Microsoft Technology Licensing, LLC

Seny Fakaba Kamara

Mariana Raykova

Melissa Erin Chase

Charalampos Papamanthou

Thomas Michael Roeder

Payman Mohassel

Tarik Moataz

Sze Ming Chow

Mariana P Raykova

Described herein is an efficient, dynamic Symmetric Searchable Encryption (SSE) scheme. A client computing device includes a plurality of files and a dictionary of keywords. An index is generated that indicates, for each keyword and each file, whether a file includes a respective keyword. The index is encrypted and transmitted (with encryptions of the files) to a remote repository. The index is dynamically updateable at the remote repository, and can be utilized to search for files that include keywords in the dictionary without providing the remote repository with information that identifies content of the file or the keyword.

Dynamic symmetric searchable encryption

A security controller controls secure processing of queries in an encrypted relational database. A query controller receives, from a client device, a secure query in a format of an encrypted token generated using a structured query language (SQL) query in a conjunctive query form, and sends an encrypted response to the secure query to the client device. A search engine generates the encrypted response to the secure query by initiating a search on the encrypted relational database, without decrypting the secure query and without decrypting the encrypted multi-maps. The encrypted relational database includes encrypted multi-maps corresponding to a relational database hosted at the client device, and an encrypted dictionary, based on structured encryption, using structured encryption, in lieu of using property-preserving encryption (PPE), and in lieu of using fully homomorphic encryption (FHE).

Controlling security in relational databases

A server module evaluates a circuit based on concealed inputs provided by respective participant modules, to provide a concealed output. By virtue of this approach, no party to the transaction (including the sever module) discovers any other party's non-concealed inputs. In a first implementation, the server module evaluates a garbled Boolean circuit. This implementation also uses a three-way oblivious transfer technique to provide a concealed input from one of the participant modules to the serer module. In a second implementation, the server module evaluates an arithmetic circuit based on ciphertexts that have been produced using a fully homomorphic encryption technique. This implementation modifies multiplication operations that are performed in the evaluation of the arithmetic circuit by a modifier factor; this removes bounds placed on the number of the multiplication operations that can be performed.

Secure computation using a server module

A server module evaluates a circuit based on concealed inputs provided by respective participant modules, to provide a concealed output. By virtue of this approach, no party to the transaction (including the sever module) discovers any other party's non-concealed inputs. In a first implementation, the server module evaluates a garbled Boolean circuit. This implementation also uses a three-way oblivious transfer technique to provide a concealed input from one of the participant modules to the server module. In a second implementation, the server module evaluates an arithmetic circuit based on ciphertexts that have been produced using a fully homomorphic encryption technique. This implementation modifies multiplication operations that are performed in the evaluation of the arithmetic circuit by a modifier factor; this removes bounds placed on the number of the multiplication operations that can be performed.

Existing private set intersection (PSI) protocol allows two parties to find intersection of their sets, but restricts learning any other information about each other's set except for its size. In general, the server-aided private set intersection with data transfer technique described herein provides a server-aided private set intersection (PSI) protocol that supports data transfers. The technique pertains to a method for providing a server-aided private set intersection protocol which allows two parties to transfer some of the information about their elements via an untrusted third party. The protocol involves (a) parties applying a shared pseudo-random permutation to each of their sets to create labels of the elements of the set, (b) sending the labels to the third party and (c) the third party performing data transfer between the two parties along with computation of intersection of sets received using a multi-share key.

Server-aided private set intersection (PSI) with data transfer

The disclosed architecture employs techniques that make secure multi-party computation (MPC) practical and scalable. In support of utilizing cloud computing, for example, for evaluating functionality, a third party server can be employed which does not have any input to the computation and does not receive any output from the computation, yet has a vast amount of computational resources. Accordingly, the secure MPC architecture can outsource as much as possible of the computation and communications burden of the parties without the server(s) learning any information about the party inputs.

Server-aided multi-party protocols

A storage system stores information about a graph in an encrypted form. A query module can submit a token to the storage system to retrieve specified information about the graph, e.g., to determine the neighbors of an entity in the graph, or to determine whether a first entity is connected to a second entity, etc. The storage system formulates its reply to the token in a lookup result. Through this process, the storage system gives selective access to information about the graph to authorized agents, yet otherwise maintains the general secrecy of the graph from the perspective of unauthorized agents, including the storage system itself. A graph processing module can produce encrypted graph information by encrypting any representation of the graph, such as an adjacency matrix, an index, etc.

Graph encryption

Anonymous information sharing systems and methods enable communication of information to parties in a privacy-preserving manner such that no one other than the designated parties can know the source, recipient, and content of the information. Furthermore, the communication can be accomplished without requiring trial decryption, and protection can be provided against of sharing of privileges.

Privacy-preserving communication

Shares for one or more data values in a dataset can be computed using evaluation point values and sharing polynomials. Lagrangian coefficients can also be computed for the evaluation point values. The shares and the Lagrangian coefficients may be used to evaluate the polynomials on the data values. The technique can also include encrypting the Lagrangian coefficients according to an encryption scheme that provides for addition operations between encrypted values. An operation on representations of coefficients of the evaluation polynomial, representations of the shares, and the encrypted representations of the Lagrangian coefficients can be delegated to a remote computing environment. The operation can be performed at the remote computing environment, such as by performing a map-reduce operation. Results of the delegated operation can be received from the remote computing environment and processed to produce representation(s) of evaluation(s) of the polynomial on the data value(s).

Polynomial evaluation delegation

This document describes techniques and apparatuses for secure computing in multi-tenant data centers. These techniques permit a client to delegate computation of a function to multiple physical computing devices without the client's information being vulnerable to exposure. The techniques prevent discovery of the client's information by a malicious entity even if that entity is a co-tenant on many of the same physical computing devices as the client.

Secure computing in multi-tenant data centers

A Searchable Symmetric Encryption (SSE) mechanism is described which allows efficient dynamic updating of encrypted index information. The encrypted index information includes pointer information that is encrypted using a malleable encryption scheme. The SSE mechanism updates the encrypted index information by modifying at least one instance of the pointer information without decrypting the pointer information, and thereby without revealing the nature of the changes being made. In one implementation, the SSE mechanism includes a main indexing structure and a deletion indexing structure. An updating operation involves patching applied to both the main indexing structure and deletion indexing structure.

Searchable symmetric encryption with dynamic updating

Counting values can be encrypted as a set of counting value cyphertexts according to a hidden vector encryption scheme using sample values of a set of samples, where each of the samples can include multiple sample values. Additionally, tokens can be generated. The tokens can be configured according to the hidden vector encryption scheme, such that each of the tokens can enable decryption of matching cyphertexts. Processing of the counting value cyphertexts and the tokens can be delegated to a map-reduce computer cluster. The cluster can run a map-reduce program to produce and return count representations. Each count representation can represent a count of a set of the counting value cyphertext(s) whose decryption was enabled by one or more of the token(s). For example, the counts may be counts that can be used in constructing a data structure such as a decision tree.

Growing community of inventors

Seattle, WA, United States of America

Seny Fakaba Kamara