Research On WSN Streaming Computing Technology Based On Compression Network Coding

In widely deployed wireless sensor networks(WSNs),nodes generate massive real-time environmental feature data streams.How to efficiently discover anomaly data from these large-scale data streams is an important research topic in streaming computing.In this thesis,a WSN streaming computing system built on Spark Stream is proposed to detect anomaly environmental data;In order to optimize the transmission and computation efficiency of data streams,the compression sensor network coding technology is introduced into system used in big data scenarios.The main contributions of this thesis are as follows:Firstly,the different terminal nodes in WSN are designed and realized to detect the anomaly environmental data,including physical structures and network topology of nodes.Basic data transmission models and protocols between nodes are analyzed too.The network topology among these nodes is planned as a ring-shaped structure;Cluster heads collect data in each annulus and upward transmission.The formation of generalized butterfly network is used to obtain compression network coding gain.Secondly,a streaming computing platform is constructed to quickly discover anomaly data.The platform receives the original environmental data streams gathered from WSNs through the data cloud gateway.Then,it pushes the synchronized data records to streaming k-means program on the Spark Stream cluster to quickly discover clusters of anomalies in a large scale data streams.Then,the k-means clustering update method is improved.This thesis proposes an optimized k-means algorithm by updating security interval labels on Spark to reduce the computation time in a single batch data stream and make the system adapt to the abnormal data discovery in big data.Finally,compressed network coding and decoding reconstruction are introduced in the transmission and processing phases of the system.In transmission stages,the transmission capacity is compressed to improve the transmission efficiency of the system.In processing stages,Spark is used to realize the reconstructing calculation,so that the performance of the large data frame is fully utilized.In this thesis,a real-time anomaly data discovery system for WSNs is developed.The system is optimized in three aspects:efficient transmission,fast processing and reliable guarantee.