多項式秘密計算に対するセキュリティ強度分析とその改善

抄録

This paper analyzes security strength of Iseki & Hayashi's secure computation which is valid for computation of polynomial as for real numbers, and proposes a new method to improve its security strength. First we derive the fact that Iseki & Hayashi's secure computation is deciphered if the value of single input variables of the computation is known. This implies that the security strength of Iseki & Hayashi's secure computation is bounded by the length of bits which can be inputted to this single variable. To improve this bound to be more secured, we propose a technique to repeat steps of Iseki & Hayashi's secure computation until its security strength becomes sufficiently powerful. A simple relationship about security strength between Iseki & Hayashi's secure computation and our proposal is found, and it shows security strength becomes quickly powerful with the increase of number of repetitions. Finally, numerical examples are demonstrated to show the effectiveness of our proposal.