Design And Implementation Of An Anomaly Detection And Diagnosis Method Based On Symptom Mining In Networked Embedded Systems

Embedded network systems,such as wireless sensor networks,are becoming widely used.But detecting and diagnosing anomalies in networked embedded systems is a very difficult task,due to the variable workloads and severe resource constraints.We propose a new method for post-deployment anomaly detection and diagnosis in networked embedded systems by combining program profiling and symptom mining.Firstly,this method employs binary instrumentation to perform lightweight function count profiling.Secondly,based on the statistics,it uses Principal Component Analysis based approach for automatically detecting network anomalies.Thirdly,it is able to point programmers closer to the most likely causes by a novel approach combining statistical tests and program call graph analysis.We implement our method based on TinyOS 2.1.1 and evaluate its effectiveness by four case studies in the development of a sensor network.Results show that our method can aid programmers to diagnose problems quickly in real-world sensor network systems.Compared with previous methods,this method provides the following features:post-deployment diagnosis ability,automatic detection,code-level diagnosis.The contributions of this work are summarized as follows.1.We propose a new method for post-deployment anomaly detection and diagnosis in networked embedded systems by combining program profiling and symptom mining.2.We propose a novel approach combining statistical tests and program call graph analysis to point programmers closer to the most likely causes inside the node.3.We implement our method and demonstrate its effectiveness using four case studies from real sensor network applications in an testbed consisting of 50 TelosB nodes.