A Survey on Task Scheduling of CPU-GPU Heterogeneous Cluster

This paper reviews task scheduling frameworks,methods,and evaluation metrics of central processing unit-graphics processing unit(CPU-GPU)heterogeneous clusters.Task scheduling of CPU-GPU heterogeneous clusters can be carried out on the system level,nodelevel,and device level.Most task-scheduling technologies are heuristic based on the experts’experience,while some technologies are based on statistic methods using machine learning,deep learning,or reinforcement learning.Many metrics have been adopted to evaluate and compare different task scheduling technologies that try to optimize different goals of task scheduling.Although statistic task scheduling has reached fewer research achievements than heuristic task scheduling,the statistic task scheduling still has significant research potential.

A New Load Balancing Scheme on Heterogeneous Database Cluster

In heterogeneous database cluster, the performance of load balancing is closely related to the computing capabilities of heterogeneous nodes and the different types of workloads. Thus, a method is introduced to evaluate the load status of nodes by the weighted load values with consideration of both the utilization of different resources and the workload types in a load balaneer and an efficient and dynamic load balancing scheme is proposed for OLTP（online transaction processing） workloads to maximize the utilization of distributed resources and achieve better performance, which need not collect the feedback of load information from the lower nodes and effectively keeps from the data skew. The simulation results for OLTP services gained by TPC-C tool show that the dynamic weighted balancing policy leads to sub-linear throughput speedup and keeps the heterogeneous cluster well balanced.

A LIMITED RESOURCE VECTOR LOAD-BALANCING ALGORITHM FOR SOFTSWITCH-BASED HETEROGENEOUS CLUSTERED MEDIA SERVER

Based on the system feature of softswitch-based heterogeneous clustered media server, this paper proposed a limited resource vector load-balancing algorithm. The purpose of the algorithm was to balance the load of clusters by utilizing all system resources effectively and to avoid violent shaking of the system per- formance. A lot of simulations on the Petri net model of load balance system are conducted and the algorithm is compared with some traditional algorithms on balancing ability for heterogeneity, system throughput, re- quest response time and performance stability. The results of simulations show that the algorithm achieves system higher performance and it has excellent ability to deal with the heterogeneity of clustered media server.

Cluster Head Selection and Multipath Routing Based Energy Efficient Wireless Sensor Network

The Wireless Sensor Network(WSN)is a network of Sensor Nodes(SN)which adopt radio signals for communication amongst themselves.There is an increase in the prominence of WSN adaptability to emerging applications like the Internet of Things(IoT)and Cyber-Physical Systems(CPS).Data secur-ity,detection of faults,management of energy,collection and distribution of data,network protocol,network coverage,mobility of nodes,and network heterogene-ity are some of the issues confronted by WSNs.There is not much published information on issues related to node mobility and management of energy at the time of aggregation of data.Towards the goal of boosting the mobility-based WSNs’network performance and energy,data aggregation protocols such as the presently-used Mobility Low-Energy Adaptive Clustering Hierarchy(LEACH-M)and Energy Efficient Heterogeneous Clustered(EEHC)scheme have been exam-ined in this work.A novel Artificial Bee Colony(ABC)algorithm is proposed in this work for effective election of CHs and multipath routing in WSNs so as to enable effective data transfer to the Base Station(BS)with least energy utilization.There is avoidance of the local optima problem at the time of solution space search in this proposed technique.Experimentations have been conducted on a large WSN network that has issues with mobility of nodes.

Energy-Efficient Routing Algorithm Based on Small-World Characteristics

Water quality sensor networks are widely used in water resource monitoring.However,due to the fact that the energy of these networks cannot be supplemented in time,it is necessary to study effective routing protocols to extend their lifecycle.To address the problem of limited resources,a routing optimization algorithm based on a small-world network model is proposed.In this paper,a small-world network model is introduced for water quality sensor networks,in which the short average path and large clustering coefficient of the model are used to construct a super link.A short average path can reduce the network’s energy consumption,and a large coefficient can improve its fault-tolerance ability.However,the energy consumption of the relay nodes near the heterogeneous node is too great,and as such the energy threshold and non-uniform clustering are constructed to improve the lifecycle of the network.Simulation results show that,compared with the low-energy adaptive clustering hierarchy routing algorithm and the best sink location clustering heterogeneous network routing algorithm,the proposed improved routing model can effectively enhance the energy-utilization.The lifecycle of the network can be extended and the data transmission amount can be greatly increased.

AN ADVANCED DYNAMIC FEEDBACK AND RANDOM DISPATCHING LOAD-BALANCING ALGORITHM FOR GMLC IN 3G

Based on the system architecture and software structure of GMLC (Gateway Mobile Location Center) in 3G (third generation), a new dynamic load-balancing algorithm is proposed. It bases on dynamic feedback and imports the increment for admitting new request into the load forecast. It dynamically adjusts the dispatching probability according to the remainder process capability of each node. Experiments on the per- formance of algorithm have been carried out in GMLC and the algorithm is compared with Pick-KX algorithm and DFB (Dynamic FeedBack) algorithm in average throughput and average response time. Experiments re- sults show that the average throughput of the proposed algorithm is about five percents higher than that of the other two algorithms and the average response time is four percents higher under high system loading condi- tion.

Self-assembled multifunctional nanotheranostics against circulating tumor clusters in metastatic breast cancer

Circulating tumor clusters(CTC)disseminating from the primary tumor are responsible for secondary tumor formation where the conventional treatments such as chemotherapy and radiotherapy does not prevent the metastasis at locally advanced stage of breast cancer.In this study,a smart nanotheranostic system has been developed to track and eliminate the CTCs before it can colonize at a new site,which would reduce metastatic progression and increase the five-year survival rate of the breast cancer patients.Targeted multiresponsive(magnetic hyperthermia and pH)nanomicelles incorporated with NIR fluorescent superparamagnetic iron oxide nanoparticles were developed based on self-assembly for dual modal imaging and dual toxicity for spontaneous killing of CTCs in blood stream.A heterogenous tumor clusters model was developed to mimic the CTCs isolated from breast cancer patients.The nanotheranostic system was further evaluated for the targeting property,drug release kinetics,hyperthermia and cytotoxicity against developed CTC model in vitro.In vivo model in BALB/c mice equivalent to stageⅢandⅣhuman metastatic breast cancer was developed to evaluate the biodistribution and therapeutic efficacy of micellar nanotheranostic system.Reduced CTCs in blood stream and low distant organ metastasis after treatment with the nanotheranostic system demonstrates its potential to capture and kill the CTCs that minimize the secondary tumor formation at distant sites.

Heterogeneous clustering via adversarial deep Bayesian generative model

This paper aims to study the deep clustering problem with heterogeneous features and unknown cluster number.To address this issue,a novel deep Bayesian clustering framework is proposed.In particular,a heterogeneous feature metric is first constructed to measure the similarity between different types of features.Then,a feature metric-restricted hierarchical sample generation process is established,in which sample with heterogeneous features is clustered by generating it from a similarity constraint hidden space.When estimating the model parameters and posterior probability,the corresponding variational inference algorithm is derived and implemented.To verify our model capability,we demonstrate our model on the synthetic dataset and show the superiority of the proposed method on some real datasets.Our source code is released on the website:Github.com/yexlwh/Heterogeneousclustering.

An Efficient Parallel Solution Framework for the Linear Solution of Large Systems on PC Clusters

In this paper, a parallel solution framework for the linear static analysis of large structures on PC clusters is presented. The framework consists of two main steps: data preparation and parallel solution. The parallel solution is performed by a substructure based method with direct solvers. The aim of the data preparation step is to create the best possible substructures so that the parallel solution time is minimized. An actual structural model was solved utilizing both homogeneous and heterogeneous PC clusters to illustrate the performance and applicability of the presented framework.

Meta-Path-Based Search and Mining in Heterogeneous Information Networks

Information networks that can be extracted from many domains are widely studied recently. Different functions for mining these networks are proposed and developed, such as ranking, community detection, and link prediction. Most existing network studies are on homogeneous networks, where nodes and links are assumed from one single type. In reality, however, heterogeneous information networks can better model the real-world systems, which are typically semi-structured and typed, following a network schema. In order to mine these heterogeneous information networks directly, we propose to explore the meta structure of the information network, i.e., the network schema. The concepts of meta-paths are proposed to systematically capture numerous semantic relationships across multiple types of objects, which are defined as a path over the graph of network schema. Meta-paths can provide guidance for search and mining of the network and help analyze and understand the semantic meaning of the objects and relations in the network. Under this framework, similarity search and other mining tasks such as relationship prediction and clustering can be addressed by systematic exploration of the network meta structure. Moreover, with user＇s guidance or feedback, we can select the best meta-path or their weighted combination for a specific mining task.

Joint Design of Clustering and In-cluster Data Route for Heterogeneous Wireless Sensor Networks

A heterogeneous wireless sensor network comprises a number of inexpensive energy constrained wireless sensor nodes which collect data from the sensing environment and transmit them toward the improved cluster head in a coordinated way. Employing clustering techniques in such networks can achieve balanced energy consumption of member nodes and prolong the network lifetimes.In classical clustering techniques, clustering and in-cluster data routes are usually separated into independent operations. Although separate considerations of these two issues simplify the system design, it is often the non-optimal lifetime expectancy for wireless sensor networks. This paper proposes an integral framework that integrates these two correlated items in an interactive entirety. For that,we develop the clustering problems using nonlinear programming. Evolution process of clustering is provided in simulations. Results show that our joint-design proposal reaches the near optimal match between member nodes and cluster heads.

A Fuzzy Neural Network Based Dynamic Data Allocation Model on Heterogeneous Multi-GPUs for Large-scale Computations

The parallel computation capabilities of modern graphics processing units （GPUs） have attracted increasing attention from researchers and engineers who have been conducting high computational throughput studies. However, current single GPU based engineering solutions are often struggling to fulfill their real-time requirements. Thus, the multi-GPU-based approach has become a popular and cost-effective choice for tackling the demands. In those cases, the computational load balancing over multiple GPU ＂nodes＂ is often the key and bottleneck that affect the quality and performance of the real=time system. The existing load balancing approaches are mainly based on the assumption that all GPU nodes in the same computer framework are of equal computational performance, which is often not the case due to cluster design and other legacy issues. This paper presents a novel dynamic load balancing （DLB） model for rapid data division and allocation on heterogeneous GPU nodes based on an innovative fuzzy neural network （FNN）. In this research, a 5-state parameter feedback mechanism defining the overall cluster and node performance is proposed. The corresponding FNN-based DLB model will be capable of monitoring and predicting individual node performance under different workload scenarios. A real=time adaptive scheduler has been devised to reorganize the data inputs to each node when necessary to maintain their runtime computational performance. The devised model has been implemented on two dimensional （2D） discrete wavelet transform （DWT） applications for evaluation. Experiment results show that this DLB model enables a high computational throughput while ensuring real=time and precision requirements from complex computational tasks.