Earticle

This study investigates the problem of leader-following consensus for multi-agent systems with output saturations. This study assumes that the agents are described as a neutrally stable system, and the leader agent generates the bounded trajectory within the saturation level. Then, the objective of the leader-following consensus is to track the trajectory of the leader by exchanging information with neighbors. To solve this problem, we propose an observer-based distributed consensus algorithm. Then, we provide a consensus analysis by applying the Lyapunov stability theorem and LaSalle's invariance principle. The result shows that the agents achieve the leader-following consensus in a global sense. Moreover, we can achieve the consensus by choosing any positive control gain. Finally, we perform a numerical simulation to demonstrate the validity of the proposed algorithm.

Distributed compressed sensing (DCS) states that we can recover the sparse signals from very few linear measurements. Various studies about DCS have been carried out recently. In many practical applications, there is no prior information except for standard sparsity on signals. The typical example is the sparse signals have block-sparse structures whose non-zero coefficients occurring in clusters, while the cluster pattern is usually unavailable as the prior information. To discuss this issue, a new algorithm, called backtracking-based adaptive orthogonal matching pursuit for block distributed compressed sensing (DCSBBAOMP), is proposed. In contrast to existing block methods which consider the single-channel signal reconstruction, the DCSBBAOMP resorts to the multi-channel signals reconstruction. Moreover, this algorithm is an iterative approach, which consists of forward selection and backward removal stages in each iteration. An advantage of this method is that perfect reconstruction performance can be achieved without prior information on the block-sparsity structure. Numerical experiments are provided to illustrate the desirable performance of the proposed method.

Consider a large-scale distributed system in which each computing device is observing triggers from an external source. Distributed Trigger Counting (DTC) algorithm is used to detect the state of the system when the aggregated number of the observed triggers reaches a predefined value. In this paper, we propose a simple and efficient DTC algorithm: Cascading Thresholds (CT). We mathematically show that CT is an optimal DTC algorithm in terms of the total number of exchanged messages among the devices (message complexity). For the maximum number of received messages per device (MaxRcv), CT is sub-optimal. The average message complexity of CT is , and MaxRcv of it is , where is the number of triggers to be detected, is the number of devices, and is the degree of a node in the tree-like structure. Compared to the previous optimal algorithm (TreeFill), CT is much simpler: in our implementation the code size is about 2.5 times smaller. Also, unlike TreeFill CT does not require complicated mechanisms including distributed locking. Experimental results show that CT has a lower message complexity and MaxRcv compared to the previous work (CoinRand and RingRand). Furthermore, CT and TreeFill show a similar performance. From its simplicity, CT is more practical than previous work including TreeFill, CoinRand and RingRand.

Mutual Exclusion is one of the most studied topics in distributed systems where processes communicate by asynchronous message passing. It is often necessary for multiple processes at different sites to access a shared resource or data called a critical section (CS) in distributed systems. A number of algorithms have been proposed to solve the mutual exclusion problem in distributed systems. In this paper, we propose the new algorithm which is modified from Garg's algorithm[1] thus works properly in a fault-tolerant system. In our algorithm, after electing the token generator, the elected process generates a new token based on the information of the myreqlist which is kept by every process and the reqdone which is received during election. Consequently, proposed algorithm tolerates any number of process failures and also does even when only one process is alive.

Load sharing is a critical resource in computer system. In sender-initiated load sharing algorithms, the sender continues to send unnecessary request messages for load transfer until a receiver is found while the system load is heavy. Meanwhile, in the receiver initiated load sharing algorithms, the receiver continues to send an unnecessary request message for load acquisition until a sender is found while the system load is light. These unnecessary request messages result in inefficient communications, low CPU utilization, and low system throughput in distributed systems. To solve these problems, we propose a genetic algorithm based approach for improved sender-initiated and receiver-initiated load sharing in distributed systems. And we expand this algorithm to an adaptive load sharing algorithm. Compared with the conventional sender-initiated and receiver-initiated algorithms, the proposed algorithm decreases the response time and task processing time.

A load sharing algorithm is one of the important factors in computer system. In sender-initiated load sharing algorithms, when a distributed system becomes to heavy system load, it is difficult to find a suitable receiver because most processors have additional tasks to send. The sender continues to send unnecessary request messages for load transfer until a receiver is found while the system load is heavy. Because of these unnecessary request messages it results in inefficient communications, low cpu utilization, and low system throughput. To solve these problems, we propose a self-adjusting evolutionary algorithm for improved sender-initiated load sharing in distributed systems. This algorithm decreases response time and increases acceptance rate. Compared with the conventional sender-initiated load sharing algorithms, we show that the proposed algorithm performs better.

생성트리와 같은 문제를 해결하는데 필요한 정보가 네트워크상의 프로세서에 분산되어 있는 상황에서 그들 정보를 교환하면서 그 문제를 해결하는 알고리즘을 분산알고리즘(Distributed Algorithm)이라고 한다. 생성트리와 강결합요소가 이미 구성되어 있는 비동기식 네트워크상에서 네트워크 형상이 변할 경우, 이로 인해 구성되어 있던 생성트리와 강결합요소를 갱신해야 해는 경우가 발생한다. 본 논문에서는 이러한 경우 생성트리와 강결합요소를 효율적으로 갱신하는 메시지 복잡도 O(n'log n'+ (n'+s+t)), 이상시간복잡도 O(n'log n')의 분산 알고리즘을 제안한다. 여기서 n'는 토폴로지 변화후의 네트워크의 프로세서수, s는 추가 링크수를 나타낸다. 또 t는 삭제 링크를 포함하는 강결합요소에 포함되어 있는 전체 링크수를 나타낸다.

Considers the problem to update the spanning tree and strongly-connected components in response to topology change of the network. This paper proposes a distributed algorithm that solves such a problem after several processors and links are added and deleted. Its message complexity and its ideal-time complexity are O(n'log n'+ (n'+s+t)) and O(n'logn') respectively where n'is the number of processors in the network after the topology change, s is the number of added links, and t is the total number of links in the strongly connected component (of the network before the topology change) including the deleted links.

본 논문에서는 유리재단 문제에 평균장 어닐링과 시뮬레이션된 어닐링 형태의 유전자 알고리즘을 결합한 합성 알고리즘을 분산 처리하여 적용한다. 유리재단 문제는 2차원 2진 패킹 문제로 주어진 원판에 요구되는 사각형 모양의 패턴들을 버려지는 부분이 최소가 되게 배치하는 조합 최적화 문제이다. 제안된 합성 알고리즘은 유전자 알고리즘의 다양한 연산자에 시뮬레이션된 어닐링의 온도개념을 추가하여 평균장 알고리즘에 의한 빠른 평형상태 도달을 유지하게 하였다. MPI를 이용한 분산 합성 알고리즘을 유리재단 문제에 적용하여 실험한 결과 기존의 평균장 어닐링 또는 유전자 알고리즘을 단독으로 사용하였을 때보다 최적의 배치 상태를 나타내었으며 최적해 접근 특성을 유지하면서 문제의 크기에 대하여 선형적인 수행시간 단축을 보여 주었다.

The proposed hybrid algorithm combines the benefits of rapid convergence property of mean filed annealing(MFA) and the effective genetic operations of simulated annealing-like genetic algorithm(SGA). This algorithm is applied to the isotropic material stock cutting problem, especially to glass cutting in distributed computing environments base on MPI called message passing interface. The glass cutting is to place the required rectangular patterns to the given large glass sheets resulting in reducing the wasted scrap area. Our experimental results show that the heuristic method improves the performance over the conventional ones by decreasing the scrap area and maximum execution time. It is also proved that the proposed distributed algorithm maintains the convergence properties of sequential one while it achieves almost linear speedup as the problem size increases.

상호배제 패러다임은 그룹통신, 원자적 커밋먼트 프로토콜이나 복제 데이터 관리 등의 하나의 객체에 대한 독점적인 이용이 필요한 실질적인 문제들의 해결방안을 위한 기본적인 수단으로 이용 될 수 있다. 이러한 문제는 그 동안 광범위하게 연구자들에게 연구되어 왔던 바, 그 이유는 많은 분산 프로토콜들이 상호배제 프로토콜을 필요로 하기 때문이다. 그러나 이러한 유용성에도 불구하고 아직까지 이동 컴퓨팅 환경에서 이러한 문제를 다룬 적은 별로 없었다. 본 논문에서는 이동 컴퓨팅 시스템 하에서 상호배제의 문제를 기술코자 한다. 본 논문에서 제시하는 해결방안은 토큰 기반의 알고리즘에 기반을 두게 된다.

The mutual exclusion (MX) paradigm can be used as a building block in many practical problems such as group communication, atomic commitment and replicated data management where the exclusive use of an object might be useful. The problem has been widely studied in the research community since one reason for this wide interest is that many distributed protocols need a mutual exclusion protocol. However, despite its usefulness, to our knowledge there is no work that has been devoted to this problem in a mobile computing environment. In this paper, we describe a solution to the mutual exclusion problem from mobile computing systems. This solution is based on the token-based mutual exclusion algorithm.

본 논문에서는 동기적인 분산시스템에서 불리 알고리즘에 기초하여 하나의 새로운 선출(election) 알고리즘을 설계하고 이를 분석하고자 한다. 동기적인 분산시스템에서 기존의 불리 알고리즘은 고장 추적 장치를 이용하여 더욱 효율적으로 설계되고 구현 될 수 있음을 보인다.

Tn this paper, we design and analyze an election algorithm, based on the Bully algorithm, in synchronous distributed systems. We show that the Bully algorithm, when using Failure Detector, is more effectively implemented than the classic Bully algorithm for synchronous distributed systems.

현재의 ICT 인프라(인터넷과 서버/Client 연동)는 다양한 장치, 서비스, 비즈니스 및 기술 진화에 따른 신속한 대응 에 어려움을 겪고 있다. 클라우드 컴퓨팅(Cloud Computing)은 구름 같은 네트워크 환경에서 원하는 작업을 요청하여 실행한다는 데서 기원하였으며, 인터넷 기술을 활용하여 IT 자원을 서비스로 제공하는 컴퓨팅을 뜻하고 오늘날 IT 트 렌드의 하나로 가장 주목 받고 있다. 이러한 분산클라우드 환경에서는 네트워크 및 컴퓨팅 자원에 대한 통합 관리 체계 를 통하여 관리 비용 증가 문제를 원천적으로 해결하고 분산된 마이크로 데이터센터(Micro DC(Data Center))를 통하여 코어 네트워크 트래픽 폭증 문제를 해결하여 비용 절감 효과를 높일 수 있다. 그러나 기존의 Flooding 방식은 인접한 모든 DC들에게 전송하기 때문에 많은 트래픽을 유발 할 수 있다. 이를 위해 Restricted Path Flooding 알고리즘이 제안 되었으나 대규모 네트워크에서는 여전히 트래픽을 발생할 수 있는 단점이 있어서 본 논문에서는 홉수 제한을 통하여 이를 개선한 Lightweight Path Flooding 알고리즘을 제안하였다.

Current ICT(Information & Communication Technology) infrastructures (Internet and server/client communicatio n) are struggling for a wide variety of devices, services, and business and technology evolution. Cloud computing or iginated simply to request and execute the desired operation from the network of clouds. It means that an IT resour ce that provides a service using the Internet technology. It is getting the most attention in today's IT trends. In the distributed cloud environments, management costs for the network and computing resources are solved fundamentall y through the integrated management system. It can increase the cost savings to solve the traffic explosion problem of core network via a distributed Micro DC. However, traditional flooding methods may cause a lot of traffic due to transfer to all the neighbor DCs. Restricted Path Flooding algorithms have been proposed for this purpose. In large networks, there is still the disadvantage that may occur traffic. In this paper, we developed Lightweight Path Floodi ng algorithm to improve existing flooding algorithm using hop count restriction..

분산 스트림 처리 시스템을 위한 고가용성 알고리즘에는 Passive Standby, Active Standby, Upstream Backup 알고리즘 등이 있다. 기존의 고가용성 알고리즘은 복구 시 필요한 데이터의 백업을 위한 bandwidth overhead가 크며 노드의 연산결과를 다수의 downstream 노드들이 공유하며 데이터의 유입률이 폭발적으로 증가하는 경우에 출력 큐의 오버플로우로 인한 데이터의 손실 문제가 발생 할 수 있다. 본 논문은 이러한 문제들을 해결하기 위해 데이터 스트림의 유입량과 노드들의 연산처리율 모니터링을 통해 백업 방법을 유동적으로 변경시키는 적응적 Upstream Backup 알고리즘을 제안한다.

There have been High-Availability algorithms for Distributed Stream Processing System such as Passive Standby, Active Standby and Upstream Backup. Existing High Availability Algorithms have high bandwidth overhead when they backup data needed for restoring from the system failure and data can be lost by overflow of output queue in case of explosive increasing of input data rates. In this paper, we suggest adaptive Upstream Backup algorithm which changes fluidly the way to backup using monitor input rates of data stream and throughput of operation to solve those problems.

하둡 맵리듀스와 같은 분산 컴퓨팅 플랫폼이 개발됨에 따라, 기존 단일 컴퓨터 상에서 수행되는 질의 처리 기법을 분산 컴퓨팅 환경에서 효율적으로 수행하는 것이 필요하다. 특히, 주어진 두 데이터 집합에서 유사도가 높은 모든 데이터 쌍을 탐색하는 유사 조인 질의를 분산 컴퓨팅 환경에서 수행하려는 연구가 있어 왔다. 그러나 분산 병렬 환경에서의 기존 유사 조인 질의처리 기법은 데이터 전송 비용만을 고려하기 때문에 클러스터 간에 비균등 연산 부하 분산의 문제점이 존재한다. 본 논문에서는 분산 컴퓨팅 환경에서 효율적인 유사 조인 처리를 위한 행렬 기반 부하 분산 알고리즘을 제안한다. 제안하는 알고리즘은 클러스터의 균등 부하 분산을 위해 행렬을 이용하여 예상되는 연산 부하를 측정하고 이에 따라 파티션을 생성한다. 아울러, 클러스터에서 질의 처리에 사용되지 않는 데이터를 필터링함으로서 연산 부하를 감소시킨다. 마지막으로 성능 평가를 통해 제안하는 알고리즘이 기존 기법에 비해 질의 처리 성능 측면에서 우수함을 보인다.

As distributed computing platforms like Hadoop MapReduce have been developed, it is necessary to perform the conventional query processing techniques, which have been executed in a single computing machine, in distributed computing environments efficiently. Especially, studies on similarity join query processing in distributed computing environments have been done where similarity join means retrieving all data pairs with high similarity between given two data sets. But the existing similarity join query processing schemes for distributed computing environments have a problem of skewed computing load balance between clusters because they consider only the data transmission cost. In this paper, we propose Matrix-based Load-balancing Algorithm for efficient similarity join query processing in distributed computing environment. In order to uniform load balancing of clusters, the proposed algorithm estimates expected computing cost by using matrix and generates partitions based on the estimated cost. In addition, it can reduce computing loads by filtering out data which are not used in query processing in clusters. Finally, it is shown from our performance evaluation that the proposed algorithm is better on query processing performance than the existing one.

In this paper, we propose a joint distributed rate control algorithm and distributed resource allocation algorithm in multiradio multichannel multihop wireless networks with network coding. Our formulation provides a method of finding active links from the source to the destination and utilizing the resources from the available ones in order to maximize the network utility. Distributed schemes are used to approximate the upper bound within the feasible solution region of the optimization problem. The performance evaluation results show that the proposed scheme can effectively exploits multiple radios and channels.

Aiming at the inherent deficiencies of large-scale wireless sensor network for distributed positioning algorithm, the DV-hop algorithm is mainly studied and improved, and two kinds of improved algorithms DV-hop and DV-hop+Lastdist DV- hop-MinMax are proposed. The DV- hop+Lastdist improves the positioning accuracy of the node by only adding a message variable, while the DV-hop-MinMax effectively reduces the costs of floating-point operations and power consumption, and can select the appropriate algorithm according to actual application environment. Finally, the paper gives the method to realize OMNeT++ in the simulation environment, and carries out the simulation analysis.

In view of the operating mode of wireless sensor networks, the node initial probability and channel listening, a distributed clustering algorithm DCOS based on the optimal search model is proposed, along with the corresponding mathematical optimization model. The goal of model optimization is to find a feasible allocation scheme for search resource so that the probability of successful communication reaches its maximum under the constrained condition for search resource. Clustering algorithm is designed to balance the network energy consumption, thereby maximizing network lifetime. In the clustering algorithm, the cluster head is generated in the comprehensive consideration of the minimum number of nodes in the overlay network, and the node residual energy and the number of neighboring nodes, during the operation of the network, setting the maximum wait time for cluster heads may avoid the their energy consumption. Analysis and simulation results show that, compared to several important clustering algorithms, the performance of DCOS is proven to be superior in balancing node energy consumption and prolonging the network lifetime.

Satellite network, especially low earth orbit (LEO) satellite network, which has advantages of global coverage and short round-trip time (RTT), has played an increasingly important role in the future generation of global communication network. Due to fast velocity, frequent handover and time-varying topology, designing effective routing algorithm for LEO satellite network has always been challenging. An Agent-based distributed Routing Algorithm with Traffic Prediction (TPARA) is proposed in this paper. TPARA algorithm, which fully takes traffic density of the surface of the Earth into account, consists of two parts: traffic prediction and routing decision. In the former, not only the heterogeneity of traffic on the ground is considered, but upcoming traffic from terrestrial terminals is predicted using an improved Kalman filtering as well. In the latter, mobile agents (MAs) are used to explore satellite network and collect routing information. The ultimate routing decision is determined by both current status and future status of satellite network. Simulation results show that, compared with the classical algorithm of ACO, TPARA not only has shorter delay, but alleviates congestion as well.

In recent years, distributed computing technology has been one of the cutting edge technologies for its low power and cost, which makes numerous IT organizations extend their hands in order to improve their financial ability. Because of these new features, grid computing, the original task scheduling mechanism, can’t work effectively in distributed computing environments, hence, we need a new task scheduling method to solve the problems. With considering the complex characters of the task in different distributed computing applications, firstly, we construct a more comprehensive task scheduling model, which has three sub objective functions. Secondly, we present an improved genetic algorithm to solve the multi- objective NP-hard problem. Finally, we implement some simulation experiments, and the evaluation results show us that the proposed model and improved GA are efficient and effective. The first part is the research status and related problems. The second part is the establishment of system architecture and task scheduling model. The last part is the experimental analysis and conclusion.

Clustering problem is one of the significant issues for wireless sensor networks concerned with energy consumption and large-scale deployment. Several energy-efficient clustering algorithms have been proposed to improve the energy utilization efficiency and prolong the network lifetime. In this paper, we propose a new clustering scheme after a comprehensive analysis on existing protocols. In our algorithm, named WPCA (Weighted Probabilistic Clustering Algorithm), every node independently decides whether to be a cluster head according to a weighted probability, which is related to the ratio between node’s residual energy and average remaining energy. The nodes with more residual energy are assigned larger weight value to further increase the chances to be elected as cluster heads. In addition, the rotation procedure of cluster heads in previous algorithms is totally abandoned. Simulation results show that WPCA achieves longer lifetime than previous probabilistic-based clustering algorithms and gets a very close approximation compared with a deterministic clustering method.

Localization of sensors nodes is a key and fundamental issue in wireless sensor networks due to random deployment. In this paper, we propose a tree based clustering (TBC) multidimensional scaling algorithm for wireless sensor networks with the purpose of overcoming the shortage of classical MDS algorithms in its localization accuracy and computing complexity. Clustering is adopted to degrade the problem scale in our approach and a moderate number of common nodes between clusters are kept during clustering. Inner cluster local coordinates are calculated and then mapped into global coordinates according to the tree structure formed by clustering. Simulations on MATLAB are conducted and the results show that the proposed algorithm has better localization coverage and higher accuracy than the traditional MDS based algorithms.