Digital watermarking of non-media data stream (applications)

This dissertation is principally concerned with investigation of digital watermarking methods suitable for non-media data types and manifesting the new trend of applications for such watermarking methods. Digital watermarking is a technique for embedding secret information within a digital content in a manner that does not interfere with normal usage of that content. It was first used as an effective method for attesting to the originality of digital contents and to discourage unauthorized copying.<br><br>A large corpus of watermarking techniques has arisen for digital images, audio and video data, but the challenges and issues of non-media data watermarking have not been fully inves- tigated. One of the contributions of this dissertation is in proposing a new taxonomy model for digital watermarking methods to cater for more complex data types such as time-series, graph-structured data, data streams, and so forth. The rationale for this taxonomy is that the ultimate goal for many of today’s data collection is gaining knowledge by means of data mining strategies. Therefore, the data watermarking process should not interfere with the data mining process and needs to be kept as transparent as possible. This implies that the type of data will not only affect the data mining methods but also data watermarking technique.<br><br>In alignment with the taxonomy to be presented, this dissertation demonstrates several novel watermarking methods for three types of data namely biomedical data, sensor stream- ing, and spatiotemporal data. Particularly in respect of sensor streaming data, two digital watermarking methods are presented that are resilient against multiple data aggregation as well as de-synchronisation. The former feature is particularly useful for secure in-network data aggregation in sensor networks, where multiple signatures from different sensors may need to be verified quickly. The latter feature is useful to discern the de-synchronisation problem between watermark encoder and decoder, for which hiding synchronisation marks serves as a solution. Finally, this dissertation examines how digital watermarking methods can be deployed as an effective lightweight security tool for Internet of Things applications.