12th International Symposium on Stabilization, Safety, and Security of Distributed Systems

This special issue contains four articles that are based on extended versions of papers presented at the 12th International Symposium on Stabilization, Safety, and Security of Distributed Systems (SSS 2010), held at Columbia University, NYC, USA, on September 20–22, 2010. The papers chosen for this special issue were among the highest ranked papers that were chosen for presentation at SSS 2010 after a rigorous peer-review process (90 papers were submitted to SSS 2010, 39 were accepted for presentation). Compared with the original papers presented at SSS 2010, the articles have been strongly revised and extended by full proofs and additional results. They also endured an additional reviewing process that followed the Information and Computation standard and was very rigorous. This additional review was performed independently from the selection process of SSS 2010. Two articles are in the area of security. One of these investigates the behavior of public-key systems where the keys have been silently compromised. The other develops algorithms for secure polling in social networks. The remaining two articles examine properties of abstract representations of computations. One of these presents algorithms for the detection of liveness properties in infinite computations. The other examines the computing capabilities of an alternative Turing machine model. The article Authenticated Broadcast with a Partially Compromised Public-Key Infrastructure, by S. Dov Gordon, Jonathan Katz, Ranjit Kumaresan, and Arkady Yerukhimovich, addresses performing a secure broadcast in a public key infrastructure when some of the public keys have been compromised. Thus, they consider three types of nodes: corrupted nodes, whose behavior is arbitrary, compromised nodes, whose keys have been stolen by corrupted nodes without their knowledge, and uncompromised nodes. The authors develop a complete characterization of the problem by establishing upper bounds on the number of corrupted and compromised nodes to perform a secure broadcast, and present broadcast protocols that achieve these bounds. The article Computing in Social Networks, by Andrei Giurgiu, Rachid Guerraoui, Kévin Huguenin, and Anne-Marie Kermarrec, defines the problem of Scalable Secure computing in a Social network (S3), and presents a protocol for its solution. In this problem, users need to compute a symmetric function from their individual inputs in a secure way. A typical function could be a binary poll (yes/no answer) of all the users. The protocol is secure if malicious users cannot significantly modify the output of the function, nor can they identify the input of a particular user, without being discovered by other users. Key to the solution is a dedicated overlay structure that enables secret sharing among users, and that users of social media care about their reputation and do not want to be tagged as misbehaving. The article Modeling, Analyzing and Slicing Periodic Distributed Computations, by Vijay K. Garg, Anurag Agarwal, and Vinit Ogale, addresses the detection of liveness properties in distributed computations. It enhances earlier work by removing the assumption that a given computation is finite. Their approach consists of three parts: identifying recurring states, obtaining a finite representation of an infinite computation, and determining if a given predicate ever becomes true during the computation. They also present an algorithm that concisely captures all the consistent global states of the computation that satisfies the given predicate. The article Effective Storage Capacity of Labeled Graphs, by Dana Angluin, James Aspnes, Rida A. Bazzi, Jiang Chen, David Eisenstat, and Goran Konjevod, presents an alternative model of computation consisting of an undirected graph, where each node can store a symbol from a finite set, and a finite-state controller (a Turing machine head) that can move along the graph. However, there is no sense of direction because the graph is undirected. The motivation for this model is selforganizing systems consisting of many communicating finite-state machines, where at any time, one machine (the location of the Turing machine head) takes a leadership role. The main question addressed is how much computing power such machines can cooperate to achieve.