The SecureNet project tries to resolve the security issues lying in the networks such that no one gains extra information from the network. In general, our research targets at 1) protecting privacy of computation and 2) achieving verifiability of computation.
1) Privacy-Preserving ComputationIn many data mining applications in our real life, we need to jointly compute some statistic results (e.g., sample skewness, k-th moment, mean square weighted deviation, regression and randomness test) without disclosing our own data to others. In the WholeCom project, we designed security protocols to allow multiple parties to jointly compute an arbitrary multivariate polynomial (i.e., its input is all participants' data) while preserving their own input data secret to anyone else. Notably, we do not require secure communication channel as other existing works did.
As an extension of this result, we are working on privacy-preserving arbitrary computation on the data.
2) Verifiable ComputationIn this era of cloud computing, many people resort to this virtualization technique to achieve powerful computation ability that individuals are almost impossible to enjoy. In such a computing model, it is highly desirable that outsourced computation is verifiable because it is impossible for outsourcer to execute the computation himself to check if the result is correct due to the huge overhead of the computation. The target of this project is to prepare a scheme which can verify one's outsourced computation with an acceptable extra computation \& communication overhead.
These are general directions of our research projects. Our project also covers myriads of applications such as privacy-preserving data mining, secure advertisement in compliance with Do-Not-Track etc.