Private Range Query by Perturbation and Matrix Based Encryption
スポンサーリンク
概要
- 論文の詳細を見る
2011 Sixth International Conference on Digital Information Management (ICDIM) : Melbourne, Australia, September, 2011.09.26-2011.09.28In this paper, we propose a novel approach for private query; IPP (inner product predicate) method. Private query is a query processing protocol to obtain requesting tuples without exposing any information about what users request to third persons including service providers. Existing works about private query such as PIR, which ensure information theoretic safety, have severe restriction because they do not support range queries nor allow tuples having a same value in queried attributes. Our IPP method, on the other hands, focuses range queries mainly and it allows tuples having a same value in any attributes.IPP method employs a query transform by trusted clients (QT) scheme and proposes transformation algorithms which make thecorrelation between plain queries and transformed queries and the correlation between plain attribute values and transformed attribute values small enough. Thus, the transformed queries and attribute values have resistance to frequency analysis attacks which implies IPP method prevents attackers, who know the plain distribution of them, from computing the plain queriesand attribute values from transformed values. IPP method adds perturbations to queries and attribute values and gives them amatrix based encryption to achieve the above property. We also confirm the computational cost on servers belongs to O(n) with the number of tuples n and is virtually no orrelation between the distributions of transformed queries and queried attribute values and the plain distributions of them by experimental evaluations.
論文 | ランダム
- ノイズを含むデータからの学習と連想記憶
- ノイズを含むデータからの学習と連想記憶
- 連想記憶のためのボルツマンマシンの学習の終了条件について
- ボルツマンマシンの学習を用いた連想記憶モデルにおける引き込み領域の大きさについて
- 学習経路と情報量の位相的性質との関係について