A Comparative Study on Microaggregation Techniques for Microdata Protection

Microaggregation is an efficient Statistical Disclosure Control (SDC) perturbative technique for microdata protection. It is a unified approach and naturally satisfies k-Anonymity without generalization or suppression of data. Various microaggregation techniques: fixed-size and data-oriented for univariate and multivariate data exists in the literature. These methods have been evaluated using the standard measures: Disclosure Risk (DR) and Information Loss (IL). Every time a new microaggregation technique was proposed, a better trade-off between risk of disclosing data and data utility was achieved. Though there exists an optimal univariate microaggregation method but unfortunately an optimal multivariate microaggregation method is an NP hard problem. Consequently, several heuristics have been proposed but no such method outperforms the other in all the possible criteria. In this paper we have performed a study of the various microaggregation techniques so that we get a detailed insight on how to design an efficient microaggregation method which satisfies all the criteria.