3D Content Encryption


Since the 1970s, a large number of encryption schemes have been proposed, among which some have been standardised and widely adopted all over the world, such as Data Encryption Standard (DES) and Advanced Encryption Standard (AES). However, due to special features of 3D content, these encryption standards may not be a suitable solution for 3D applications. The problem of 3D content encryption is beyond the application of established and well-known encryption algorithms. This is primarily due to the structure of 3D content and the way it is used commercially. Unlike data encryption, where a complete bitstream is encrypted, 3D content encryption introduces several challenges. One of the greatest challenges of 3D content encryption is that, in comparison with traditional data and 2D images, 3D content implies a higher level representation or semantics, and in many 3D applications, it is necessary to maintain 3D semantics, such as the spatial and dimensional requirements. In this project, we conduct research as to address such requirements.





Anomaly Detection in Wireless Sensor Networks


The recent advancements in technology have enhanced the use of Wireless Sensor Network (WSN) in several application domains such as environment, home, industry, military, health, and infrastructure monitoring. In these application domains, sensor nodes and sensor readings are vulnerable to various attacks and anomalies. These anomalies may degrade overall performance of a Wireless Sensor Network (WSN) application. Mobile Agents (MA) are being effectively employed in WSNs for variety of purposes such as distributed data fusion, code and data dissemination, localization, security, collaborative signal, and information processing. In this research, MAs are employed for in-situ verification of suspicious behaviour of a sensor node as part of an anomaly detection system. This approach is particularly useful for large scale WSNs where physical verification of anomalous node is cumbersome and time consuming. Furthermore, this technique is also beneficial in infrastructure monitoring WSN applications where mobile agent can non-intrusively verify the anomalous behaviour of sensor nodes before taking an appropriate action against them.





Body Sensor Networks


The availability of small, low-cost networked sensors combined with advanced signal processing and information extraction is driving a revolution in physiological monitoring and intervention. Body Sensor Networks (BSN) are enabling technologies for precision healthcare, enhanced sports and fitness training, novel life-style monitoring, and individualized security. Expected growth of elderly populations and the corresponding increase in healthcare costs mandate systems for automated monitoring of physiological conditions, triage, and remote diagnosis.





Semantics-Based Document Classification for Data Leakage Detection



The protection of confidential data from being leaked to the public is conventionally done through firewalls, virtual private networks and intrusion detection/prevention systems. However, these systems lack dedicated and proactive protection for confidential data when it travels through legitimate channels. An emerging technology in the field of information security called Data Leakage Prevention has been developed to overcome these problems. It deals with tools working under a central policy, which analyse networked environments to detect sensitive data, prevent unauthorized access to it and block channels associated with potential leak. This requires special data classification capabilities to distinguish between sensitive and normal data. Not only this task needs prior knowledge of the sensitive data, but also requires knowledge of potentially evolved and unknown data. Most current DLPs use content-based analysis in order to detect sensitive data. This mainly involves the use of regular expressions, data fingerprinting and statistical analysis. Regular expressions and data fingerprinting are widely used as robust detection techniques when the confidential data is known and unmodified. However, they usually become ineffective if the confidential data is ambiguous or largely modified. Many forms of advanced regular expression and data fingerprinting have been studied in the literature, but they suffer from being inadequate when data is extremely modified. Also, they require huge amount of indexing and computations, which introduce significant system overheads.   Content statistical analysis on the other hand deals with nebulous forms of data, while preserving exact data detection capabilities. It focuses on analysing frequencies and relationships of words “terms” within a large corpus, to construct a semantic weight for data. It also facilitates the use of machine learning algorithms and Bayesian analysis, in order to classify data.  Although proven effective in other fields like information retrieval (IR), statistical analysis is not fully explored when it comes to detecting confidential data leakage. Therefore, in our study we evaluate the effectiveness of using content statistical analysis in constructing semantics of confidential data. Moreover, we propose a semantic-aware document classification model for data leakage detection. This model utilizes statistical analysis of term weighting, to semantically compare documents with existing category centroids.