Abstract:
With widespread use of databases in organizations and businesses, sensitive data in
a database could be easily accessed by malicious or unauthorized users. Hence,
security and privacy issues have become indispensable ingredients to the survival of
any organization. However, treating security issues as an afterthought often result in
inefficiency and non-secure design and implementation. In addition, high level
security is likely to conflicts with high efficiency.
Encryption over sensitive data could provide a way to protect databases from
unauthorized accesses including those from the database administrator and service
providers. It also puts some limitations on data access since appropriate keys must be
used for decrypting those data. But since most encryption techniques have high
overheads and are complicated, the encryption usually degrades database performance.
Generally, the cost of applying encryption to the database depends on data granularity,
the encryption algorithm itself, and the number of encryptions performed.
In this thesis, the trusted query model for encrypted database where is proposed
only sensitive attributes are encrypted and the role-based security policy is enforced
via the key manager who maintains the authority table and the key table for allowing
authorized accesses according to the security policy defined by the organization
executives. The experimental results illustrate that the encryption plays a significant
role in the query response time since the query response time of selective encrypted
attributes in this model is much lower than that of the whole encrypted database. Thus,
this model can be practically applied to real-world applications where few sensitive
attributes are focused, and the database performance is not greatly reduced.