Spam has become a big problem for electronic mail. Computational proof for fighting spam is an innovative solution that makes senders "pay" for sending mail using computational effort rather than money. This works by attaching to the mail a short bitstring which proves that a certain amount of computational effort has been expended, just for that message, and just for that recipient. The recipient can accept only mail that has a valid proof attached. Researchers have developed the theory, but there was no way to integrate it into real e-mail systems.

Ali's project implements these ideas as a Mozilla Thunderbird extension, which enables users to compute such proofs using an existing e-mail infrastructure. When the extension is activated, it automatically performs the computation task and attaches the required proof to the outgoing message.

There is a huge number of collaborative tools on the web allowing users to share resources such as videos, documents and ideas. However, they are not very suitable for the enterprise environment, which requires more stringent security mechanisms in place to safeguard business content. For example, wikis don't have sufficiently fine-grained policy-driven protection of content.

Cherk's project implemented a prototype designed by colleagues at Hewlett Packard (Bristol) for a fine-grained access control for enterprise collaborative environments. His implementation allows the HP system to be investigated more thoroughly. This project was supervised by colleagues at HP as well as a supervisor in Birmingham.

Biometric authentication systems are available on modern laptops, but so far they can be used only locally (e.g., for authenticating users at login time). Remote biometric authentication poses additional problems, since an attacker has access to the communication channel (e.g., the Internet) and can forge messages or replay old messages. This project aimed to design a protocol for remote biometric authentication that resists such attacks.

This dissertation investigates different approaches to issues of the systems for enforcing access control. Such a mechanism is called stack inspection and is present in current Java JVM and Microsoft .NET CLR implementations. Its main purpose is to verify that only sufficiently authorized code may perform security-sensitive operations. In general, there are two main approaches.

The first one is based on certain modifications of stack-based access control systems, i.e. adding and maintaining a set of permissions at run-time. The othe rapproach is about static type systems that can enforce access control, statically verifying that code can access only those resources that are granted by an access control policy, before run-time. this dissertation presents details of each of those systems and discusses its properties, limitations and possible implementation problems.

In this project I have reviewed the JPEG compression algorithm and how it lends itself to effective steganography techniques. Once understood, I analysed the available steganalysis techniques that attempt to distinguish between normal images and those with information embedded.

In particular I have concentrated on developing an understanding of which images are most likely to produce better cover images. In choosing a cover image carefully, through statistical analysis a steganalyst is able to reduce the chances of any information being detected through steganalysis techniques.

This dissertation discusses defences against SQL command injection attacks with a focus to adapt a proposed defence which presents a sound and complete algorithm for dealing with injection attacks to another web technology.

The aim of this project was to extend this defence and to use the idea of parse tree validation in developing a simple Java tool; XPath Check to prevent XPath injection attacks.