Laurence Pilard

Blocking vs. non-blocking coordinated checkpointing for large-scale fault tolerant MPI

A long-term trend in high-performance computing is the increasing number of nodes in parallel computing platforms, which entails a higher failure probability. Fault programming environments should be used to guarantee the safe execution of critical applications. Research in fault tolerant MPI has led to the development of several fault tolerant MPI environments. Different approaches are being proposed using a variety of fault tolerant message passing protocols based on coordinated checkpointing or message logging. The most popular approach is with coordinated checkpointing. In the literature, two different concepts of coordinated checkpointing have been proposed: blocking and non-blocking. However they have never been compared quantitatively and their respective scalability remains unknown. The contribution of this paper is to provide the first comparison between these two approaches and a study of their scalability. We have implemented the two approaches within the MPICH environments and evaluate their performance using the NAS parallel benchmarks

Blocking vs. non-blocking coordinated checkpointing for large-scale fault tolerant MPI Protocols

A long-term trend in high-performance computing is the increasing number of nodes in parallel computing platforms, which entails a higher failure probability. Fault tolerant programming environments should be used to guarantee the safe execution of critical applications. Research in fault tolerant MPIs has led to the development of several fault tolerant MPI environments. Different approaches are being proposed using a variety of fault tolerant message passing protocols based on coordinated checkpointing or message logging. The most popular approach is with coordinated checkpointing. In the literature, two different concepts of coordinated checkpointing have been proposed: blocking and non-blocking. However they have never been compared quantitatively, and their respective scalabilities remain unknown. The contribution of this paper is to provide the first comparison between these two approaches and a study of their scalabilities. We have implemented the two approaches within the MPICH environments and evaluate their performance using the NAS parallel benchmarks.

Self-stabilizing algorithm for efficient topology control in Wireless Sensor Networks

Abstract Wireless Sensor Networks lifetime mainly depends on energy saving efficiency. In this paper, we propose an energy-efficient self-stabilizing topology control protocol for WSN. We reduce the transmission power of each node so as to maintain network connectivity while saving maximum energy. Besides, we propose an approximation algorithm for minimum weighted connected dominating set that builds a virtual backbone formed by sensors with maximum energy. This backbone is used for efficient routing purpose. We proved the algorithm correctness and through our simulation results, we showed the efficiency of our proposed solution.

A Model for Large Scale Self-Stabilization

We introduce a new model for distributed algorithms designed for large scale systems that need a low-overhead solution to allow the processes to communicate with each other. We assume that every process can communicate with any other process provided it knows its identifier, which is usually the case in e.g. a peer to peer system, and that nodes may arrive or leave at any time. To cope with the large number of processes, we limit the memory usage of each process to a small constant number of variables, combining this with previous results concerning failure detectors and resource discovery. We illustrate the model with a self-stabilizing algorithm that builds and maintains a spanning tree topology. We provide a formal proof of the algorithm and the results of experiments on a cluster.

Temporal partition in sensor networks

Sensor networks are composed of nodes embedded in physical environments where applications may be tasked to run for years without human maintenance and without continuous external power supply. Strategies for power conservation are thus important in sensor network protocols and system architecture. One such strategy is to arrange node sleeping schedules so that radios are powered off until communication is necessary. Nodes cannot receive messages during periods when the radio is turned off. In this setting, there can arise situations where groups of network nodes have somehow become temporally partitioned: due to having different sleeping schedules, groups of nodes could be unaware of each other. The paper presents several self-stabilizing protocols to solve the problem of temporal partition; starting from an arbitrary temporally partitioned state, these protocols lead the network to a state in which all nodes have a perfectly aligned sleep schedule. Our techniques include using randomly chosen relatively prime sleep periods and occasional, and possibly random, probing of extra time slots. Our protocols aim for fast convergence while imposing only a small energy consumption overhead.

Brief announcement: self-stabilizing spanning tree algorithm for large scale systems

We introduce a self-stabilizing algorithm that builds and maintains a spanning tree topology on any large scale system. We assume that the existing topology is a complete graph and that nodes may arrive or leave at any time. To cope with the large number of processes of a grid or a peer to peer system, we limit the memory usage of each process to a small constant number of variables, combining this with previous results concerning failure detectors and resource discovery.

A distributed clustering algorithm for large-scale dynamic networks

We propose an algorithm that builds and maintains clusters over a network subject to mobility. This algorithm is fully decentralized and makes all the different clusters grow concurrently. The algorithm uses circulating tokens that collect data and move according to a random walk traversal scheme. Their task consists in (i) creating a cluster with the nodes it discovers and (ii) managing the cluster expansion; all decisions affecting the cluster are taken only by a node that owns the token. The size of each cluster is maintained higher than m nodes (m is a parameter of the algorithm). The obtained clustering is locally optimal in the sense that, with only a local view of each clusters, it computes the largest possible number of clusters (i.e. the sizes of the clusters are as close to m as possible). This algorithm is designed as a decentralized control algorithm for large scale networks and is mobility-adaptive: after a series of topological changes, the algorithm converges to a clustering. This recomputation only affects nodes in clusters where topological changes happened, and in adjacent clusters.

Self-stabilizing algorithm for energy saving in Wireless Sensor Networks

Wireless Sensor Networks lifetime mainly depends on energy saving efficiency. In this paper, we propose an energy-efficient self-stabilizing topology control protocol for WSN. We reduce the transmission power of each node so as to maintain network connectivity while saving maximum energy. Besides, we propose an approximation algorithm for minimum weighted connected dominating set that builds a virtual backbone formed by sensors with maximum energy. This backbone is used for efficient routing purpose. Through our simulation results, we show the efficiency of our proposed algorithm.

A self-stabilizing 23-approximation algorithm for the maximum matching problem

The matching problem asks for a large set of disjoint edges in a graph. It is a problem that has received considerable attention in both the sequential and the self-stabilizing literature. Previous work has resulted in self-stabilizing algorithms for computing a maximal (12-approximation) matching in a general graph, as well as computing a 23-approximation on more specific graph types. In this paper, we present the first self-stabilizing algorithm for finding a 23-approximation to the maximum matching problem in a general graph. We show that our new algorithm, when run under a distributed adversarial daemon, stabilizes after at most O(n2) rounds. However, it might still use an exponential number of time steps.

A Self-stabilizing

The matching problem asks for a large set of disjoint edges in a graph. It is a problem that has received considerable attention in both the sequential and self-stabilizing literature. Previous work has resulted in self-stabilizing algorithms for computing a maximal (