Wireless Sensor Network Autonomic Agent Based on In-Cluster Directional Evidence Reasoning

The sensor nodes performance evaluation and uncertainty reasoning are vital to dynamic, fault tolerance wireless sensor network management. An efficient directional evidence reasoning algorithm is put forward based on in-cluster directional hierarchical architecture, which is adapted to capture epistemic uncertainty in sensor-network. Accounting for NP complete of combination of Dempster Shafer theory (DST) for central algorithm, one distribute in-cluster algorithm is applied with fast Mobius transform (FMT) based on coarsening combination of evidences. This method not only offers an efficient framework for uncertainty reasoning with low computation cost, but also balances energy budget between ubiquitous computation and communication. Simulation verified it efficiency for self-management, and lifetime of wireless sensor network is prolonged obviously