The distributed spanning tree (DST) is an overlay structure designed to be scalable. It supports the growth from a small number of nodes to a large one. The DST is a tree without bottlenecks which automatically balanc...
详细信息
The distributed spanning tree (DST) is an overlay structure designed to be scalable. It supports the growth from a small number of nodes to a large one. The DST is a tree without bottlenecks which automatically balances the load between its nodes. This sound paradoxale, but the DST breaks the common assumption that a tree is build of leaves and intermediate nodes. In a DST every nodes are equal. The nodes are put together into small cliques. Then, the cliques are put together into small cliques of higher level recursively. The cliques are represented in each node by a routing table. The memory space complexity of the routing tables is O(log(n)) for a n nodes DST. A theoretical description of the DST was already published but it does not provide enough information about the capacities of the DST. The purpose of this article is to give a practical view of what can be done with a DST. This document can be considered as a DST traversal catalog. It also outlines some characteristics and shares the lessons learned from our implementation errors.
In this correspondence, we first point out an error in Jia's algorithm (1998) by a counterexample. Second, we provide a fix and make an improvement to the part of Dynamic-Join. Finally, we give an analysis and a c...
详细信息
In this correspondence, we first point out an error in Jia's algorithm (1998) by a counterexample. Second, we provide a fix and make an improvement to the part of Dynamic-Join. Finally, we give an analysis and a correctness proof of our algorithm.
This paper describes a solution to workforce allocation, modelled as a distributed system. The novelty of the solution resides in the fact that it is a natural and versatile formulation that combines an agent-based mo...
详细信息
This paper describes a solution to workforce allocation, modelled as a distributed system. The novelty of the solution resides in the fact that it is a natural and versatile formulation that combines an agent-based model with rule-based expressions of allocation strategies and multi-criteria optimisation expressions of allocation objectives. The problem space is structured on disjoint domains. Each domain has two agents: one in charge of its engineers and one in charge of its jobs. In the implemented system, agents have complete decision power over the allocation of their resources/requirements. However, the general solution is readily extendable towards the inclusion of central agents with specialist roles. Workforce allocation is modelled here as an iterative communication process, based on a 4-step communication protocol, between workforce/engineers' agents, on one hand, and jobs agents, on the other. Choices (decisions) regarding the allocation are made locally, within each agent. A global allocation emerges from this interaction/communication. The paper describes the context of this solution (including the problem description), the general model and solution, and specific implementations. The latter illustrate the emergence of good global solutions from the distributed algorithm.
This paper proposes a exible stochastic automaton-based network partitioning algorithm that is capable of find-ing the optimal k-way partition with respect to a broad range of cost functions, and given various constra...
详细信息
ISBN:
(纸本)0780389166
This paper proposes a exible stochastic automaton-based network partitioning algorithm that is capable of find-ing the optimal k-way partition with respect to a broad range of cost functions, and given various constraints, in directed and weighted graphs. Further, this iterative algorithm requires only local computation, with respect to the graph. Hence, by incorporating a distributed stopping criterion, we have been able to solve certain partitioning problems in a totally distributed manner. In this article, this inuence model-based partitioning algorithm is motivated and introduced, and is shown to find the optimal partition for a large class of problems. Also, a conceptual discussion of why the algorithm might be expected to find good partitions quickly is included, and the performance of the algorithm is illustrated through examples. Applications in partitioning distributed communicatin-agent networks, sensor systems, and power grids are discussed.
Classification based on decision trees is one of the important problems in data mining and has applications in many fields. In recent years, database systems have become highly distributed, and distributed system para...
详细信息
ISBN:
(纸本)9780898715934
Classification based on decision trees is one of the important problems in data mining and has applications in many fields. In recent years, database systems have become highly distributed, and distributed system paradigms such as federated and peer-to-peer databases are being adopted. In this paper, we consider the problem of inducing decision trees in a large distributed network of high dimensional databases. Our work is motivated by the existence of distributed databases in healthcare and in bioinformatics, and by the vision that these database are soon to contain large amounts of genomic data, characterized by its high dimensionality. Current decision tree algorithms would require high communication bandwidth when executed on such data, which is not likely to exist in large-scale distributed systems. We present an algorithm that sharply reduces the communication overhead by sending just a fraction of the statistical data. A fraction which is nevertheless sufficient to derive the exact same decision tree learned by a sequential learner on all the data in the network. Extensive experiments using standard synthetic SNP data show that the algorithm utilizes the high dependency among attributes, typical to genomic data, to reduce communication overhead by up to 99%. Scalability tests show that the algorithm scales well with both the size of the dataset, the dimensionality of the data, and the size of the distributed system.
Finding a small dominating set is one of the most fundamental problems of classical graph theory. In this paper, we present a new fully distributed approximation algorithm based on LP relaxation techniques. For an arb...
详细信息
ISBN:
(纸本)9781581137088
Finding a small dominating set is one of the most fundamental problems of classical graph theory. In this paper, we present a new fully distributed approximation algorithm based on LP relaxation techniques. For an arbitrary, possibly constant parameter k and maximum node degree., our algorithm computes a dominating set of expected size O (k Delta(2/k) log(Delta)vertical bar DSOPT vertical bar) in O (k(2)) rounds. Each node has to send O (k(2)Delta) messages of size O( log Delta). This is the first algorithm which achieves a non-trivial approximation ratio in a constant number of rounds.
In this paper we present a distributed algorithm for constructing efficient multicast trees in large networks. The algorithm allows the set up of effective distribution trees using a coordination protocol among the ne...
详细信息
ISBN:
(纸本)0780389328
In this paper we present a distributed algorithm for constructing efficient multicast trees in large networks. The algorithm allows the set up of effective distribution trees using a coordination protocol among the network nodes. The algorithm has been implemented and extensively tested both in simulation, and on experimental networks. Performance evaluation indicates that our approach outperforms the state-of-the-art distributed algorithms for multicast tree setup, providing good levels of convergence time and communication complexity. Furthermore, we introduce a distributed technique for dynamically updating the multicast tree. The approach monitors the accumulated damage as nodes join or leave, and it triggers a "stirring" process that locally rearranges the tree, when its degradation exceeds a given threshold.
This paper shows a new distributed algorithm for deadlock detection and resolution under the single-resource request model that highly improves the complexity measurements of previous proposals. The algorithm has a co...
详细信息
ISBN:
(纸本)076952513X
This paper shows a new distributed algorithm for deadlock detection and resolution under the single-resource request model that highly improves the complexity measurements of previous proposals. The algorithm has a communication cost of 2n - I messages and a latency of n - T for a deadlock cycle of n processes, where T is the inter-site communication delay. The algorithm achieves this improvement even verifying the strongest correctness criteria considered in previous works: it resolves all deadlocks in finite time and does not resolve false deadlocks.
The application of the tolerance paradigm to security - intrusion tolerance - has been raising a reasonable amount of attention in the dependability and security communities. In this paper we present a novel approach ...
详细信息
The application of the tolerance paradigm to security - intrusion tolerance - has been raising a reasonable amount of attention in the dependability and security communities. In this paper we present a novel approach to intrusion tolerance. The idea is to use privileged components - generically designated by wormholes - to support the execution of intrusion-tolerant protocols, often called Byzantine-resilient in the literature. The paper introduces the design of wormhole-aware intrusion-tolerant protocols using a classical distributed systems problem: consensus. The system where the consensus protocol runs is mostly asynchronous and can fail in an arbitrary way, except for the wormhole, which is secure and synchronous. Using the wormhole to execute a few critical steps, the protocol manages to have a low time complexity: in the best case, it runs in two rounds, even if some processes are malicious. The protocol also shows how often theoretical partial synchrony assumptions can be substantiated in practical distributed systems. The paper shows the significance of the TTCB as an engineering paradigm, since the protocol manages to be simple when compared with other protocols in the literature.
Accurate and low-cost sensor localization is a critical requirement for the deployment of wireless sensor networks in a wide variety of applications. In cooperative localization, sensors work together in a peer-to-pee...
详细信息
Accurate and low-cost sensor localization is a critical requirement for the deployment of wireless sensor networks in a wide variety of applications. In cooperative localization, sensors work together in a peer-to-peer manner to make measurements and then forms a map of the network. Various application requirements influence the design of sensor localization systems. In this article, the authors describe the measurement-based statistical models useful to describe time-of-arrival (TOA), angle-of-arrival (AOA), and received-signal-strength (RSS) measurements in wireless sensor networks. Wideband and ultra-wideband (UWB) measurements, and RF and acoustic media are also discussed. Using the models, the authors have shown the calculation of a Cramer-Rao bound (CRB) on the location estimation precision possible for a given set of measurements. The article briefly surveys a large and growing body of sensor localization algorithms. This article is intended to emphasize the basic statistical signal processing background necessary to understand the state-of-the-art and to make progress in the new and largely open areas of sensor network localization research.
暂无评论