This paper proposes a delay-sensitive communication approach based on distributed processing for real-time applications that provide interactive services for multiple users in order to minimize the delay considering b...
详细信息
This paper proposes a delay-sensitive communication approach based on distributed processing for real-time applications that provide interactive services for multiple users in order to minimize the delay considering both admissible delay and delay variation rate. The proposed approach considers two scenarios, namely, simultaneous participation and successive participation. In the simultaneous participation, all users and servers are given, and the application is processed in different distributed servers;a user accesses a suitable server as a solution of the server selection problem. In the successive participation, where all servers are given, different users will be participated sequentially in a greedy manner with variation of time, while executing the currently applications. We formulate an integer linear programming (ILP) problem in the simultaneous participation scenario for the distributed server selection when all users and servers are given considering the parameter of admissible delay and delay-variation rate. We prove that the distributed server selection problem is NP-complete. By using a high-performance optimization solver, we solve the introduced ILP problem within a practical time for 800 users. We provide a method for the successive participation scenario by utilizing the ILP formulated in the simultaneous participation. Numerical results indicate that the proposed delay-sensitive communication approach based on distributed processing outperforms the conventional centralized processing approach in terms of delay.
Video synopsis is one of the popular research topics in the field of digital video and has broad application prospects. Current research of it focuses on the methods of generating video synopsis or studying to utilize...
详细信息
Video synopsis is one of the popular research topics in the field of digital video and has broad application prospects. Current research of it focuses on the methods of generating video synopsis or studying to utilize optimization algorithms such as fuzzy theory, minimum sparse reconstruction, and genetic algorithm to optimize its computing steps. This paper mainly studies the object-based video synopsis technology in distributed environment. We propose an effective video synopsis algorithm and a distributed processing model to accelerate the computing speed of video synopsis. The algorithm is proposed for studies of surveillance videos, which focuses on several key algorithmic steps, for instance, initialization of original video resources, background modeling, moving object detecting, and nonlinear rearrangement. These steps can be performed in parallel. In order to obtain good video synopsis effect and fast computing speed, some optimization methods are applied to these steps. With the aim of employing much more computing resources, we propose a distributed processing model, which splits the original video file into multiple segments and distributes them to different computing nodes to improve the computing performance by leveraging the multi-core and multi-thread capabilities of CPU. Experimental results show that the proposed distributed model can significantly improve the computing speed of video synopsis.
Parallel distributed processing (PDP) models in psychology are the precursors of deep networks used in computer science. However, only PDP models are associated with two core psychological claims, namely that all know...
详细信息
Parallel distributed processing (PDP) models in psychology are the precursors of deep networks used in computer science. However, only PDP models are associated with two core psychological claims, namely that all knowledge is coded in a distributed format and cognition is mediated by non-symbolic computations. These claims have long been debated in cognitive science, and recent work with deep networks speaks to this debate. Specifically, single-unit recordings show that deep networks learn units that respond selectively to meaningful categories, and researchers are finding that deep networks need to be supplemented with symbolic systems to perform some tasks. Given the close links between PDP and deep networks, it is surprising that research with deep networks is challenging PDP theory.
Efficient distributed processing is vital for collaborative searching tasks of robotic swarm systems. Typically, those systems are decentralized, and the members have only limited communication and processing capaciti...
详细信息
Efficient distributed processing is vital for collaborative searching tasks of robotic swarm systems. Typically, those systems are decentralized, and the members have only limited communication and processing capacities. What is illustrated in this paper is a distributed processing paradigm for robotic swarms moving in a line or v-shape formation. The introduced concept is capable of exploits the line and v-shape formations for 2-D filtering and processing algorithms based on a modified multi-dimensional Roesser model. The communication is only between nearest adjacent members with a simple state variable. As an example, we applied a salient region detection algorithm to the proposed framework. The simulation results indicate the designed paradigm can detect salient regions by using a moving line or v-shape formation in a scanning way. The requirement of communication and processing capability in this framework is minimal, making it a good candidate for collaborative exploration of formatted robotic swarms.
The amount of data generated worldwide related to geolocalization has exponentially increased. However, the fast processing of this amount of data is a challenge from the programming perspective, and many available so...
详细信息
ISBN:
(纸本)9781728137735
The amount of data generated worldwide related to geolocalization has exponentially increased. However, the fast processing of this amount of data is a challenge from the programming perspective, and many available solutions require learning a variety of tools and programming languages. This paper introduces the support for parallel and distributed processing in a DSL for Geospatial Data Visualization to speed up the data pre-processing phase. The results have shown the MPI version with dynamic data distribution performing better under medium and large data set files, while MPI-I/O version achieved the best performance with small data set files.
The problem of estimating the state of a linear system whose measured outputs are distributed across a network has been under study in one form or another for a number of years. Despite this, only recently have provab...
详细信息
The problem of estimating the state of a linear system whose measured outputs are distributed across a network has been under study in one form or another for a number of years. Despite this, only recently have provably correct distributed state observers emerged which solve this problem under reasonably non - restrictive assumptions. The aim of this talk is to describe some of these observers and the conditions under which they can provide asymptotically correct state estimates. For any of these observers to function robustly in the face of small modeling errors, it is necessary for the process whose state is to be estimated to be stable. Interestingly this stability requirement is also necessary for centralized {robust} state estimation, whether the estimator is a classical observer or even a Kalman filter.
We consider the problem of evaluating new improvements to distributed processing platforms like Spark and Hadoop. One approach commonly used when evaluating these systems is to use workloads published by companies wit...
详细信息
ISBN:
(纸本)9781538619568
We consider the problem of evaluating new improvements to distributed processing platforms like Spark and Hadoop. One approach commonly used when evaluating these systems is to use workloads published by companies with large data clusters, like Google and Facebook. These evaluations seek to demonstrate the benefits of improvements to critical framework components like the job scheduler, under realistic workloads. However, published workloads typically do not contain information on dependencies between the jobs. This is problematic, as ignoring dependencies could lead to significantly misestimating the speedup obtained from a particular improvement. In this position paper, we discuss why it is important to include job dependency information when evaluating distributed processing frameworks, and show that workflow mining techniques can be used to obtain dependencies from job traces that lack them. As a proof-of-concept, we show that the proposed methodology is able to find workflows in traces published by Google.
Travelling-wave and time-domain-based protection functions provide significant response time improvements over conventional phasor-based protection functions in power systems. These types of protection functions requi...
详细信息
ISBN:
(纸本)9781509066841
Travelling-wave and time-domain-based protection functions provide significant response time improvements over conventional phasor-based protection functions in power systems. These types of protection functions require very high sampling rates in the order of several hundreds of kHz. This paper proposes a novel centralized substation protection architecture (CPC) based on distributed signal processing units (DSPU) that enables the deployment of these high sampling rate applications in digital substations utilizing an Ethernet-based process-level network. The design of the DSPU is elaborated, and its signal processing algorithms are discussed. Moreover, the performance of the DSPU is analysed through dynamic tests and verified through a numerical electromagnetic transient simulation.
The spread of various sensors and the development of cloud computing technologies enable the accumulation and use of many live logs in ordinary homes. In addition, deep learning technologies have been widely used for ...
详细信息
ISBN:
(纸本)9781538650356
The spread of various sensors and the development of cloud computing technologies enable the accumulation and use of many live logs in ordinary homes. In addition, deep learning technologies have been widely used for image and speech recognition processing. However, a key issue for deep learning is heavy processing loads. To operate a service that utilizes sensor data, those data are transmitted from sensors in ordinary homes to a cloud and analyzed in the cloud. However, services that involve moving image analysis require large amounts of data to be transferred continuously and high computing power for the analysis; hence, it is difficult to process them in real time in the cloud using a conventional stream data processing framework. First, we perform preliminary experiments using Apache Spark [3] (hereinafter called Spark), which is a representative cluster computing platform that is designed to be fast and versatile, and Ray [4] , which is a distributed execution framework. We investigate the characteristics of their distributed recognition processing and demonstrate that Ray enables scalable distributed processing. Next, We implement a prototype system of the proposed distributed stream processing infrastructure using Ray and Apache Kafka [1] (hereinafter called Kafka), which is a distributed messaging system, and demonstrate its performance.
暂无评论