A GPU-Accelerated In-Memory Metadata Management Scheme for Large-Scale Parallel File Systems

Driven by the increasing requirements of high-performance computing applications,supercomputers are prone to containing more and more computing nodes.Applications running on such a large-scale computing system are likely to spawn millions of parallel processes,which usually generate a burst of I/O requests,introducing a great challenge into the metadata management of underlying parallel file systems.The traditional method used to overcome such a challenge is adopting multiple metadata servers in the scale-out manner,which will inevitably confront with serious network and consistence problems.This work instead pursues to enhance the metadata performance in the scale-up manner.Specifically,we propose to improve the performance of each individual metadata server by employing GPU to handle metadata requests in parallel.Our proposal designs a novel metadata server architecture,which employs CPU to interact with file system clients,while offloading the computing tasks about metadata into GPU.To take full advantages of the parallelism existing in GPU,we redesign the in-memory data structure for the name space of file systems.The new data structure can perfectly fit to the memory architecture of GPU,and thus helps to exploit the large number of parallel threads within GPU to serve the bursty metadata requests concurrently.We implement a prototype based on BeeGFS and conduct extensive experiments to evaluate our proposal,and the experimental results demonstrate that our GPU-based solution outperforms the CPU-based scheme by more than 50％under typical metadata operations.The superiority is strengthened further on high concurrent scenarios,e.g.,the high-performance computing systems supporting millions of parallel threads.     

Pirax: framework for application piracy control in mobile cloud environment

Mobile cloud computing is an emerging technology that is gaining popularity as a means to extend the capabilities of resource-constrained mobile devices such as a smartphone. Mobile cloud computing requires specialized application development models that support computation offloading from a mobile device to the cloud. The computation offloading is performed by means of offloading application process, application component, entire application, or clone of the smartphone. The offloading of an entire application or clone of the smartphone to cloud may raise application piracy issues, which, unfortunately, have not been addressed in the existing literature. This paper presents a piracy control framework for mobile cloud environment, named Pirax, which prevents mobile applications from executing on unauthenticated devices and cloud resources. Pirax is formally verified using High Level Petri Nets, Satisfiability Modulo Theories Library and Z3 solver. Pirax is implemented on Android platform and analyzed from security and performance perspectives. The performance analysis results show that Pirax is lightweight and easy to integrate into existing mobile cloud application development models.     

MEC中资源分配与卸载决策联合优化策略

针对移动边缘计算(MEC)中用户任务处理时延与能耗过高的问题,提出了"云-边-端"三层MEC计算卸载结构下的资源分配与卸载决策联合优化策略.首先,考虑系统时延与能耗,将优化问题规划为系统总增益(任务处理时延与能耗相对减少的加权和)最大化问题;其次,为用户任务设置优先级,并根据任务数据量初始化卸载决策方案;然后,采用均衡...     

Wi-Fi网络下多AP协作边缘计算资源分配策略

近年来, AR/VR、在线游戏、4K/8K超高清视频等计算密集且时延敏感型应用不断涌现,而部分移动设备受自身硬件条件的限制,无法在时延要求内完成此类应用的计算,且运行此类应用会带来巨大的能耗,降低移动设备的续航能力.为了解决这一问题,本文提出了一种Wi-Fi网络多AP (access point)协作场景下边缘计算卸载和资源分配方案.首先,通过遗传算法确定用户的任务卸载决策.随后,利用匈牙利算法为进行任务卸载的用户分配通信资源.最后,根据任务处理时延限制,为进行任务卸载的用户分配边缘服务器计算资源,使其满足任务处理时延限制要求.仿真结果表明,所提出的任务卸载与资源分配方案能够在满足任务处理时延限制的前提下有效降低移动设备的能耗.     

区块链赋能物联网中联合资源分配与控制的智能计算迁移研究

大数据场景下,远程云服务器通常被部署用于数据处理与价值挖掘,但在面对时延敏感型或需要动态频繁交互的业务时,该种处理模式显得力不从心.作为对云计算模式的补充,雾计算因其可有效降低任务处理时延、能耗与带宽消耗而备受关注;同时,面向雾计算的计算迁移机制因其能有效缓解节点的处理负担并改善用户体验而成为领域研究焦点.在雾计算模式...     

基于光纤-无线网络的协同计算卸载算法

随着无源光网络的发展,光纤-无线网络能同时支持集中式云和边缘云计算技术,成为一种具有发展前景的网络结构。但是,现有的基于光纤-无线网络的任务协同计算卸载研究主要以最小化移动设备的能耗为目标,忽略了实时性高的任务的需求。针对实时性高的任务,提出了以最小化任务的总处理时间为目标的集中式云和边缘云协同计算卸载问题,并对其进行形式化描述。同时,通过将该问题归约为装箱问题,从而证明其为NP难解问题。提出一个启发式协同计算卸载算法,该算法通过比较不同卸载策略的任务处理时间,优先选择时间最短的任务卸载策略。同时,提出一个定制的遗传算法,获得一个更优的任务卸载策略。实验结果表明,与现有的算法相比,本文提出的启发式算法得到的任务卸载策略平均减少4.34%的任务总处理时间,而定制的遗传算法的卸载策略平均减少18.41%的任务总处理时间。同时,定制的遗传算法的卸载策略与本文提出的启发式算法相比平均减少14.49%的任务总处理时间。     

A Cloud-Manager-Based Re-Encryption Scheme for Mobile Users in Cloud Environment: a Hybrid Approach

Cloud computing is an emerging computing paradigm that offers on-demand, flexible, and elastic computational and storage services for the end-users. The small and medium-sized business organization having limited budget can enjoy the scalable services of the cloud. However, the migration of the organizational data on the cloud raises security and privacy issues. To keep the data confidential, the data should be encrypted using such cryptography method that provides fine-grained and efficient access for uploaded data without affecting the scalability of the system. In mobile cloud computing environment, the selected scheme should be computationally secure and must have capability for offloading computational intensive security operations on the cloud in a trusted mode due to the resource constraint mobile devices. The existing manager-based re-encryption and cloud-based re-encryption schemes are computationally secured and capable to offload the computationally intensive data access operations on the trusted entity/cloud. Despite the offloading of the data access operations in manager-based re-encryption and cloud-based re-encryption schemes, the mobile user still performs computationally intensive paring-based encryption and decryption operations using limited capabilities of mobile device. In this paper, we proposed Cloud-Manager-based Re-encryption Scheme (CMReS) that combines the characteristics of manager-based re-encryption and cloud-based re-encryption for providing the better security services with minimum processing burden on the mobile device. The experimental results indicate that the proposed cloud-manager-based re-encryption scheme shows significant improvement in turnaround time, energy consumption, and resources utilization on the mobile device as compared to existing re-encryption schemes.     

MobiByte: An Application Development Model for Mobile Cloud Computing

Mobile cloud computing presents an effective solution to overcome smartphone constraints, such as limited computational power, storage, and energy. As the traditional mobile application development models do not support computation offloading, mobile cloud computing requires novel application development models that can facilitate the development of cloud enabled mobile applications. This paper presents a mobile cloud application development model, named MobiByte, to enhance mobile device applications’ performance, energy efficiency, and execution support. MobiByte is a context-aware application model that uses multiple data offloading techniques to support a wide range of applications. The proposed model is validated using prototype applications and detailed results are presented. Moreover, MobiByte is compared with the most recent application models with a conclusion that it outperforms the existing application models in many aspects like energy efficiency, performance, generality, context awareness, and privacy.     

C2OF2N: a low power cooperative code offloading method for femtolet-based fog network

Power and delay aware cloud service provisioning to mobile devices has become a promising domain today. This paper proposes and implements a cooperative offloading approach for indoor mobile cloud network. In the proposed work mobile devices register under femtolet which is a home base station with computation and data storage facilities. The resources of the mobile devices are collaborated in such a way that different mobile devices can execute different types of computations based on cooperative federation. The proposed offloading scheme is referred as cooperative code offloading in femtolet-based fog network. If none of the mobile device can execute the requested computation, then femtolet executes the computation. Use of femtolet provides the mobile devices voice call service as well as cloud service access. Femtolet is used as the fog device in our approach. The proposed model is simulated using Qualnet version 7. The simulation results demonstrate that the proposed scheme minimizes the energy by 15% and average delay up to 12% approximately than the existing scheme. Hence, the proposed model is referred as a low power offloading approach.     

车辆边缘计算环境下任务卸载研究综述

计算密集和延迟敏感型车辆应用的出现对车辆设备有限的计算能力提出了严峻的挑战,将任务卸载到传统的云平台会有较大的传输延迟,而移动边缘计算专注于将计算资源转移到网络的边缘,为移动设备提供高性能、低延迟的服务,因此可作为处理计算密集和延迟敏感的任务的一种有效方法.同时,鉴于城市地区拥有大量智能网联车辆,将闲置的车辆计算资源充分利用起来可以提供巨大的资源和价值,因此在车联网场景下,结合移动边缘计算产生了新的计算模式——车辆边缘计算.近年来,智能网联车辆数量的增长和新兴车辆应用的出现促进了对车辆边缘计算环境下任务卸载的研究,本文对现有车辆边缘计算环境下任务卸载研究进展进行综述,首先,从计算模型、任务模型和通信模型三个方面对系统模型进行梳理、比较和分析.然后介绍了最小化卸载延迟、最小化能量消耗和应用结果质量三种常见的优化目标,并按照集中式和分布式两种不同的决策方式对现有的研究进行了详细的归类和比较.此外,本文还介绍了几种常用的实验工具,包括SUMO、Veins和VeinsLTE.最后,本文围绕卸载决策算法复杂度、安全与隐私保护和车辆移动性等方面对车辆边缘计算任务卸载目前面临的挑战进行了总结,并展望了车辆边缘计算环境下任务卸载未来的发展方向与前景.     

边缘计算中任务卸载研究综述

近年来,随着移动智能设备的普及以及5G等无线通信技术的发展,边缘计算作为一种新兴的计算模式被提出,作为传统的云计算模式的扩展与补充.边缘计算的基本思想是将移动设备上产生的计算任务从卸载到云端转变为卸载到网络边缘端,从而满足实时在线游戏、增强现实等计算密集型应用对低延迟的要求.边缘计算中的计算任务卸载是一个关键的研究问题...     

Multiobjective computation offloading for workflow management in cloudlet‐based mobile cloud using NSGA‐II

Cloudlet is a novel computing paradigm, introduced to the mobile cloud service framework, which moves the computing resources closer to the mobile users, aiming to alleviate the communication delay between the mobile devices and the cloud platform and optimize the energy consumption for mobile devices. Currently, the mobile applications, modeled by the workflows, tend to be complicated and computation‐intensive. Such workflows are required to be offloaded to the cloudlet or the remote cloud platform for execution. However, it is still a key challenge to determine the offloading resolvent for the deadline‐constrained workflows in the cloudlet‐based mobile cloud, since a cloudlet often has limited resources. In this paper, a multiobjective computation offloading method, named MCO, is proposed to address the above challenge. Technically, an energy consumption model for the mobile devices is established in the cloudlet‐based mobile cloud. Then, a corresponding computation offloading method, by improving Nondominated Sorting Genetic Algorithm II, is designed to achieve the goal of energy saving for all the mobile device while satisfying the deadline constraints of the workflows. Finally, extensive experimental evaluations are conducted to demonstrate the efficiency and effectiveness of our proposed method.     

Smart mobile computation offloading: Centralized selective and multi-objective approach

Although mobile devices have been considerably upgraded to more powerful terminals, yet their lightness feature still impose intrinsic limitations in their computation capability, storage capacity and battery lifetime. With the ability to release and augment the limited resources of mobile devices, mobile cloud computing has drawn significant research attention allowing computations to be offloaded and executed on remote resourceful infrastructure. Nevertheless, circumstances like mobility, latency, applications execution overload and mobile device state; any can affect the offloading decision, which might dictate local execution for some tasks and remote execution for others. We present in this article a novel system model for computations offloading which goes beyond existing works with smart centralized, selective, and optimized approach. The proposition consists of (1)hotspots selection mechanism to minimize the overhead of the offloading evaluation process yet without jeopardizing the discovery of the optimal processing environment of tasks, (2)a multi-objective optimization model that considers adaptable metrics crucial for minimizing device resource usage and augmenting its performance, and (3)a tailored centralized decision maker that uses genetics to intelligently find the optimal distribution of tasks. The scalability, overhead and performance of the proposed hotspots selection mechanism and hence its effect on the decision maker and tasks dissemination are evaluated. The results show its ability to notably reduce the evaluation cost while the decision maker was able in turn to maintain optimal dissemination of tasks. The model is also evaluated and the experiments prove its competency over existing models with execution speedup and significant reduction in the CPU usage, memory consumption and energy loss.     

I/O-aware bandwidth allocation for petascale computing systems

In the Big Data era, the gap between the storage performance and an application’s I/O requirement is increasing. I/O congestion caused by concurrent storage accesses from multiple applications is inevitable and severely harms the performance. Conventional approaches either focus on optimizing an application’s access pattern individually or handle I/O requests on a low-level storage layer without any knowledge from the upper-level applications. In this paper, we present a novel I/O-aware bandwidth allocation framework to coordinate ongoing I/O requests on petascale computing systems. The motivation behind this innovation is that the resource management system has a holistic view of both the system state and jobs’ activities and can dynamically control the jobs’ status or allocate resource on the fly during their execution. We treat a job’s I/O requests as periodical sub-jobs within its lifecycle and transform the I/O congestion issue into a classical scheduling problem. Based on this model, we propose a bandwidth management mechanism as an extension to the existing scheduling system. We design several bandwidth allocation policies with different optimization objectives either on user-oriented metrics or system performance. We conduct extensive trace-based simulations using real job traces and I/O traces from a production IBM Blue Gene/Q system at Argonne National Laboratory. Experimental results demonstrate that our new design can improve job performance by more than 30%, as well as increasing system performance.     

基于双层Stackelberg博弈的MEC计算卸载方案

多接入边缘计算(multi-access edge computing, MEC)中的计算卸载问题已经成为当前研究的热点之一.目前的计算卸载方案仅考虑云、边、端结构中的计算卸载问题,而未考虑到其公、私有云的属性.提出了一种新的计算卸载方案,所提方案考虑了边缘计算中公有云与私有云之间的关系,将公有云作为了私有云资源的补充,可以缓解由于私有云资源局限性带来的算力不足问题;并通过建立双层Stackelberg博弈来解决计算卸载问题.对公有云、私有云以及用户的策略和收益进行了分析,求出了各参与人的最优策略,证明了双层博弈的纳什均衡解的存在性及唯一性.仿真结果和分析也验证了基于双层Stackelberg博弈的计算卸载方案的可行性,且相较基于单层Stackelberg博弈的卸载方案更高效,更适合可扩展的边缘计算的环境.     

A mobile edge computing-based applications execution framework for Internet of Vehicles

Mobile edge computing (MEC) is a promising technology for the Internet of Vehicles, especially in terms of application offloading and resource allocation. Most existing offloading schemes are sub-optimal, since these offloading strategies consider an application as a whole. In comparison, in this paper we propose an application-centric framework and build a finer-grained offloading scheme based on application partitioning. In our framework, each application is modelled as a directed acyclic graph, where each node represents a subtask and each edge represents the data flow dependency between a pair of subtasks. Both vehicles and MEC server within the communication range can be used as candidate offloading nodes. Then, the offloading involves assigning these computing nodes to subtasks. In addition, the proposed offloading scheme deal with the delay constraint of each subtask. The experimental evaluation show that, compared to existing non-partitioning offloading schemes, this proposed one effectively improves the performance of the application in terms of execution time and throughput.     

面向优先级任务的移动边缘计算资源分配方法

目前移动边缘计算中的资源分配方法,多数按照任务请求计算卸载的时间顺序分配计算资源,未考虑实际应用中任务存在优先级的问题。针对此类情况下的计算需求,提出一种面向优先级任务的资源分配方法。根据任务平均处理价值赋予其相应的优先级,对不同优先级的任务进行计算资源加权分配,在保证高优先级任务获取充足计算资源的同时,减少完成所有任务计算的总时间及能耗,从而提高服务质量。仿真结果表明,与平均分配、按任务数据量分配和本地计算方法相比,该方法的计算时延分别降低83.76%、15.05%和99.42%,能耗分别降低84.78%、17.37%和87.69%。     

A lightweight active service migration framework for computational offloading in mobile cloud computing

Cloud computing enables access to the widespread services and resources in cloud datacenters for mitigating resource limitations in low-potential client devices. Computational cloud is an attractive platform for computational offloading due to the attributes of scalability and availability of resources. Therefore, mobile cloud computing (MCC) leverages the application processing services of computational clouds for enabling computational-intensive and ubiquitous mobile applications on smart mobile devices (SMDs). Computational offloading frameworks focus on offloading intensive mobile applications at different granularity levels which involve resource-intensive mechanism of application profiling and partitioning at runtime. As a result, the energy consumption cost (ECC) and turnaround time of the application is increased. This paper proposes an active service migration (ASM) framework for computational offloading to cloud datacenters, which employs lightweight procedure for the deployment of runtime distributed platform. The proposed framework employs coarse granularity level and simple developmental and deployment procedures for computational offloading in MCC. ASM is evaluated by benchmarking prototype application on the Android devices in the real MCC environment. It is found that the turnaround time of the application reduces up to 45 % and ECC of the application reduces up to 33 % in ASM-based computational offloading as compared to traditional offloading techniques which shows the lightweight nature of the proposed framework for computational offloading.     

高效的格上基于身份的签名方案

基于身份的签名(IBS)方案可广泛应用于移动电子商务等资源受限的场合。利用Micciancio和Peikert在Eurocrypt'12上提出的陷门生成算法GenTrap、原像抽样算法SampleD和陷门委托算法DelTrap构造了一个新的基于格的IBS方案, 在标准模型下基于小整数解(SIS)问题证明了所提出的方案满足选择身份和固定选择消息攻击下的强不可伪造性, 并比较了所提出的方案与现有基于格的IBS方案的计算性能, 结果表明所提方案的效率最高。     

An Approach to Automatic Performance Prediction for Cloud-Enhanced Mobile Applications with Sparse Data

In mobile cloud computing (MCC), offloading compute-intensive parts of a mobile application onto the cloud is an attractive method to enhance application performance. To make good offloading decisions, history-based machinelearning techniques are proposed to predict application performance under various offloading schemes. However, the data sparsity problem is common in a realistic MCC scenario but is rarely the concern of existing work. In this paper, we employ a two-phase hybrid framework to predict performance for cloud-enhanced mobile applications, which is designed to be robust to the data sparsity. By training several multi-layer neural networks with historical execution records, the first phase automatically predicts some intermediate parameters for each execution of an application. The models learned by these neural networks can be shared among different applications, thus alleviating the data sparsity. Based on these predicted intermediate parameters and the application topology, the second phase deterministically calculates the estimated values of the performance metrics. The deterministic algorithm can partially guarantee the prediction accuracy of newly published applications even with no execution records. We evaluate our approach with a cloud-enhanced object recognition application and show that our approach can precisely predict the application performance and is robust to data sparsity.