Seasonal wireless sensor network link performance in boreal forest phenology monitoring

Abstract

While indoor wireless sensor network (WSN) research has recently flourished for monitoring civil and industrial infrastructure, considerably less attention has been given to the development of reliable outdoor WSNs capable of long-term operation in challenging remote locations. We present wireless sensor network link performance results from the first year of monitoring micro-meteorological conditions alongside the 802.15.4 link received signal strength indicator (RSSI) within an old growth stand of deciduous boreal Aspen forest (Populus tremuloides) in Northern Alberta, Canada. Thirty-six weather proof nodes were equipped with meteorological sensors and distributed across one hectare in the forest understory to assess the application of WSNs for observing high resolution changes in seasonal ecosystem productivity and forest phenology. We describe here the density distribution of node RSSI using Gaussian kernel density estimates in relation to node antenna- receiver orientation and vegetation seasonality. RSSI across the network displays a lognormal distribution with an increasing bimodal tendency with path length through the forest stand. Spatial variability in RSSI is discussed with respect to forest structure. A strong temporal relationship between RSSI variability and plant canopy development is observed with a 20dBm or 100 fold difference in mean network radio signal power from spring leaf presence to fall leaf absence. The meteorological and biophysical factors associated with this trend are explored using multiple regression and relative factor importance analysis. Our results indicate that in addition to meteorological data, spectral vegetation density metrics are useful in assisting deployment planning and network