In the originally published version of this manuscript, there was an error in the affiliations. Affiliation 2 and 3 were inadvertently erroneously reversed. This has now been corrected.

Abstract Electronic voting (e-voting) has been studied for many years. Recently, researchers find that blockchain can provide an alternative secure platform for e-voting systems, because of its properties of tamper resistance and transparency. However, existing blockchain-based schemes either require central authorities to tally ballots or can only handle a limited number of voters. This paper tries to propose a self-tallying e-voting system, i.e. the public can verify the validity of all ballots and tally the ballots without a centralized authority. To achieve this goal, we solve two challenges, namely how to cancel out all random numbers used for ballots and to prove the validity of ballots using a non-interactive zero knowledge proof. Our scheme is proved to be secure and shown to be practical by experiments.

Searchable compressed data structures (e.g.Burrows-Wheeler Transform) enable one to create a memory-efficient index for large datasets such as human genomes. On the other hand, storing such an index in a third-party server, e.g., cloud, may have the privacy and confidentiality issues. An open problem in the community is to construct a secure variant of such a data structure. This problem is challenging as most of the existing works were shown to be insecure and none of them is able to perform pattern matching. In this paper, we provide the first solution based on Burrows-Wheeler Transform (BWT) to solve this problem (our scheme can do both compression and pattern matching). A new security definition, called isomophism-restricted IND-CPA security, is proposed. We show that our scheme is secure under this definition and our scheme is practical by experiments.

With the popularity of cloud computing, an increasing number of institutions outsource their data to a third-party cloud system which could be untrusted. The institutions encrypt their data before outsourcing to protect data privacy. On the other hand, data mining techniques are used widely but computationally intensive, especially for large datasets. Combining data from different institutions for a big and varied training set helps enhance data mining performance. Therefore, it is important to make the cloud system which has powerful computing abilities run data mining algorithms on the encrypted data from multiple institutions. Two challenges need attention – how to compute on encrypted data under multiple keys and how to verify the correctness of the result. There are no existing methods that solve the two challenges at the same time. Elastic net is a useful linear regression tool to find genomic biomarkers. In this paper, we propose the first privacy-preserving verifiable elastic net protocol based on reduction to support vector machine using two non-colluding servers. We construct a homomorphic cryptosystem that supports one multiply operation and multiple add operations under both single key and different keys. We allow the involved institutions to verify the correctness of the final result. The collaboration between multiple institutions is made possible without jeopardizing the privacy of data records. We formally prove that our protocol is secure and implement the protocol. The experimental results show that our protocol runs reasonably fast, and thus can be applied in practice.

Searching encrypted database is an important topic as more users want to leverage a third-party cloud system to store and process their data in encrypted form. Despite a lot of wonderful results, there are still a number of unsolved problems. In particular, the problems of pattern matching (not keyword search), e.g. with wildcards, that supports secure boolean queries and how to determine the value of k automatically of a top-k search for different queries on encrypted data are not properly addressed. In this paper, we provide solutions to solve these problems. Also, most existing secure databases employ different encryption functions to support different operators. The only exception is SDB (SIGMOD'2014) that was designed to support data interoperability between integers with a unified encryption scheme so that sophisticated queries can be answered by the database. However, SDB does not support string matching queries. We show that our solutions can be made compatible with SDB to fill this gap. To the best of our knowledge, we are the first to investigate these problems. We prove that our scheme is secure against chosen query attack. We have evaluated the performance of our scheme on large (105strings) real-world datasets, and showed that our scheme can achieve a high search quality of 99.9% recall and 98.6% accuracy with reasonable response time.

Blockchain technology has attracted a lot of attention in recent years. Applications of blockchain are not only restricted to cybercurrencies, but have also been extended to other areas such as finance, e-health, music, and other business. One of the key components of blockchain is the design for miners who are responsible for adding new transactions (blocks) by solving a puzzle and receive some rewards in return. As a result, miners tend to join centralized mining pools to outsource their computing resources in order to gain more steady rewards, which may affect the security and fairness of the system. This motivates the researchers to propose nonoutsourceable puzzles. However, existing nonoutsourceable puzzles do not work well under the high-rate transaction processing protocol (GHOST). In this paper, we propose the first nonoutsourceable puzzle that can satisfy all security requirements of GHOST. Our experimental results show that our puzzle is practical.