Abstract Wireless sensor networks (WSNs) offer us a potential for greater awareness of our surroundings, collecting, measuring, and aggregating parameters beyond our current abilities, and provide an opportunity to enrich our experience through context-awareness. As a typical sensor node is small with limited processing power, memory, and energy resources, in particular, these WSNs must be very energy-efficient for practical deployment. Medium access control (MAC) protocols are central to the energy-efficiency objective of WSNs, as they directly control the most energy consuming part of a sensor node: communications over the shared medium. This thesis focuses on evaluating MAC protocols within the WSN domain by, firstly, surveying a representative number of MAC protocols and their features. Secondly, three novel MAC protocols are proposed, one for layered contention-based access, one for layered scheduled access, and one for cross-layer contention-based access. Thirdly, a novel energy consumption model is proposed, and fourthly, a holistic MAC protocol evaluation model is proposed that takes into account application emphasis on performance metrics. The MAC protocols are evaluated analytically. In addition, the layered contention-based MAC protocol has been implemented and measured, and the cross-layer contention-based protocol operating over an impulse radio-ultra wideband (IR-UWB) physical layer has been verified by simulations with relevant physical layer characteristics. The energy consumption evaluation model proposed is straightforward to modify for evaluating delay, and it can reuse state transition probabilities derived from throughput analysis. The holistic application-driven MAC protocol evaluation model uses a novel single compound metric that represents a MAC protocol's relative performance in a given application scenario. The evaluations have revealed several significant flaws in sensor MAC protocols that are adapted to sensor networking from ad hoc networks. Furthermore, it has been shown that, when taking sufficient details into account, single hop communications can outperform multi-hop communications in the energy perspective within the feasible transmission ranges provided by sensor nodes. The impulse radio physical layer introduces characteristics to MAC protocols that invalidate traditional techniques which model the physical layer in terms of simple collisions. Hence, these physical layer characteristics have been modelled and included in the analysis, which improves the level of agreements with simulated results.

The increasing reliance on information technology to support business processes has emphasised the need for information security mechanisms. This, however, has resulted in an ever-increasing workload in terms of security administration. Security administration encompasses the activity of ensuring the correct enforcement of access control within an organisation. Access rights and their allocation are dictated by the security policies within an organisation. As such, security administration can be seen as a policybased approach. Policy-based approaches promise to lighten the workload of security administrators. Separation of duties is one of the principles cited as a criterion when setting up these policy-based mechanisms. Different types of separation of duty policies exist. They can be categorised into policies that can be enforced at administration time, viz. static separation of duty requirements and policies that can be enforced only at execution time, viz. dynamic separation of duty requirements. This dissertation deals with the specification of both static separation of duty requirements and dynamic separation of duty requirements in role-based workflow environments. It proposes a model for the specification of separation of duty requirements, the expressions of which are based on set theory. The model focuses, furthermore, on the enforcement of static separation of duty. The enforcement of static separation of duty requirements is modelled in terms of invariant conditions. The invariant conditions specify restrictions upon the elements allowed in the sets representing access control requirements. The sets are themselves expressed as database tables within a relational database management system. Algorithms that stipulate how to verify the additions or deletions of elements within these sets can then be performed within the database management system. A prototype was developed in order to demonstrate the concepts of this model. This prototype helps demonstrate how the proposed model could function and flaunts its effectiveness.

Security policy specification languages are a response to today's complex and vulnerable software climate. These languages allow an individual or organization to restrict and modify the behavior of third-party applications such that they adhere to the rules specified in the policy. As software grows in complexity, so do the security policies that govern them. Existing policy specification languages have not adapted to the growing complexity of the software they govern and as a result do not scale well, often resulting in code that is overly complex or unreadable. Writing small, isolated policies as separate modules and combining them is known as policy composition, and is an area in which existing policy specification languages have a number of drawbacks. Policy composition is unpredictable and nonstandard with existing languages. PoCo is a new policy specification language that uses signed regular expressions to return sets of allowed and denied actions as output from its policies, allowing policies to be combined with standard set operations in an algebraic way. This thesis covers my contribution to the PoCo project in creating a formal grammar for the language, developing a static analysis tool for policy designers, and implementation of the first PoCo language compiler and runtime for the Java platform.

M.Com. (Computer Auditing) / The need to evaluate today's complex computer environments from an audit perspective has increased, particularly in viewof the disappearance of a paper audit trail and the inefficiencies of auditing "around" the computer in these environments. By making use of the Access Path and the Path Context Models, it was possible to carry out an evaluation of the MVS/XA environment. This evaluation was carried out using the methodology developed in this research essay. This methodology may have universal applicability in the evaluation of computer security. The concept of each layer in the access path being a "net", which only allows authorised users to drop to the next layer, was applied. It was found that each systems software component had sufficient facilities to meet the control objectives. The operating system itself, however, was found to present the installation with more risk factors than controls. It was therefore concluded that an external access control software package needs to be implemented to supplement the controls in this environment, if the control objectives are to be met. It was also concluded that the implementation of this package would not, in itself, solve all the security issues, and that the matrices developed should be used in the implementation of this package. This is a further indication of the usefulness of the model and the methodology. The applicability of the Access Path and the Path Context Models in the evaluation of the predefined environment has therefore been established.

The individual of today incessantly insists on more protection of his/her personal privacy than a few years ago. During the last few years, rapid technological advances, especially in the field of information technology, directed most attention and energy to the privacy protection of the Internet user. Research was done and is still being done covering a vast area to protect the privacy of transactions performed on the Internet. However, it was established that almost no research has been done on the protection of the privacy of personal data that are stored in tables of a relational database. Until now the individual had no say in the way his/her personal data might have been used, indicating who may access the data or who may not. The individual also had no way to indicate the level of sensitivity with regard to the use of his/her personal data or exactly what he/she consented to. Therefore, the primary aim of this study was to develop a model to protect the personal privacy of the individual in relational databases in such a way that the individual will be able to specify how sensitive he/she regards the privacy of his/her data. This aim culminated in the development of the Hierarchical Privacy-Sensitive Filtering (HPSF) model. A secondary aim was to test the model by implementing the model into query languages and as such to determine the potential of query languages to support the implementation of the HPSF model. Oracle SQL served as an example for text or command-based query languages, while Oracle SQL*Forms served as an example of a graphical user interface. Eventually, the study showed that SQL could support implementation of the model only partially, but that SQL*Forms was able to support implementation of the model completely. An overview of the research approach employed to realise the objectives of the study: At first, the concepts of privacy were studied to narrow down the field of study to personal privacy and the definition thereof. Problems that relate to the violation or abuse of the individual’s personal privacy were researched. Secondly, the right to privacy was researched on a national and international level. Based on the guidelines set by organisations like the Organisation for Economic Co-operation and Development (OECD) and the Council of Europe (COE), requirements were determined to protect the personal privacy of the individual. Thirdly, existing privacy protection mechanisms like privacy administration, self-regulation, and automated regulation were studied to see what mechanisms are currently available and how they function in the protection of privacy. Probably the most sensitive data about an individual is his/her medical data. Therefore, to conclude the literature study, the privacy of electronic medical records and the mechanisms proposed to protect the personal privacy of patients were investigated. The protection of the personal privacy of patients seemed to serve as the best example to use in the development of a privacy model. Eventually, the Hierarchical Privacy-Sensitive Filtering model was developed and introduced, and the potential of Oracle SQL and Oracle SQL*Forms to implement the model was investigated. The conclusion at the end of the dissertation summarises the study and suggests further research topics. / Prof. M.S. Olivier

Businesses are continually striving to become more efficient. In an effort to achieve optimal efficiency, many companies have been forced to re-evaluate the efficiency of their business processes. Consequently, the term “business process re-engineering” (BPR) has been given to the activity of restructuring organizational policies and methods for conducting business. The refinement of business processes is the primary motivation behind the development of automated work- flow systems that ensure the secure and efficient flow of information between activities and participants that constitute the business process. A workflow is an automated business process that comprises a number of related tasks. When these tasks are executed in a systematic way, they contribute to the fulfilment of some goal. The order in which workflow tasks execute is of great significance because these tasks are typically dependent on each other. A workflow management system (WFMS) is responsible for scheduling the systematic execution of workflow tasks whilst considering the dependencies that exist between them. Businesses are realizing the necessity of information management in the functioning and general management of a company. They are recognizing the important role that information security has to play in ensuring that accurate information that is relevant is gathered, applied and maintained to enhance the company’s service to its customers. In a workflow context, information security primarily involves the implementation of access control security mechanisms. These mechanisms help ensure that task dependencies are coordinated and that tasks are performed by authorized subjects only. In doing so, they also assist in the maintenance of object integrity. TheWorkflow Authorization Model (WAM) was developed by Atluri and Huang [AH96b, HA99] with the specific intention of addressing the security requirements of workflow environments. It primarily addresses the granting and revoking of authorizations in a WFMS. TheWAM satisfies most criteria that are required of an optimal access control model. These criteria are the enforcement of separation of duties, the handling of temporal constraints, a role-based application and the synchronization of workflow with authorization flow. Some of these conditions cannot be met through pure role-based access control (RBAC) mechanisms. This dissertation addresses the delegation of task authorizations within a work- flow process by subject roles in the organizational structure. In doing this, a role may have the authority to delegate responsibility for task execution to another individual in a role set. This individual may potentially belong to a role other than the role explicitly authorized to perform the task in question. The proposed model will work within the constraints that are enforced by the WAM. Therefore, the WAM will play a part in determining whether delegation may be approved. This implies that the delegation model may not override any dynamically defined security constraints. The Delegation Authorization Model (DAM) proposed assists in distributing workloads amongst subject roles within an organization, by allowing subjects to delegate task responsibilities to other subjects according to restrictions imposed by security policies. As yet, this area of research has not received much attention. / Prof. M.S. Olivier

D.Phil. (Computer Science) / The need for security in a database is obvious. Object-orientation enables databases to be used in applications where other database models are not adequate. It is thus clear that security of object-oriented databases must be investigated...

Workflow systems have gained popularity not only as a research topic, but also as a key component of Enterprize Resource Planning packages and e- business. Comprehensive workflow products that automate intra- as well inter-organizational information flow are now available for commercial use. Standardization efforts have centered mostly around the interoperability of these systems, however a standard access control model have yet to be adopted. The research community has developed several models for access control to be included as part of workflow functionality. Commercial systems, however, are still implementing access control functionality in a proprietary manner. This dissertation investigates whether a comprehensive model for gain- ing context-sensitive access control, namely CoSAWoE, can be purposefully implemented in a commercial workflow product. Using methods such as an exploratory prototype, various aspects of the model was implemented to gain an understanding of the di±culties developers face when attempting to map the model to existing proprietary software. Oracle Workflow was chosen as an example of a commercial workflow product. An investigtion of the features of this product, together with the prototype, revealed the ability to affect access control in a similar manner to the model: by specifying access control constraints during administration and design, and then enforcing those constraints dynamically during run-time. However, only certain components within these two aspects of the model directly effected the commercial workflow product. It was argued that the first two requirements of context-sensitive access control, order of events and strict least privilege, addressed by the object design, role engineering and session control components of the model, can be simulated if such capabilities are not pertinently available as part of the product. As such, guidelines were provided for how this can be achieved in Oracle Workflow. However, most of the implementation effort focussed on the last requirement of context-sensitive access control, namely separation of duties. The CoSAWoE model proposes SoD administration steps that includes expressing various business rules through a set of conflicting entities which are maintained outside the scope of the workflow system. This component was implemented easily enough through tables which were created with a relational database. Evaluating these conflicts during run-time to control worklist generation proved more di±cult. First, a thorough understanding of the way in which workflow history is maintained was necessary. A re-usable function was developed to prune user lists according to user involvement in previous tasks in the workflow and the conflicts specified for those users and tasks. However, due to the lack of a central access control service, this re- usable function must be included in the appropriate places in the workflow process model. Furthermore, the dissertation utilized a practical example to develop a prototype. This prototype served a dual purpose: firstly, to aid the author's understanding of the features and principles involved, and secondly, to illustrate and explore the implementation of the model as described in the previous paragraphs. In conclusion the dissertation summarized the CoSAWoE model's compo- nents which were found to be product agnostic, directly or indirectly imple- mentable, or not implemented in the chosen workflow product. The lessons learnt and issues surrounding the implementation effort were also discussed before further research in terms of XML documents as data containers for the workfow process were suggested.

The goal of this thesis is to find provably correct methods for detecting conflicts between XACML rules. A conflict occurs when one rule permits a request and another denies that same request. As XACML deals with access control, we can help prevent unwanted access by verifying that it contains rules that do not have unintended conflicts. In order to help with this, we propose an algorithm to find these conflicts then use the Coq Proof Assistant to prove correctness of this algorithm. The algorithm takes a rule set specified in XACML and returns a list of pairs of indices denoting which rules conflict. It is then up to the policy writer to see if the conflicts are intended, or if they need modifying. Since we will prove that this algorithm is sound and complete, we can be assured that the list we obtain is complete and only contains true conflicts.

Ubiquitous computing (UbiComp) envisions a new computing environment, where computing devices and related technology are widespread (i.e. everywhere) and services are provided at anytime. The technology is embedded discreetly in the environment to raise users' awareness. UbiComp environments support the proliferation of heterogeneous devices such as embedded computing devices, personal digital assistants (PDAs), wearable computers, mobile phones, laptops, office desktops (PCs), and hardware sensors. These devices may be interconnected by common networks (e.g. wired, wireless), and may have different levels of capabilities (i.e. computational power, storage, power consumption, etc). They are seamlessly integrated and interoperated to provide smart services (i.e. adaptive services). A UbiComp environment provides smart services to users based on the users' and/or system's current contexts. It provides the services to users unobtrusively and in turn the user's interactions with the environment should be as non-intrusive and as transparent as possible. Access to such smart services and devices must be controlled by an effective access control system that adapts its decisions based on the changes in the surrounding contextual information. This thesis aims at designing an adaptive fine-grained access control solution that seamlessly fits into UbiComp environments. The solution should be flexible in supporting the use of different contextual information and efficient, in terms of access delays, in controlling access to resources with divergent levels of sensitivity. The main contribution of this thesis is the proposal of the Context-Risk-Aware Access Control (CRAAC) model. CRAAC achieves fine-grained access control based upon the risk level in the underlying access environment and/or the sensitivity level of the requested resource object. CRAAC makes new contributions to the access control field, those include 1) introducing the concept of level of assurance based access control, 2) providing a method to convert the contextual attributes values into the corresponding level of assurance, 3) Proposing two methods to aggregate the set of level of assurance into one requester level of assurance, 4) supporting four modes of working each suits a different application context and/or access control requirements, 5) a comprehensive access control architecture that supports the CRAAC four modes of working, and 6) an evaluation of the CRAAC performance at runtime.