The recent growing size of datasets requires scalability of data mining algorithms, such as clustering algorithms. The MapReduce programing model provides the scalability needed, alongside with portability as well as ...
详细信息
The recent growing size of datasets requires scalability of data mining algorithms, such as clustering algorithms. The MapReduce programing model provides the scalability needed, alongside with portability as well as automatic data safety and management. k-means is one of the most popular algorithms in data mining and can be easily adapted to the MapReduce model. Nevertheless, k-means has drawbacks, such as the need to provide the number of clusters (k) in advance and the sensitivity of the algorithm to the initial cluster prototypes. This paper presents two evolutionary scalable metaheuristics in MapReduce that automatically seek the solution with the optimal number of clusters and best clustering structure for scalable datasets. The first consists in an algorithm able to iteratively enhance k-means clusterings through evolutionary operators designed to handle distributed data. The second consists in applying evolutionary k-means to cluster each distributed portion of a dataset in an independent way, combining the obtained results into an ensemble afterwards. The proposed techniques are compared asymptotically and experimentally with other state-of-the-art clustering algorithms also developed in MapReduce. The results are analyzed by statistical tests and show that the first proposed metaheuristic yielded results with the best quality, while the second achieved the best computing times. (C) 2017 Elsevier B.V. All rights reserved.
An effective and robust time synchronization scheme is essential for many wireless sensor network (WSN) applications. Conventional synchronization methods assume the use of highly accurate crystal oscillators (10-100 ...
详细信息
An effective and robust time synchronization scheme is essential for many wireless sensor network (WSN) applications. Conventional synchronization methods assume the use of highly accurate crystal oscillators (10-100 ppm) for synchronization, only correcting for small errors. This paper suggests a novel method for time synchronization in a multihop, fully-distributed WSN using imprecise CMOS oscillators (up to 15 000 ppm). The DiStiNCT technique is power-efficient, computationally simple, and robust to packet loss and complex topologies. Effectiveness has been demonstrated in simulations of fully connected, grid, and unidirectional ring topologies. The method has been validated in hardware on a grid of nine sensor nodes, synchronizing to within a mean error of 6.6 ms after 40 iterations.
Motivated by both distributed computation and decentralized control applications, we studied the distributed linear iterative algorithms with memory. Specifically, we showed that the system of linear equations Gx = b ...
详细信息
Motivated by both distributed computation and decentralized control applications, we studied the distributed linear iterative algorithms with memory. Specifically, we showed that the system of linear equations Gx = b b can be solved through a distributed linear iteration for arbitrary invertible G using only a single memory element at each processor. Further, we demonstrated that the memoried distributed algorithm can be designed to achieve much faster convergence than a memoryless distributed algorithm. Two small simulation examples were included to illustrate the results. Copyright (c) 2011 John Wiley & Sons, Ltd.
Energy harvesting is a key technology to enable long-term wireless sensor network applications. In the case of multi-hop networks, each node both performs measurements to produce data to be sent to a sink, and relays ...
详细信息
Energy harvesting is a key technology to enable long-term wireless sensor network applications. In the case of multi-hop networks, each node both performs measurements to produce data to be sent to a sink, and relays data packets from other nodes. In this letter, we propose a distributed algorithm for computation of fair packet rates for multi-hop energy harvesting wireless sensor networks. The packet rate computation problem is formulated as a convex optimization problem, and using the fast alternating direction method of multipliers, the original problem is decomposed into smaller subproblems that can be solved in parallel. Simulations using real indoor light energy traces show that the algorithm computes high accuracy solutions, even with a low median number of iterations (10 or less). By setting the stop criteria parameter, a compromise can be set between the accuracy of the solution and the number of iterations required.
Wireless sensor networks (WSNs) can provide numerous benefits in industrial automation. By removing the cable infrastructure, the wireless architecture enables the possibility for nodes in a network to dynamically and...
详细信息
Wireless sensor networks (WSNs) can provide numerous benefits in industrial automation. By removing the cable infrastructure, the wireless architecture enables the possibility for nodes in a network to dynamically and autonomously group into clusters according to the communication features and the data they collect. This capability allows to leverage the flexibility and robustness of industrial WSNs in supervisory intelligent systems for high-level tasks, such as, for example, environmental sensing, condition monitoring, and process automation. In this paper, a clustering strategy is studied that partitions a sensor network into a nonfixed number of nonoverlapping clusters according to the communication network topology and measurements distribution: To this aim, both a centralized and a distributed algorithm are designed that do not require a cluster-head structure or other network assumptions. As a validation, these strategies are tested on a real dataset coming from a structured environment and the effectiveness of the clustering procedure is also investigated to perform anomalies detection in an industrial production process.
In this letter, we derive distributed synchronous and asynchronous algorithms for computing quantiles of the agents' local values. These algorithms are based on the formulation of a suitable problem, explicitly so...
详细信息
In this letter, we derive distributed synchronous and asynchronous algorithms for computing quantiles of the agents' local values. These algorithms are based on the formulation of a suitable problem, explicitly solvable by the alternating direction method of multipliers (ADMM), and recent randomized optimization methods.
Motivated by various applications in wireless sensor networks and edge computing, we study distributed optimization problems over a network of nodes, where the goal is to optimize a global objective function composed ...
详细信息
ISBN:
(纸本)9781538665961
Motivated by various applications in wireless sensor networks and edge computing, we study distributed optimization problems over a network of nodes, where the goal is to optimize a global objective function composed of a sum of local functions. In these problems, due to the large scale of the network, both computation and communication must be implemented locally resulting in the need for distributed algorithms. In addition, the algorithms should be efficient enough to tolerate the limitation of computing resources, memory capacity, and communication bandwidth shared between the nodes. To cope with such limitations, we consider in this paper distributed subgradient methods under quantization. Our main contribution is to provide a sufficient condition for the sequence of quantization levels, which guarantees the convergence of distributed subgradient methods. Our results, while complementing existing results, suggest that distributed subgradient methods achieve desired convergence properties even under quantization, as long as the quantization levels become finer and finer with a proper rate. We also provide numerical simulations to compare the convergence properties of such methods with and without quantization for solving the wellknown least square problems over networks.
We devise new algorithms for the single-source shortest paths (SSSP) problem with non-negative edge weights in the CONGEST model of distributed computing. While close-to-optimal solutions, in terms of the number of ro...
详细信息
ISBN:
(纸本)9781538642306
We devise new algorithms for the single-source shortest paths (SSSP) problem with non-negative edge weights in the CONGEST model of distributed computing. While close-to-optimal solutions, in terms of the number of rounds spent by the algorithm, have recently been developed for computing SSSP approximately, the fastest known exact algorithms are still far away from matching the lower bound of (Omega) over tilde(root n+ D) rounds by Peleg and Rubinovich [SIAM Journal on Computing 2000], where n is the number of nodes in the network and D is its diameter. The state of the art is Elkin's randomized algorithm [STOC 2017] that performs (O) over tilde (n(2/3)D(1/3) + n(5/6)) rounds. We significantly improve upon this upper bound with our two new randomized algorithms for polynomially bounded integer edge weights, the first performing (O) over tilde(root nD) rounds and the second performing (O) over tilde(root nD(1/4) + n(3/5) + D) rounds. Our bounds also compare favorably to the independent result by Ghaffari and Li [STOC 2018]. As side results, we obtain a (1 + epsilon)-approximation (O) over tilde((root nD(1/4) + D)/epsilon)-round algorithm for directed SSSP and a new work/depth trade-off for exact SSSP on directed graphs in the PRAM model.
The dynamic spectrum access (DSA) algorithms aim to maximize network throughput by ensuring orthogonal channel allocation among secondary users (SUs) in cognitive radio network (CRN). The DSA is challenging in the inf...
详细信息
ISBN:
(纸本)9781538617342
The dynamic spectrum access (DSA) algorithms aim to maximize network throughput by ensuring orthogonal channel allocation among secondary users (SUs) in cognitive radio network (CRN). The DSA is challenging in the infrastructure-less CRN (ICRN) due to lack of coordination among SUs. Most of the existing algorithms assume prior knowledge of the number of active SUs to guarantee orthogonalization. The musical chair (MC) algorithm is the state-of-the-art algorithm where the number of SUs are unknown and estimated independently at each SU. However, MC algorithm incurs a significant number of SU collisions and hence, the aggregate throughput is poor compared to centralized algorithms. In this paper, we setup DSA in ICRN as a multi-player multi-armed Bandit problem and develop distributed algorithm which allows SUs to orthogonalize in optimal channels with a negligible number of collisions. We also develop realistic USRP based testbed to validate the performance of the DSA algorithms in the real radio environment. Experimental results validate the superiority of the proposed algorithm over existing algorithms in terms of network throughput and fairness in channel access. Furthermore, a fewer number of collisions and hence, fewer retransmissions lead to efficient use of battery power, spectrum and time.
This paper performs further improvement to a distributed algorithm for solving linear algebraic equations via multi-agent networks recently developed by Mou et al., in which all agents' states converge exponential...
详细信息
This paper performs further improvement to a distributed algorithm for solving linear algebraic equations via multi-agent networks recently developed by Mou et al., in which all agents' states converge exponentially fast to the same solution to a group of linear equations by assuming each agent knows only part of the linear equations and its nearby neighbors' states. We first prove that the algorithm proposed by Mou et al. with special initialization is able to achieve the solution that is closest to a given point in the Euclidean distance. Second, we eliminate the required initialization step used by Mou et al. by a modification to the update equation. Both analytical and numerical results are provided for validation.
暂无评论