Orchestration systems in cloud platforms are responsible for creating, managing and assigning the computational and network bandwidth resources to the requesting services. Conventional orchestration approaches in data...
详细信息
ISBN:
(纸本)9783903176157
Orchestration systems in cloud platforms are responsible for creating, managing and assigning the computational and network bandwidth resources to the requesting services. Conventional orchestration approaches in data centers are based on centralized solutions where they are a single point of failure, and a potential performance bottleneck. In this paper, using the notions of Markov approximation method and auction theory, we propose a fully distributed resource management scheme for data centers. The proposed solution takes into account the operational and economic constraints of the services and the servers in the data center and maximizes a global system utility function in a fully distributed manner. Simulation results show the effectiveness of the proposed solution in terms of speed of convergence, accuracy and resource utilization for applicability in next generation cloud systems.
In this paper, a distributed optimization problem is investigated via input feedforward passivity. First, an input-feedforward-passivity-based continuous-time distributed algorithm is proposed. It is shown that the er...
详细信息
ISBN:
(数字)9781728113982
ISBN:
(纸本)9781728113982
In this paper, a distributed optimization problem is investigated via input feedforward passivity. First, an input-feedforward-passivity-based continuous-time distributed algorithm is proposed. It is shown that the error system of the proposed algorithm can be interpreted as output feedback interconnections of a group of Input Feedforward Passive (IFP) systems. Second, based on this IFP framework, the distributed algorithm is studied over weight-balanced directed and uniformly jointly strongly connected switching topologies. Specifically, the continuous-time distributed algorithm for uniformly jointly strongly connected digraphs has never been considered before. Sufficient convergence conditions are derived for the design of a suitable coupling gain.
Percolation analysis is a valuable tool to study the statistical properties of turbulent flows. It is based on computing the percolation function for a derived scalar field, thereby quantifying the relative volume of ...
详细信息
ISBN:
(纸本)9781728126050
Percolation analysis is a valuable tool to study the statistical properties of turbulent flows. It is based on computing the percolation function for a derived scalar field, thereby quantifying the relative volume of the largest connected component in a superlevel set for a decreasing threshold. We propose a novel memory-distributed parallel algorithm to finely sample the percolation function. It is based on a parallel version of the union-find algorithm interleaved with a global synchronization step for each threshold sample. The efficiency of this algorithm stems from the fact that operations in-between threshold samples can be freely reordered, are mostly local and thus require no inter-process communication. Our algorithm is significantly faster than previous algorithms for this purpose, and is neither constrained by memory size nor number of compute nodes compared to the conceptually related algorithm for extracting augmented merge trees. This makes percolation analysis much more accessible in a large range of scenarios. We explore the scaling of our algorithm for different data sizes, number of samples and number of MPI processes. We demonstrate the utility of percolation analysis using large turbulent flow data sets.
We design distributed algorithms to compute approximate solutions for several related graph optimization problems. All our algorithms have round complexity being logarithmic in the number of nodes of the underlying gr...
详细信息
ISBN:
(纸本)9781728112466
We design distributed algorithms to compute approximate solutions for several related graph optimization problems. All our algorithms have round complexity being logarithmic in the number of nodes of the underlying graph and in particular independent of the graph diameter. By using a primal-dual approach, we develop a 2(1 + epsilon)-approximation algorithm for computing the coreness values of the nodes in the underlying graph, as well as a 2(1 + epsilon)-approximation algorithm for the min-max edge orientation problem, where the goal is to orient the edges so as to minimize the maximum weighted in-degree. We provide lower bounds showing that the aforementioned algorithms are tight both in terms of the approximation guarantee and the round complexity. Finally, motivated by the fact that the densest subset problem has an inherent dependency on the diameter of the graph, we study a weaker version that does not suffer from the same limitation.
There is a dearth of research in developiiig low-cost solutions for distributed decision making in loT networks. Most studies in the literature require the deployment of additional sensors for data collection. In this...
详细信息
ISBN:
(纸本)9781728155845
There is a dearth of research in developiiig low-cost solutions for distributed decision making in loT networks. Most studies in the literature require the deployment of additional sensors for data collection. In this study, we propose to leverage available sensors built-in smartphones, to properly collect and broadcast data for different decision-making purposes in smart cities infrastructures, for example, intelligent transportation net. works, smart health services, security and emergencies, industrial control, smart agriculture, home automation and so on. To this end, we first introduce our new platform (including software and mobile app implementation) to identify available sensors at each end-user device. We have identified a wide range of sensors including gyroscope, ambient light sensor, temperature, magnetic field sensor, orientation sensor, game rotation vector, linear acceleration, relative humidity, gravity, geomagnetic rotation vector, etc. As the sensors are already integrated within the phone, therefore, using these sensors can be beneficial considering the complexity, efficiency, and cost of the overall system. The challenge is to design a system that can trigger distributed devices to be self-activated and agreed to generate all available sensors data. Besides, as devices can send a continuous stream of data, therefore, size of data could be mounted and could be in the haphazard structure, which would give us hurdles to identify a device sensor data from another and to make an intelligent decision. To tackle all of these, we propose a distributed sensing approach that is capable to identify a device using token, can activate distributed end-user devices to send data to the cloud whenever it requires and store data in the cloud server maintaithng proper format. This approach enables remote data collection leveraging available end-user devices and reduces the cost of installing new sensors for autonomous loT applications. We then build on our efficien
Consider a network of N sensors that collect data. Each sensor transmits this data to one out of G gateways. The gateways perform the preprocessing of the data and store it in the cloud. In order not to lose data, sen...
详细信息
ISBN:
(纸本)9781728144450
Consider a network of N sensors that collect data. Each sensor transmits this data to one out of G gateways. The gateways perform the preprocessing of the data and store it in the cloud. In order not to lose data, sensors need to route their packets to the gateway that has higher chances of successfully receiving them. However, this might overload some gateways that are the best gateway for many sensors, forcing them to drop packets. We design an asynchronous distributed algorithm where each sensor randomizes its destination, taking into account both the transmission success probabilities and the average input data rate of each gateway. Sensors estimate the transmission success probabilities online by sending pilot sequences and gateways broadcast the input data rates to all sensors. We show that our algorithm converges to a close to optimal solution. Specifically, when the number of sensors N approaches infinity, the ratio of the total throughput of our algorithm to the optimal throughput converges in probability to one.
We study linear programming and general LP-type problems in several big data (streaming and distributed) models. We mainly focus on low dimensional problems in which the number of constraints is much larger than the n...
详细信息
ISBN:
(纸本)9781450362276
We study linear programming and general LP-type problems in several big data (streaming and distributed) models. We mainly focus on low dimensional problems in which the number of constraints is much larger than the number of variables. Low dimensional LP-type problems appear frequently in various machine learning tasks such as robust regression, support vector machines, and core vector machines. As supporting large-scale machine learning queries in database systems has become an important direction for database research, obtaining efficient algorithms for low dimensional LP-type problems on massive datasets is of great value. In this paper we give both upper and lower bounds for LP-type problems in distributed and streaming models. Our bounds are almost tight when the dimensionality of the problem is a fixed constant.
We consider a microgrid that consists of N providers and B consumers. Each provider has a certain supply and each consumer has a certain demand. The efficiency of transmitting energy between providers and consumers is...
详细信息
ISBN:
(纸本)9781538680889
We consider a microgrid that consists of N providers and B consumers. Each provider has a certain supply and each consumer has a certain demand. The efficiency of transmitting energy between providers and consumers is modeled using a bipartite graph G. Our goal is to maximize the amount of utilized energy using a distributed algorithm that each provider runs locally. We propose a non-cooperative energy allocation game, and adopt the best-response dynamics for this game as our distributed algorithm. We prove that the best-response dynamics converge in no more than N steps to one of at most N! pure Nash equilibria of our game. Despite the fact that some of these Nash equilibria are suboptimal, we are able to prove that our algorithm achieves near-optimal performance in "almost all" games. We do so by analyzing the best-response dynamics in a random game, where the network is generated using a random model for the graph G. We prove that the ratio between the utilized energy of our algorithm and that of the optimal solution converges to one in probability as B increases (and N is any function of B). Using numerical simulations, we demonstrate that our asymptotic analysis is valid even for B = 10 consumers.
In this paper, the coordinated operation of agents representing residential household energy management systems in the electricity market is considered. Each agent has its local decision vector that may contain both c...
详细信息
ISBN:
(纸本)9781728111568
In this paper, the coordinated operation of agents representing residential household energy management systems in the electricity market is considered. Each agent has its local decision vector that may contain both continuous and discrete variables. The problem of minimizing the energy procurement cost of the community of agents is represented by a Mixed Integer Linear Program with local and global constraints. A dual decomposition method with a tightening of the global constraint is applied to solve the problem with guarantees for the feasibility of the obtained solutions. The problem is decomposed and solved in a distributed way with coordination by a community coordinator. The method is applied to a realistic case study of 15 households.
In this paper, we presents a fixed-time convergent distributed algorithm to achieve least square solutions of networked linear equations. Each agent in the network only knows a subset of the equations and can only exc...
详细信息
ISBN:
(纸本)9781728101071;9781728101064
In this paper, we presents a fixed-time convergent distributed algorithm to achieve least square solutions of networked linear equations. Each agent in the network only knows a subset of the equations and can only exchange messages with its nearest neighbors. Compared with the finite-time algorithm, it is shown that the settling time is independent to the initial states of the algorithm and can be preassign according to requirements of the task. A numerical example is provided to illustrate the effectiveness of the theoretical result.
暂无评论