一种基于视联AI业务的算网调度方案

在国家“东数西算”战略持续推进且全国一体化算力网络深化实施背景下,本文设计了一种基于视联AI业务的算网调度方案,为算力网络的度量和建模提供一种实践参考。方案以视联业务调用AI算法推理场景为切入点,设计整体体系架构和流程,同时构建业务场景需求到算网资源度量的映射关系模型,最后综合资源使用率、算网成本和绿色低碳等因子构建多因子评价模型,为业务需求匹配最佳算网节点和路径,实现运营成本最低。

基于云计算与大数据技术的算网调度管理系统建设研究

算网调度管理系统是一种高度智能化、高效化、安全化、可靠化的企业级资源调度和管理平台,可以为企业提供全方位的资源管理和调度服务,从而实现企业的可持续发展和高质量的业务增长。随着大带宽、低时延、强算力的需求越来越广,传统的系统由于算力资源与网络资源分布杂乱,没有一体化统一呈现算网资源及服务,导致组网不可视。同时,因为当前资源多以固定规格、配额为主,无法面对业务需求灵活、动态地调整算网资源。因此,在信息技术高速发展的背景下,如何利用新兴信息技术推进算网调度管理系统建设就成为各行各业亟须解决的重要课题。

基于云电脑业务场景的算网智能调度研究

本文旨在研究云电脑业务场景的算网感知和调度能力,通过云电脑业务场景特点分析,结合算网融合机制,提出了基于算网调度系统的创新解决方案,实现了资源的智能调度与优化,大幅提升了用户体验与业务效率。本文中的算网调度系统以其高效的信息处理能力、智能化的自我优化及强大的协同能力,为云电脑服务提供了强大的技术支撑,也为后续实践应用提供了有益参考。

算力网络一体化支撑方案及应用场景探索

【目的】本文在“东数西算”国家工程背景下,从电信运营商的定位及挑战出发,重点介绍算力网络一体化支撑体系及应用场景的探索成果。【方法】本文从底层算力基础设施、中层算力编排管理层(算网大脑)、上层算力运营服务层三方面,阐述了算力网络意图受理、可信交易、统一纳管、统一度量、能力封装、编排调度、算方感知七大方面能力及多种落地场景,为算力供给方、需求方、运营方提供指导建议。【结果】电信运营商在算力网络发展历程中具备多重优势,通过体系化的构建,未来算力网络将在新媒体、云游戏、工业视觉、智慧家庭等丰富场景中落地。【局限】算力网络分三个阶段在不断演进,在技术路线、技术标准等多方面未形成统一,如何稳定推向市场商用还仍在试点,需于当前背景下在算网感知、算网运营、算网编排等核心能力方面不断提升加强,在不同场景中进行探索。【结论】在国家政策、技术发展双重驱动下,算力网络已成为新的发展方向,算力网络的发展将会上升到前所未有的高度。未来算力网络将成为新基建的一个核心的组成部分,渗透到生活的方方面面,加速驱动产业数字化转型。

Modeling and performance evaluation of QoS-aware job scheduling of computational grids

To achieve high quality of service （QoS） on computational grids, the QoS-aware job scheduling is investigated for a hierarchical decentralized grid architecture that consists of multilevel schedulers. An integrated QoS-aware job dispatching policy is proposed, which correlates priorities of incoming jobs used for job selecting at the local scheduler of the grid node with the job dispatching policies at the global scheduler for computational grids. The stochastic high-level Petri net （SHLPN） model of a two-level hierarchy computational grid architecture is presented, and a model refinement is made to reduce the complexity of the model solution. A performance analysis technique based on the SHLPN is proposed to investigate the QoS-aware job scheduling policy. Numerical results show that the QoS-aware job dispatching policy outperforms the QoS-unaware job dispatching policy in balancing the high-priority jobs, and thus enables priority-based QoS.

Throughput scheduling in cognitive radio networks based on immune optimization

To study the throughput scheduling problem under interference temperature in cognitive radio networks, an immune algorithm-based suboptimal method was proposed based on its NP-hard feature. The problem is modeled as a constrained optimization problem to maximize the total throughput of the secondary users（ SUs）. The mapping between the throughput scheduling problems and the immune algorithm is given. Suitable immune operators are designed such as binary antibody encoding, antibody initialization based on pre-knowledge, a proportional clone to its affinity and an adaptive mutation operator associated with the evolutionary generation. The simulation results showthat the proposed algorithm can obtain about 95% of the optimal throughput and operate with much lower liner computational complexity.

A flexible architecture for job management in a grid environment

Job management is a key issue in computational grids, and normally involves job definition, scheduling, executing and monitoring. However, job management in the existing grid middleware needs to be improved in terms of efficiency and flexibility. This paper addresses a flexible architecture for job management with detailed design and implementation. Frameworks for job scheduling and monitoring, as two important aspects, are also presented. The proposed job management has the advantages of reusability of job definition, flexible and automatic file operation, visual steering of file transfer and job execution, and adaptive application job scheduler. A job management wizard is designed to implement each step. Therefore, what the grid user needs to do is only to define the job by constructing necessary information at runtime. In addition, the job space is adopted to ensure the security of the job management. Experimental results showed that this approach is user-friendly and system efficient.

EEFA： Energy Efficiency Frame Aggregation Scheduling Algorithm for IEEE 802.11n Wireless Network

Packet size is restricted due to the error-prone wireless channel which drops the network energy utilization. Furthermore, the frequent packet retransmissions also lead to energy waste. In order to improve the energy efficiency of wireless networks and save the energy of wireless devices, EEFA （Energy Efficiency Frame Aggregation）, a frame aggregation based energy-efficient scheduling algorithm for IEEE 802.11n wireless network, is proposed. EEFA changes the size of aggregated frame dynamically according to the frame error rate, so as to ensure the data transmission and retransmissions completed during the TXOP and reduce energy consumption of channel contention. NS2 simulation results show that EEFA algorithm achieves better performance than the original frame-aggregation algorithm.

A hybrid differential evolution algorithm for meta-task scheduling in grids

Task scheduling is one of the core steps to effectively exploit the capabilities of heterogeneous re-sources in the grid.This paper presents a new hybrid differential evolution(HDE)algorithm for findingan optimal or near-optimal schedule within reasonable time.The encoding scheme and the adaptation ofclassical differential evolution algorithm for dealing with discrete variables are discussed.A simple but ef-fective local search is incorporated into differential evolution to stress exploitation.The performance of theproposed HDE algorithm is showed by being compared with a genetic algorithm(GA)on a known staticbenchmark for the problem.Experimental results indicate that the proposed algorithm has better perfor-mance than GA in terms of both solution quality and computational time,and thus it can be used to de-sign efficient dynamic schedulers in batch mode for real grid systems.

Coordinated node scheduling for energy-conserving in large wireless sensor networks

Aiming at developing a node scheduling protocol for sensor networks with fewer active nodes,we propose a coordinated node scheduling protocol based on the presentation of a solution and its optimization to determine whether a node is redundant.The proposed protocol can reduce the number of working nodes by turning off as many redundant nodes as possible without degrading the coverage and connectivity.The simulation result shows that our protocol outperforms the peer with respect to the working node number and dynamic coverage percentage.

A Chord-based resource scheduling approach in drug discovery grid

This paper presents a resource scheduling approach in grid computing environment. Using P2P technology, this novel approach can schedule dynamic grid computing resources efficiently. Grid computing resources in different domains are organized into a structured P2P overlay network. Available resource information is published in type of grid services. Task requests for computational resources are also presented as grid services. Problem of resources scheduling is translated into services discovery. Different from central scheduling approaches that collect available resources information, this Chord-based approach forwards task requests in the overlay network and discovers satisfied resources for these tasks. Using this approach, the computational resources of a grid system can be scheduled dynamically according to the real- time workload on each peer. Furthermore, the application of this approach is introduced into DDG, a grid system for drug discovery and design, to evaluate the performance. Experimental results show that computational resources of a grid system can be managed efficiently, and the system can hold a perfect load balance state and robustness.

Distributed fair queuing algorithm based on compensation coordination scheduling in WMN

This paper proposes a distributed fair queuing algorithm which is based on compensation coordi- nation scheduling in wireless mesh networks, considering such problems as the location-dependent competition and unfair channel bandwidth allocation among nodes. The data communication process requiring the establishment of compensation coordination scheduling model is divided into three periods： the sending period, the compensation period and the dormancy period. According to model parameters, time constraint functions are designed to limit the execution length of each period. The algorithms guarantee that the nodes complete fair transmission of network packets together in accordance with the fixed coordination scheduling rule of the model. Simulations and analysis demonstrate the effectiveness of the proposed algorithm in network throughput and fairness.

Multicast Scheduling Algorithm Supporting Spatial Mini-slot Reuse for IEEE 802.16 Mesh Networks

Abstract： In this paper, we propose a mecha- nism for multicast data transmission in IEEE 802.16 mesh networks aimed at increasing the throughput by incorporating mini-slot spatial reuse. The proposed mechanism includes two novel algorithms： a source-based multicast tree topology construction algorithm followed by an interference-aware multieast scheduling algorithm. The proposed multicast interfer- ence-aware scheduling algorithm can be ap- plied to both source-based and rendez- vous-based multicast tree topologies. Results of our simulation study show that in compari- son to the mechanism used for the IEEE 802.16＇s standard, the proposed multicast tree generation algorithm reduces the number of consumed mini-slots by 64% on average. Moreover, using the proposed interfer- ence-aware scheduling algorithm decreases the number of required mini-slots by a further 22% on average. Therefore, the proposed mul- ticast scheduling mechanism shows a higher throughput than the previous approaches and it is more scalable with respect to increasing the number of multicast groups as well as in- creasing the number of members inside each multicast group.

Joint Dynamic Clustering and User Scheduling for Downlink Cloud Radio Access Network with Limited Feedback

In limited feedback-based CloudRAN(C-RAN) systems,the inter-cluster and intra-cluster interference together with the quantification error can seriously deteriorates the system spectral efficiency.We,in this paper,propose an efficient three-phase framework and corresponding algorithms for dealing with this problem.Firstly,a greedy scheduling algorithm based on the lower bound of the ergodic rate is performed for generating an elementary cluster in the first phase.And then the elementary cluster is divided into many small clusters according to the following proposed algorithms based on the short term instantaneous information in the second phase.In the end,based on the limited feedback two zero-forcing(ZF) precoding strategies are adopted for reducing the intra-cluster interference in the third phase.The provided Monte Carlo simulations show the effectiveness of our proposed algorithms in the respect of system spectral efficiency and average user rate.

Resource Scheduling Algorithm and Ecnomic Model in ForCES Networks

Resource scheduling algorithm for ForCES(Forwarding and Control Element Separation) networks need to meet the flexibility,programmability and scalability of node resources.DBC(Deadline Budget Constrain) algorithm relies on users select cost or time priority,then scheduling to meet the requirements of users.However,this priority strategy of users is relatively simple,and cannot adapt to dynamic change of resources,it is inevitable to reduce the QoS.In order to improve QoS,we refer to the economic model and resource scheduling model of cloud computing,use SAL(Service Level Agreement) as pricing strategy,on the basis of DBC algorithm,propose an DABP(Deadline And Budget Priority based on DBC) algorithm for ForCES networks,DABP combines both budget and time priority to scheduling.In simulation and test,we compare the task finish time and cost of DABP algorithm with DP(Deadline Priority) algorithm and BP(Budget Priority) algorithm,the analysis results show that DABP algorithm make the task complete with less cost within deadline,benifical to load balancing of ForCES networks.

A novel deadline and budget constrained scheduling heuristics for computational grids

The conventional deadline and budget constrained （DBC） scheduling heuristics for economic-based computational grids does not take the inconsistency of grid heterogeneity into account, which can lead to decline of application completion ratios. Motivated by this fact, a novel DI3C scheduling heuristics was proposed to deal with sequential workflow applications. In order to valuate the inconsistency, the relative cost （RC） metric was introduced, which was used to indicate the task-starving degree for resources. The new algorithm assigns tasks to resources, considering completion time, budget and RC together. The GridSim toolkit and the benchmark suites of the standard performance evaluation corporation （SPEC） were used to simulate the heterogeneous grid environment and applications. The experimental results show that the task and workflow completion ratios of the new heuristics are higher than those of the conventional heuristics.

Efficient Broadcast Retransmission Based on Network Coding for InterPlaNetary Internet

In traditional wireless broadcast networks,a corrupted packet must be retransmitted even if it has been lost by only one receiver.Obviously,this is not bandwidth-efficient for the receivers that already hold the retransmitted packet.Therefore,it is important to develop a method to realise efficient broadcast transmission.Network coding is a promising technique in this scenario.However,none of the proposed schemes achieves both high transmission efficiency and low computational complexity simultaneously so far.To address this problem,a novel Efficient Opportunistic Network Coding Retransmission(EONCR)scheme is proposed in this paper.This scheme employs a new packet scheduling algorithm which uses a Packet Distribution Matrix(PDM)directly to select the coded packets.The analysis and simulation results indicate that transmission efficiency of EONCR is over 0.1,more than the schemes proposed previously in some simulation conditions,and the computational overhead is reduced substantially.Hence,it has great application prospects in wireless broadcast networks,especially energyand bandwidth-limited systems such as satellite broadcast systems and Planetary Networks(PNs).