Security in Pervasive Computing Environment

Abstract

This thesis proposes a novel security model for trust based access control for an entelp se by using statistical process control approach. The proposed model provides secuity via authenticating the users / services, deflning authorization rules and policies for available resources' confidentiality and message integrity. The proposed model provides privacy framework for instant knowledge system to avoid the risk of unauthorized use ofthe usel's personally identifiable iDformation by analyzing the global privacy regulations, specifically made in UK and EU Detailed threat model using STRIDE approach is developed for personal distributed environment Based on the threat model' misuse cases are created to cater for the ways the system could be attacked' The proposed model incoryorates the use ofdigital certificates; synmetdc and asymmetric ciphers; trusted third parly for providing authentication, authorization and attribute assertions by using secu ty assertion markup language. Statistical process cofltlol monito ng concepts are applied to monitor and adjust users' trust values between upper control limit and lower contol limit of a paticular authorized trust category. Shewhart and exponentially weighted moving avemge control charts have been used to analyze the performance on simulated data. Simulation results have been obtained by extracting the sample trust values ofindividual users, group ofusers in a pafiicular hust category and sample average values acloss different trust categories. The proposed model once compared with existing solutions proves that the model is highly adaptable, provides dynamic access to resources and enhances security. 'fhe critical analysis of results shows that by adjusting trust values based oil mean and standard deviation, 9570 of users' average trust values lie within waming limits and 99.8olo ofusers' average fust values lie within action limits.