Sharding is a popular way to achieve scalability in blockchain protocols, increasing their throughput by partitioning the set of transaction validators into a number of smaller committees, splitting the workload. Existing approaches for blockchain sharding, however, do not scale well when concurrent transactions alter the same replicated state component—a common scenario in Ethereum-style smart contracts.

The rise of programmable open distributed consensus platforms based on the blockchain technology has aroused a lot of interest in replicated stateful computations, aka smart contracts. As blockchains are used predominantly in financial applications, smart contracts frequently manage millions of dollars worth of virtual coins. Since smart contracts cannot be updated once deployed, the ability to reason about their correctness becomes a critical task. Yet, the de facto implementation standard, pioneered by the Ethereum platform, dictates smart contracts to be deployed in a low-level language, which renders independent audit and formal verification of deployed code infeasible in practice. We report an ongoing experiment held with an industrial blockchain vendor on designing, evaluating, and deploying Scilla, a new programming language for safe smart contracts. Scilla is positioned as an intermediate-level language, suitable to serve as a compilation target and also as an independent programming framework. Taking System F as a foundational calculus, Scilla offers strong safety guarantees by means of type soundness. It provides a clean separation between pure computational, state-manipulating, and communication aspects of smart contracts, avoiding many known pitfalls due to execution in a byzantine environment. We describe the motivation, design principles, and semantics of Scilla, and we report on Scilla use cases provided by the developer community. Finally, we present a framework for lightweight verification of Scilla programs, and showcase it with two domain-specific analyses on a suite of real-world use cases.

Data compression is ubiquitous to any information and communication system. It often reduces resources required to store and transmit data. However, the efficiency of compression algorithms also makes them an obvious target for hackers to mount denial-of-service attacks. In this work, we consider decompression quines, a specific class of compressed files that decompress to themselves. We analyze all the known decompression quines by studying their structures, and their impact on anti-viruses. Our analysis reveals that most of the anti-viruses do not have a suitable architecture in place to detect decompression quines. Even worse, some of them are vulnerable to denial-of-service attacks exploiting quines. Motivated by our findings, we study several quine detectors and propose a new one that exploits the fact that quines and non-quine files do not share the same underlying structure. Our evaluation against different datasets shows that the detector incurs no performance overhead at the expense of a low false positive rate.

Accountability plays a key role in dependable distributed systems. It allows to detect, isolate and churn malicious/selfish nodes that deviate from a prescribed protocol. To achieve these properties, several accountable systems use at their core cryptographic primitives that produce non-repudiable evidence of inconsistent or incorrect behavior.

Google and Yandex Safe Browsing are popular services included in many web browsers to prevent users from visiting phishing or malware websites. If these services protect their users from losing private information, they also require that their servers receive browsing information on the very same users. In this paper, we analyze Google and Yandex Safe Browsing services from a privacy perspective. We quantify the privacy provided by these services by analyzing the possibility of re-identifying URLs visited by a client. We thereby challenge Google's privacy policy which claims thatGoogle cannot recover URLs visited by its users. Our analysis and experimental results show that Google and Yandex Safe Browsing canpotentially be used as a tool to track specific classes of individuals. Additionally, our investigations on the data currently included in Google and Yandex Safe Browsing provides a concrete set of URLs/domains that can be re-identified without much effort.

A Bloom filter is a probabilistic hash-based data structure extensively used in software including online security applications. This paper raises the following important question: Are Bloom filters correctly designed in a security context? The answer is no and the reasons are multiple: bad choices of parameters, lack of adversary models and misused hash functions. Indeed, developers truncate cryptographic digests without a second thought on the security implications. This work constructs adversary models for Bloom filters and illustrates attacks on three applications, namely SCRAPY web spider, BITLY DABLOOMS spam filter and SQUID cache proxy. As a general impact, filters are forced to systematically exhibit worst-case behavior. One of the reasons being that Bloom filter parameters are always computed in the average case. We compute the worst-case parameters in adversarial settings, show how to securely and efficiently use cryptographic hash functions and propose several other countermeasures to mitigate our attacks.

Cryptographic accumulators are space/time efficient data structures used to verify if a value belongs to a set. They have found many applications in networking and distributed systems since their introduction by Benaloh and de Mare in 1993. Despite this popularity, there is currently no thorough performance evaluation of the different existing designs. Symmetric and asymmetric accumulators are used likewise without any particular argument to support either of the design. We aim to establish the speed of each design and their application's domains in terms of their size and the size of the values.