Enhancing the security of electronic commerce transactions

This thesis looks at the security of electronic commerce transaction process- ing. It begins with an introduction to security terminology used in the thesis. Security requirements for card payments via the Internet are then described, as are possible protocols for electronic transaction processing. It appears that currently the Secure Socket Layer (SSL) protocol together with its standardised version Transport Layer Security (TLS) are the most widely used means to se- cure electronic transactions made over the Internet. Therefore, the analysis and discussions presented in the remainder of the thesis are based on the assumption that this protocol provides a `baseline' level of security, against which any novel means of security should be measured. The SSL and TLS protocols are analysed with respect to how well they satisfy the outlined security requirements. As SSL and TLS provide transport layer security, and some of the security requirements are at the application level, it is not surprising that they do not address all the identi¯ed security requirements. As a result, in this thesis, we propose four protocols that can be used to build upon the security features provided by SSL/TLS. The main goal is to design schemes that enhance the security of electronic transaction processing whilst imposing minimal overheads on the involved parties. In each case, a description of the new scheme is given, together with its advantages and limitations. In the ¯rst protocol, we propose a way to use an EMV card to improve the security of online transactions. The second protocol involves the use of the GSM subscriber authentication service to provide user authentication over the Internet. Thirdly, we propose the use of GSM data con¯dentiality service to protect sensitive information as well as to ensure user authentication. Regardless of the protection scheme employed for the transactions, there exist threats to all PCs used to conduct electronic commerce transactions. These residual threats are examined, and motivate the design of the fourth protocol, proposed speci¯cally to address cookie threats