small network overhead (typically smaller than 1%) and its robustness to the size of the network.