Citation Context

...1, and we also prove the security properties of our solutions under the adversary model. The study of “search on encrypted data” is closely related to our work. Specifically, Song, Wagner, and Perrig =-=[28]-=- propose practical techniques for finding keywords in encrypted files, which allow a user, when given a trapdoor for a keyword, to check the existence of the key word in a file. But their solution nee...

Citation Context

...y Chang and Mitzenmacher [9] has interesting analysis but their solution is restricted to searching for a keyword chosen from a pre-determined set. Boneh et al. present a searchable public key scheme =-=[6]-=-; the scenario they considered is analogous to that of [28] but uses public-key encryption rather than symmetric-key encryption. In the same scenario, Goh demonstrates a method for secure indexes usin...

Citation Context

...nsitive data, are very useful in preventing unauthorized users from accessing sensitive data [3]. With data stored in an encrypted form, a crucial question is how to perform queries. Hacigumus et al. =-=[19]-=- studied querying encrypted data in the database-as-service (DAS) model where sensitive data is outsourced to an untrusted server [20]. Their solution divides attribute domains into partitions and map...

Citation Context

...om a look-up dictionary by using hash functions. The basic idea is that a tuple can only be retrieved if a valid key is provided. 3sIn contrast to the goals of our work, private information retrieval =-=[10,8, 26]-=- is designed to hide entirely from the database which queries a user is making. As we discuss later, we take a more pragmatic view that allows more efficient solutions. 1.2 Our Contributions In this p...

Citation Context

...om a look-up dictionary by using hash functions. The basic idea is that a tuple can only be retrieved if a valid key is provided. 3sIn contrast to the goals of our work, private information retrieval =-=[10,8, 26]-=- is designed to hide entirely from the database which queries a user is making. As we discuss later, we take a more pragmatic view that allows more efficient solutions. 1.2 Our Contributions In this p...

Citation Context

... privacy policies for different data records, Hippocratic databases, which combine privacy policies with sensitive data, are very useful in preventing unauthorized users from accessing sensitive data =-=[3]-=-. With data stored in an encrypted form, a crucial question is how to perform queries. Hacigumus et al. [19] studied querying encrypted data in the database-as-service (DAS) model where sensitive data...

Citation Context

...m, a crucial question is how to perform queries. Hacigumus et al. [19] studied querying encrypted data in the database-as-service (DAS) model where sensitive data is outsourced to an untrusted server =-=[20]-=-. Their solution divides attribute domains into partitions and maps partitions ids to random numbers to achieve privacy. This idea is simple, practical, and elegant. However, it relies on an implicit ...

Citation Context

... check the existence of the key word in a file. But their solution needs to scan the entire file sequentially and no provably secure index technique is provided. A follow-up by Chang and Mitzenmacher =-=[9]-=- has interesting analysis but their solution is restricted to searching for a keyword chosen from a pre-determined set. Boneh et al. present a searchable public key scheme [6]; the scenario they consi...

Citation Context

...in the results of queries) to the user. If the partitions are smaller, then the database server needs to send fewer false positives, but more information is leaked. (This issue is further explored in =-=[23]-=-.) Furthermore, no precise quantifications are given of either of the information leak relative to the 2ssize of partitions or of the amount of communication overhead. In comparison, the solutions we ...

Citation Context

...hey considered is analogous to that of [28] but uses public-key encryption rather than symmetric-key encryption. In the same scenario, Goh demonstrates a method for secure indexes using Bloom filters =-=[15]-=-. These solutions are possibly useful in searching for keywords in a file; however, it is unclear how to apply them to the problem of efficiently querying encrypted relational databases. Yet another p...

Citation Context

...on of the definition of secure protocol in the semi-honest model (i.e., assuming the intruder does not modify the database software but attempts to violate data privacy by analyzing what he observes) =-=[17]-=-. However, note that a secure one-round query protocol as defined here remains secure even in the case the intruder is fully malicious (i.e., even when the intruder modifies the database software such...

Citation Context

...of attributes is large, this is a heavy burden for the user. To eliminate this problem, we use a pseudorandom permutation, which is by definition indistinguishable from a uniformly random permutation =-=[16]-=-. The advantage of this approach is that it requires the user to memorize the random seed. In fact, we note that we do not need to permute all the attributes in the 〈1〉 as defined in the basic solutio...

Citation Context

...r. Such methods are very useful in strengthening the protection of sensitive data in databases. 1.1 Related Work Various methods have been proposed recently for securing databases in various settings =-=[7,27,30,22,25,21]-=-. In particular, encryption is an important technique to protect sensitive data [13]. An analysis of how to encrypt and securely store data in relational database management systems has been given in ...

Citation Context

...om a look-up dictionary by using hash functions. The basic idea is that a tuple can only be retrieved if a valid key is provided. 3sIn contrast to the goals of our work, private information retrieval =-=[10,8, 26]-=- is designed to hide entirely from the database which queries a user is making. As we discuss later, we take a more pragmatic view that allows more efficient solutions. 1.2 Our Contributions In this p...

Citation Context

...Foundation under Grant No. CCR0331584. In Proceedings of the 11th European Symposium On Research In Computer Security (Esorics), 2006.sEncryption is a well-studied technique to protect sensitive data =-=[13]-=- so that even if a database is compromised by an intruder, data remains protected even in the event that a database is successfully attacked or stolen. Provided that the encryption is done properly an...

Citation Context

...r. Such methods are very useful in strengthening the protection of sensitive data in databases. 1.1 Related Work Various methods have been proposed recently for securing databases in various settings =-=[7,27,30,22,25,21]-=-. In particular, encryption is an important technique to protect sensitive data [13]. An analysis of how to encrypt and securely store data in relational database management systems has been given in ...

Citation Context

...r. Such methods are very useful in strengthening the protection of sensitive data in databases. 1.1 Related Work Various methods have been proposed recently for securing databases in various settings =-=[7,27,30,22,25,21]-=-. In particular, encryption is an important technique to protect sensitive data [13]. An analysis of how to encrypt and securely store data in relational database management systems has been given in ...

Citation Context

...r. Such methods are very useful in strengthening the protection of sensitive data in databases. 1.1 Related Work Various methods have been proposed recently for securing databases in various settings =-=[7,27,30,22,25,21]-=-. In particular, encryption is an important technique to protect sensitive data [13]. An analysis of how to encrypt and securely store data in relational database management systems has been given in ...

Citation Context

...r. Such methods are very useful in strengthening the protection of sensitive data in databases. 1.1 Related Work Various methods have been proposed recently for securing databases in various settings =-=[7,27,30,22,25,21]-=-. In particular, encryption is an important technique to protect sensitive data [13]. An analysis of how to encrypt and securely store data in relational database management systems has been given in ...

Citation Context

...ference and linking attacks as is pointed out in [11]. In comparison, our solution in Section 5 speeds up queries using metadata without introducing any additional information leakage. Agrawal et al. =-=[4]-=- propose a solution for range queries on numerical data that allows convenient indexing. Their solution is built on an encoding that preserves the order of the numerical data in each column. Consequen...

Citation Context

...are tested in the “where” conditions. 8sThe security of our scheme derives from the security of the block cipher we use. In cryptography, secure block ciphers are modeled as pseudorandom permutations =-=[18]-=-. Here, encryption key of the block cipher is the random seed for the pseudorandom permutation. For each value of the key, the mapping from the cleartext blocks to the ciphertext blocks is the permuta...

Citation Context

...ching for keywords in a file; however, it is unclear how to apply them to the problem of efficiently querying encrypted relational databases. Yet another piece of related work is by Feigenbaum et al. =-=[14]-=-, in which an encryption scheme was proposed to efficiently retrieve tuples from a look-up dictionary by using hash functions. The basic idea is that a tuple can only be retrieved if a valid key is pr...

Citation Context

.... In particular, encryption is an important technique to protect sensitive data [13]. An analysis of how to encrypt and securely store data in relational database management systems has been given in =-=[24]-=-. Recognizing the importance of encryption techniques, some database vendors have included encryption functionality in their products [1, 2]. By considering different privacy policies for different da...

Citation Context

...yptographic measure. Another issue is that, although the partition ids in [19] can be used for indexing to speed up queries, such an index can incur inference and linking attacks as is pointed out in =-=[11]-=-. In comparison, our solution in Section 5 speeds up queries using metadata without introducing any additional information leakage. Agrawal et al. [4] propose a solution for range queries on numerical...

Citation Context

...of the largest of these revealed sensitive information of millions of individuals. For example, in June 2005, names and credit card numbers of more than 40 million MasterCard cardholders were exposed =-=[12]-=-. In May 2006, disks containing the names, social security numbers, and dates of birth of more than 26 million United States veterans were stolen from the home of an employee of the Department of Vete...

Citation Context

...r. Such methods are very useful in strengthening the protection of sensitive data in databases. 1.1 Related Work Various methods have been proposed recently for securing databases in various settings =-=[7,27,30,22,25,21]-=-. In particular, encryption is an important technique to protect sensitive data [13]. An analysis of how to encrypt and securely store data in relational database management systems has been given in ...

Citation Context

..., disks containing the names, social security numbers, and dates of birth of more than 26 million United States veterans were stolen from the home of an employee of the Department of Veterans Affairs =-=[29]-=-. In the wrong hands, this kind of sensitive information can be to carry out identity theft and other fraudulent activities that harm the individuals involved and have a large cost to society. Techniq...