Study On The Application Of Digital Watermarking In Security Of Wireless Sensor Networks

Wireless sensor networks(WSNs) consisting of a large number of tiny sensor nodes can achieve data sensing, collecting, processing and transmitting in specific monitoring area through multi-hop and self-organizing networks. WSNs have broad application prospects in military, environmental monitoring, health care and some other fields, so its security problem must be paid more and more attention by researchers. There are various security issues involved in sensor networks. Among them, integrity and copyright protection of sensing data transmitted in unreliable networks are both security problems worthy of study.Sensor nodes are characterized by weak computing and storage capabilities and limited energy supply, and most energy consumption comes from wireless communication. That means traditional cryptography is not the best way to deal with the security problems in WSNs. Digital Watermarking is relatively simple, and brings no additional transmission overhead, which is wildly used in multimedia information security. And it also can be used to address some security issues of sensor networks with low cost. So, this study has some academic value in security of networks with limited resource and has some practical value in many applications of sensor networks.This dissertation firstly introduces the basic concept of WSNs, and proposes that the current researches on the security of sensor networks mainly focus on secure routing and data fusion, while data authentication still rely on traditional cryptology methods, and even less literature is related to the copyright protection issue. And the basic theoretical knowledge on cryptography, information hiding and digital watermarking involved in this dissertation are illustrated, and the data stream in WSNs is compared with multicast data stream in respect of network structure and security requirement. Then this dissertation reviews the existing authentication and copyright protection schemes based on digital watermarking technology in WSNs.According to the network characteristics and security requirements, combined with some existing digital image watermarking schemes, this dissertation mainly studies authentication schemes based on fragile watermarking and copyright protection schemes based on robust watermarking in sensor networks. It proposes(1) a watermarking authentication scheme validating individual data items which overcomes the shortcomings of grouping schemes,(2) a reversible watermarking authentication scheme which can restore original data, and(3) a 1-bit robust watermarking scheme which realizes copyright protection, and verifies the performance of these schemes through theoretical analysis and simulation experiments.The main contributes of this dissertation are:(1) Integrity and copyright protection in WSNs are studied on an abstract uniform data model. There are kinds of sensor nodes and network protocols in WSNs. In order to avoid the complexity due to various hardware and protocols, this dissertation implements authentication in the application layer. Each data item sampled by sensor nodes will be regarded as a basic data granularity regardless of the data processing in data link layer. So these proposed schemes have a wide range of adaptability, especially in restricted network environment existing in the applications of Internet of things. Secondly, data transmitted in WSNs will be considered as abstract numerical data stream from the sensor nodes to the sink node. Watermarks are embedded in the sensor nodes and verified in the sink node without considering the data fusion. The abstraction provides the convenience and simplicity for the research of digital watermarking in diverse sensor network environment.(2) Individual data items authentication schemes based on watermarking in WSNs are proposed. Current watermarking schemes in sensor networks usually authenticate stream at the level of data group. Some unchanged data items are misjudged in a tampered group and robustness of the data group is poor. This dissertation proposes an individual data items watermarking authentication scheme based on “double diffusion” and an improved version without grouping. Our schemes overcome the disadvantages of grouping schemes and achieve authentication of individual data items in the stream with low actual false alarm rate.(3) A reversible watermarking authentication scheme in WSNs is proposed. Traditional watermarking schemes, which modifies the original data, cannot meet the demand of some sensitive applications. According to the reversible image watermarking scheme based on prediction-error expansion, by utilizing the correlation of adjacent data in the stream, the watermark can be embedded through expanding the difference between the original value and the predicted value of the data item. And the sink can restore the data after authentication.(4) Copyright protection schemes based on 1-bit robust watermarking in WSNs are proposed. Based on patchwork algorithm, the proposed scheme selects some data items in the stream uniformly and pseudo-randomly, and divides the selected data into two groups. The 1-bit watermark can be embedded through changing the difference between the two groups, and also can be detected through some statistical methods. The proposed approaches show high robustness to the common attacks such as interception, injection, sampling and modification, and ensure that the watermark can not be erased without making the stream unusable.(5) A performance evaluation mechanism for watermarking schemes in WSNs is established. Through analysis and experiments, this dissertation establishes an evaluation system composed of the false negative rate, false alarm rate, robustness, cost and quality of service, and evaluates these proposed watermarking schemes through analysis and simulations.