摘要:
以机器学习、人工智能、物联网等技术所构建的智能环境正在改变人们的生活、工作及思维方式.智能环境下数据存储和处理的方式也在不断改变，其中安全和效率是2个重要的因素.就安全而言，在数据共享的前提下保护隐私势在必行.就效率而言，智能环境中存在诸多资源受限的设备，针对这些设备如何设计高效的算法或协议直接决定其可行性.从以上2个需求出发，研究适用于智能环境中的安全高效模式匹配问题.传统的安全模式匹配协议中模式持有方需要执行大量的公钥操作，因此不适用于手机等资源受限设备作为模式持有方的场景.首次在双云服务器辅助的安全两方计算模型下给出安全模式匹配协议的功能函数，并基于茫然传输(oblivious transfer, OT)给出协议的具体构造.假设云服务器和参与方之间不合谋，协议在半诚实敌手模型下是安全的.协议需要4轮交互，模式方仅需要执行少量的异或操作，而复杂的OT协议主要集中在数据库方和云服务器之间.此外，使用OT扩展(OT extension)技术可以将所有OT协议的数量从O(nm)降至O(k)，其中n和m是数据库方和模式方的输入长度，k是OT扩展协议中基础OT的数目，其远小于nm.
Abstract:
The intelligent environment built by machine learning, artificial intelligence, Internet of things is changing the way people live, work and think. The way of data storage and processing in intelligent environment is changing constantly, where security and efficiency are two important factors. In terms of security, privacy must be guaranteed during the sharing and analysis of data. In addition, many devices with limited resources exist in the intelligent environment, whose feasibility is directly influenced by how to design suitable algorithms or protocols. Based on the above two requirements, this paper studies the problem of secure pattern matching for intelligent environment. In the traditional secure pattern matching protocol, the pattern holder needs to compute lots of public-key operations，which is unsuitable for a resource-limited device such as a mobile phone. In this paper, we formalize the functionality of the secure pattern matching protocol under the two-cloud-assisted secure two-party computation model for the first time, and construct an efficient protocol via oblivious transfer (OT). The protocol is secure in semi-honest adversary model, assuming that no collusion exists between the cloud servers and the participants. The protocol requires 4 rounds, and the pattern holder performs only a small number of XOR operations, and the complex OT protocols are mainly executed between the database and the cloud servers. Furthermore, the OT extension technology can reduce the number of all OT protocols from O(nm) to O(k), where n and m are the input lengths of the two participants, and k is the number of base OT in OT extension protocol, which is much smaller than nm.
下载:
