A Research On Key Techniques Of In-Network Information Processing In Wireless Sensor Networks

ABSTRACT:As an important part of the Internet of Things(IoT),Wireless Sensor Networks(WSNs)have been widely researched in the literatures.In the field of WSNs,in-network information processing can greatly improve the quality of service(QoS),decrease the energy consumption and prolong the lifetime of the networks,and hence it is a hot research area.In this thesis,we research the approaches of target locating and tracking,secure data fusion,data query and source location-privacy protection.Correspondingly,four approaches are proposed which perform better than existing methods and they are more practical and reasonable.1)We design a new target locating and tracking scheme based on the theory of planar graph,and the coverage holes are also taken into consideration.In our scheme,two modes are designed based on the different regions in the network,i.e.,target tracking mode and round up mode.In target tracking mode,we improve the tracking accuracy by employing data fusion technique.In round up mode,we transmit the information by both left-hand and right-hand rules to improve the tracking continuity.By integrating the above two modes,we can always provide a good service to the data users.Simulation results illustrate that our scheme performs well in terms of tracking accuracy,continuity and energy-efficiency.2)We propose a secure data fusion framework for WSNs.In this framework,the nodes in the networks are organized by clusters and each cluster runs a reputation system independently to record the reputation value of each node.We design a cluster head selection scheme to decrease the probability of selecting the unreliable nodes as the cluster heads.In the cluster heads,only the credible data are fused and the results are sent to the sink node.By analyzing the fusion results,the sink node provides a feedback to the nodes in the clusters to accelerate the convergence procedure of reputation systems.We improve the security of data fusion by making the common nodes in the cluster,cluster heads and the sink nodes collaborate with each other.3)Considering that the filter-based top-k query schemes perform very badly if the readings of the nodes fluctuate drastically or the readings decrease in a period,in this thesis,we mine the spatial correlation of the data and design a direct-walk-based top-k data query scheme.In the network,a set of tokens walk independently and dynamically calculate the optimal walking directions based on the collected readings to find the local optimal points quickly.By collecting all the tokens,the sink node can finally get the top-k readings of the whole network.In addition,we propose a smart selection algorithm to choose different schemes for different data sets and hence we can make fulluse of the advantages of different schemes.Simulation results illustrate the effectiveness of our scheme.4)We propose a source-location privacy protection approach based on local cloud and anchor nodes to defend against the smart adversary.The source node constructs a local cloud based on secret sharing technique to hide itself and the behaviors of the nodes in the cloud are carefully designed to make the nodes indistinguishable.At the border of the cloud,the fake source nodes are responsible of delivering the secret shares to the sink node based on the designed random routing algorithm.When the sink node successfully receives a set of secret shares,it can reconstruct the original message.Simulation results demonstrate that the proposed scheme can strongly protect the source-location privacy and performs well in terms of data transmission amount and energy consumption.In addition,the employment of the secret sharing scheme improves the tolerance in terms of node failure.At last,we conclude this thesis.First,we analyze the drawbacks of the proposed schemes and discuss the methods to improve the schemes in the future.The development direction of in-network information processing techniques is also discussed.In addition,we summarize the challenges of out-network information processing including data storage,transmission,processing and security.The seamless combination of WSNs,cloud computing and edge computing is also discussed.