Securing Data in Future Computer Systems

点击翻译按钮获取中文摘要

Protection and control of digital data are increasingly threatened in two fundamental ways. First, the computational assumptions that underlie the security of today's cryptosystems will likely to be substantially weakened or completely broken in the near future. In particular, the advent of quantum computing will break the computational assumptions that are the basis of the security algorithms that are the backbone of secure communication: RSA encryption, and Diffie-Hellman (key establishment) protocol. Second, massive data publication and sharing will result in unwanted and uncontrollable inferences. Using search engines and data analysis techniques could reveal information that data owners had intended to protect, making it increasingly hard to maintain individual privacy and to implement laws on freedom of information that require protection of sensitive information. Protected data publishing (PDP) systems work on the assumption that removing terms from documents, or modifying parts of it, will provide the required protection. It has been shown that with sufficient related data, this assumption will not be valid. The long-term objectives of my research program are: To contribute security models for data protection that are inspired by real life scenarios; to develop theories, tools, and techniques to analyze and understand them; and to design systems with verifiable security properties that last through the coming decades. My research will use two complementary approaches, cryptographic and non-cryptographic. A cryptographic system with information theoretic security does not rely on any computational assumption and its security is proven through mathematical theorems that will not be affected with advances in computer algorithms, or development of new models of computation such as quantum computers. A large class of data protection problems, however, cannot be solved by cryptographic systems. I will use probabilistic interpretation of data and information theoretic measures to propose models, theories and protection systems for such problems. I will implement the algorithms and experiment on real data to verify the models and test the theories. Using the two complementary approaches will enable me to explore and contribute to a wide range of emerging problems, with the unifying thread being the use of information theoretic measures, arguments, and proofs. My short term objectives are: 1. Cryptographic systems for long-term data security. Propose models that capture two-party and multi-party data communication scenarios, analyze the models, and propose constructions with optimal performance, provable security, and efficiency. 2. Non-cryptographic systems for PDP. Explore information leakage, model and analyze the leakage in PDP, and propose protection systems to control the leakage. Evaluate the models and constructions, and test theories using experiments on real data. The proposed research is essential for long term security of digital data. Next generation information communication infrastructure will operate in an environment of massive data and abundant computation, and require cryptographic systems that remain secure irrespective of the computational power of the adversary, and tools and systems that can control information leakage in published data. The need for cryptographic technologies for future has been recognized by standardization bodies such as ETSI (http://www.etsi.org/news-events/events/648-crypto-workshop2013), and the need for technologies for controlling unwanted inferences is the today's need of individuals and institutions, which will grow in the coming years.