IJUC HomeIssue Contents

Model Accuracy Assessment in Reaction-Diffusion Pattern Formation in Wireless Sensor Networks
Thomas C. Henderson, Anshul Joshi, Kirril Rashkeev, Narong Boonsirisumpun, Kyle Luthy and Edward Grant

We propose to exploit reaction-diffusion (RD) patterns as part of the wireless sensor network (S-Net) high-level structure building toolkit; e.g., to support leader selection or to provide pathways through the network. In particular, we study the formation of RD spot and stripe patterns in S-Nets for which no coordinate frame exists; i.e., the nets have only topological connectivity determined by the inter-node broadcast range. We further demonstrate how macro-features of the RD patterns can be used for Bayesian model accuracy assessment of the difference between a uniform grid layout of the nodes versus an irregular grid due to error in node placement.

Keywords: Bayesian computational sensor networks, reaction diffusion, wireless sensor networks, pattern synthesis, model accuracy, chemical morphogenesis, ermentrout stripe

Full Text (IP)