Data Flow Watermarking In Wireless Sensor Networks

With the continuous development of wireless network communication technology,wireless sensor networks(WSNs)have been widely used in the Internet of Things and environmental detection due to their low power consumption and characteristics of being distributed and self-organizing characteristics.Wireless sensor networks are usually deployed in complex open environments and use simple micro-sensors,resulting in problems such as vulnerability and difficulty to guarantee network security performance.In particular,the historical features of data such as the creator of the data,ownership,etc.are recorded in the data source,so implementing data traceability is one of the important methods for protecting sensor data.Although encoding source information by data hiding technology can effectively protect data integrity in WSN,existing methods still face challenges in terms of secure data transmission and storage overhead.Therefore,this thesis focuses on the research on data stream watermarking in wireless sensor networks.The specific research contents are as follows.This thesis first proposes a Memory Incentive Provenance(MIP)method for data source management.The MIP method models the trailed data source as a real-time provenance framework based on the memory consumption within the specified data payload of a sensor data packet.Also,the consumed memory state is labeled as the data provenance,which demonstrates a hierarchical relation between the data fields of each packet and the packets of each data stream.Additionally,the MIP method utilizes a logistic regression-based source classification approach to identify the data provenance at the data receiver end to confirm whether the data originates from the valid source.Finally,experiments are conducted to demonstrate the effectiveness of the proposed MIP method.In addition,this thesis proposes a data integrity protection method based on adaptive watermark embedding,namely the Adaptive Embedding Algorithm(AEA).AEA examines the required bits of data by comparing the length of its primitive data type and detects nonoccupied bits to embed watermarks with variable lengths in a distributed approach,and on the basis of ensuring data security,it verifies the authenticity and integrity of the data.Additionally,the AEA shows its robustness against the threats of the payload segments because freed space eventually increases the embedding capacity.The experimental results based on the data payload utilization rate and embedding ability show the adaptability,scalability,and robustness of AEA.In summary,to target data stream security in WSN,this thesis proposes a memory incentive source method for data source management and a data integrity protection method using adaptive watermark embedding.Experimental results show that the two methods ensure the data security of wireless sensor networks without the interference of data accuracy,storage overhead,and transmission efficiency.They are of practical value and have promising prospects.