Searcharxiv⌕ Search

arXiv subjects

Lewis Tseng

Publications and source records attributed to Lewis Tseng.

34 records · Page 2Linked to original sources
arXiv2017

An Improved Approximate Consensus Algorithm in the Presence of Mobile Faults

This paper explores the problem of reaching approximate consensus in synchronous point-to-point networks, where each pair of nodes is able to communicate with each other directly and reliably. We consider the mobile Byzantine fault model proposed by Garay '94 -- in the model, an omniscient adversary can corrupt up to $f$ nodes in each round, and at the beginning of each round, faults may "move" in the system (i.e., different sets of nodes may become faulty in different rounds). Recent work by Bonomi et al. '16 proposed a simple iterative approximate consensus algorithm which requires at least $4f+1$ nodes. This paper proposes a novel technique of using "confession" (a mechanism to allow others to ignore past behavior) and a variant of reliable broadcast to improve the fault-tolerance level. In particular, we present an approximate consensus algorithm that requires only $\lceil 7f/2\rceil + 1$ nodes, an $\lfloor f/2 \rfloor$ improvement over the state-of-the-art algorithms. Moreover, we also show that the proposed algorithm is optimal within a family of round-based algorithms.

cs.DC↗
arXiv2016

Recent Results on Fault-Tolerant Consensus in Message-Passing Networks

Fault-tolerant consensus has been studied extensively in the literature, because it is one of the most important distributed primitives and has wide applications in practice. This paper surveys important results on fault-tolerant consensus in message-passing networks, and the focus is on results from the past decade. Particularly, we categorize the results into two groups: new problem formulations and practical applications. In the first part, we discuss new ways to define the consensus problem, which includes larger input domains, link fault models, different network models . . . etc, and briefly discuss the important techniques. In the second part, we focus on Crash Fault-Tolerant (CFT) systems that use Paxos or Raft, and Byzantine Fault-Tolerant (BFT) systems. We also discuss Bitcoin, which can be related to solving Byzantine consensus in anonymous systems, and compare Bitcoin with BFT systems and Byzantine consensus.

cs.DC↗
arXiv2016

Characterizing and Adapting the Consistency-Latency Tradeoff in Distributed Key-value Stores

The CAP theorem is a fundamental result that applies to distributed storage systems. In this paper, we first present and prove two CAP-like impossibility theorems. To state these theorems, we present probabilistic models to characterize the three important elements of the CAP theorem: consistency (C), availability or latency (A), and partition tolerance (P). The theorems show the un-achievable envelope, i.e., which combinations of the parameters of the three models make them impossible to achieve together. Next, we present the design of a class of systems called PCAP that perform close to the envelope described by our theorems. In addition, these systems allow applications running on a single data-center to specify either a latency SLA or a consistency SLA. The PCAP systems automatically adapt, in real-time and under changing network conditions, to meet the SLA while optimizing the other C/A metric. We incorporate PCAP into two popular key-value stores -- Apache Cassandra and Riak. Our experiments with these two deployments, under realistic workloads, reveal that the PCAP system satisfactorily meets SLAs, and performs close to the achievable envelope. We also extend PCAP from a single data-center to multiple geo-distributed data-centers.

cs.DC↗
arXiv2015

Application-Aware Consistency: An Application to Social Network

This work weakens well-known consistency models using graphs that capture applications' characteristics. The weakened models not only respect application semantic, but also yield a performance benefit. We introduce a notion of dependency graphs, which specify relations between events that are important with respect to application semantic, and then weaken traditional consistency models (e.g., causal consistency) using these graphs. Particularly, we consider two types of graphs: intra-process and inter-process dependency graphs, where intra-process dependency graphs specify how events in a single process are related, and inter-process dependency graphs specify how events across multiple processes are related. Then, based on these two types of graphs, we define new consistency model, namely {\em application-aware} consistency. We also discuss why such application-aware consistency can be useful in social network applications. This is a work in progress, and we present early ideas regarding application-aware consistency here.

cs.DC↗
arXiv2015

Asynchronous Convex Consensus in the Presence of Crash Faults

This paper defines a new consensus problem, convex consensus. Similar to vector consensus [13, 20, 19], the input at each process is a d-dimensional vector of reals (or, equivalently, a point in the d-dimensional Euclidean space). However, for convex consensus, the output at each process is a convex polytope contained within the convex hull of the inputs at the fault-free processes. We explore the convex consensus problem under crash faults with incorrect inputs, and present an asynchronous approximate convex consensus algorithm with optimal fault tolerance that reaches consensus on an optimal output polytope. Convex consensus can be used to solve other related problems. For instance, a solution for convex consensus trivially yields a solution for vector consensus. More importantly, convex consensus can potentially be used to solve other more interesting problems, such as convex function optimization [5, 4].

cs.DC↗
arXiv2015

Byzantine Broadcast Under a Selective Broadcast Model for Single-hop Wireless Networks

This paper explores an old problem, {\em Byzantine fault-tolerant Broadcast} (BB), under a new model, {\em selective broadcast model}. The new model "interpolates" between the two traditional models in the literature. In particular, it allows fault-free nodes to exploit the benefits of a broadcast channel (a feature from reliable broadcast model) and allows faulty nodes to send mismatching messages to different neighbors (a feature from point-to-point model) simultaneously. The {\em selective broadcast} model is motivated by the potential for {\em directional} transmissions on a wireless channel. We provide a collection of results for a single-hop wireless network under the new model. First, we present an algorithm for {\em Multi-Valued} BB that is order-optimal in bit complexity. Then, we provide an algorithm that is designed to achieve BB efficiently in terms of message complexity. Third, we determine some lower bounds on both bit and message complexities of BB problems in the {\em selective broadcast model}. Finally, we present a conjecture on an "exact" lower bound on the bit complexity of BB under the {\em selective broadcast} model.

cs.DC↗
arXiv2015

Crash-Tolerant Consensus in Directed Graphs

This work considers a point-to-point network of n nodes connected by directed links, and proves tight necessary and sufficient conditions on the underlying communication graphs for achieving consensus among these nodes under crash faults. We identify the conditions in both synchronous and asynchronous systems

cs.DC↗
arXiv2014

Exact Byzantine Consensus in Directed Graphs

Consider a synchronous point-to-point network of n nodes connected by directed links, wherein each node has a binary input. This paper proves a tight necessary and sufficient condition on the underlying communication topology for achieving Byzantine consensus among these nodes in the presence of up to f Byzantine faults. We derive a necessary condition, and then we provide a constructive proof of sufficiency by presenting a Byzantine consensus algorithm for directed graphs that satisfy the necessary condition. Prior work has developed analogous necessary and sufficient conditions for undirected graphs. It is known that, for undirected graphs, the following two conditions are together necessary and sufficient [8, 2, 6]: (i) n ? 3f + 1, and (ii) network connectivity greater than 2f. However, these conditions are not adequate to completely characterize Byzantine consensus in directed graphs.

cs.DC↗
arXiv2014

Iterative Approximate Consensus in the presence of Byzantine Link Failures

This paper explores the problem of reaching approximate consensus in synchronous point-to-point networks, where each directed link of the underlying communication graph represents a communication channel between a pair of nodes. We adopt the transient Byzantine link failure model [15, 16], where an omniscient adversary controls a subset of the directed communication links, but the nodes are assumed to be fault-free. Recent work has addressed the problem of reaching approximate consen- sus in incomplete graphs with Byzantine nodes using a restricted class of iterative algorithms that maintain only a small amount of memory across iterations [22, 21, 23, 12]. However, to the best of our knowledge, we are the first to consider approximate consensus in the presence of Byzan- tine links. We extend our past work that provided exact characterization of graphs in which the iterative approximate consensus problem in the presence of Byzantine node failures is solvable [22, 21]. In particular, we prove a tight necessary and sufficient condition on the underlying com- munication graph for the existence of iterative approximate consensus algorithms under transient Byzantine link model. The condition answers (part of) the open problem stated in [16].

cs.DC↗
arXiv2013

Broadcast Using Certified Propagation Algorithm in Presence of Byzantine Faults

We explore the correctness of the Certified Propagation Algorithm (CPA) [6, 1, 8, 5] in solving broadcast with locally bounded Byzantine faults. CPA allows the nodes to use only local information regarding the network topology. We provide a tight necessary and sufficient condition on the network topology for the correctness of CPA. To the best of our knowledge, this work is the first to solve the open problem in [8]. We also present some simple extensions of this result

cs.DC↗
arXiv2013

Byzantine Convex Consensus: An Optimal Algorithm

Much of the past work on asynchronous approximate Byzantine consensus has assumed scalar inputs at the nodes [4, 8]. Recent work has yielded approximate Byzantine consensus algorithms for the case when the input at each node is a d-dimensional vector, and the nodes must reach consensus on a vector in the convex hull of the input vectors at the fault-free nodes [9, 13]. The d-dimensional vectors can be equivalently viewed as points in the d-dimensional Euclidean space. Thus, the algorithms in [9, 13] require the fault-free nodes to decide on a point in the d-dimensional space. In our recent work [arXiv:/1307.1051], we proposed a generalization of the consensus problem, namely Byzantine convex consensus (BCC), which allows the decision to be a convex polytope in the d-dimensional space, such that the decided polytope is within the convex hull of the input vectors at the fault-free nodes. We also presented an asynchronous approximate BCC algorithm. In this paper, we propose a new BCC algorithm with optimal fault-tolerance that also agrees on a convex polytope that is as large as possible under adversarial conditions. Our prior work [arXiv:/1307.1051] does not guarantee the optimality of the output polytope.

cs.DC↗
arXiv2013

Byzantine Convex Consensus: Preliminary Version

Much of the past work on asynchronous approximate Byzantine consensus has assumed scalar inputs at the nodes [3, 7]. Recent work has yielded approximate Byzantine consensus algorithms for the case when the input at each node is a d-dimensional vector, and the nodes must reach consensus on a vector in the convex hull of the input vectors at the fault-free nodes [8, 12]. The d-dimensional vectors can be equivalently viewed as points in the d-dimensional Euclidean space. Thus, the algorithms in [8, 12] require the fault-free nodes to decide on a point in the d-dimensional space. In this paper, we generalize the problem to allow the decision to be a convex polytope in the d-dimensional space, such that the decided polytope is within the convex hull of the input vectors at the fault-free nodes. We name this problem as Byzantine convex consensus (BCC), and present an asynchronous approximate BCC algorithm with optimal fault tolerance. Ideally, the goal here is to agree on a convex polytope that is as large as possible. While we do not claim that our algorithm satisfies this goal, we show a bound on the output convex polytope chosen by our algorithm.

cs.DC↗
arXiv2012

Iterative Approximate Byzantine Consensus under a Generalized Fault Model

In this work, we consider a generalized fault model that can be used to represent a wide range of failure scenarios, including correlated failures and non-uniform node reliabilities. This fault model is general in the sense that fault models studied in prior related work, such as f -total and f -local models, are special cases of the generalized fault model. Under the generalized fault model, we explore iterative approximate Byzantine consensus (IABC) algorithms in arbitrary directed networks. We prove a necessary and sufficient condition for the existence of IABC algorithms. The use of the generalized fault model helps to gain a better understanding of IABC algorithms.

cs.DC↗
arXiv2012

Iterative Approximate Byzantine Consensus in Arbitrary Directed Graphs - Part II: Synchronous and Asynchronous Systems

This report contains two related sets of results with different assumptions on synchrony. The first part is about iterative algorithms in synchronous systems. Following our previous work on synchronous iterative approximate Byzantine consensus (IABC) algorithms, we provide a more intuitive tight necessary and sufficient condition for the existence of such algorithms in synchronous networks1. We believe this condition and the previous results also hold in partially asynchronous algorithmic model. In the second part of the report, we explore the problem in asynchronous networks. While the traditional Byzantine consensus is not solvable in asynchronous systems, approximate Byzantine consensus can be solved using iterative algorithms.

cs.DC↗
arXiv2012

Iterative Approximate Byzantine Consensus in Arbitrary Directed Graphs

In this paper, we explore the problem of iterative approximate Byzantine consensus in arbitrary directed graphs. In particular, we prove a necessary and sufficient condition for the existence of iterative byzantine consensus algorithms. Additionally, we use our sufficient condition to examine whether such algorithms exist for some specific graphs.

cs.DC↗