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。     

Emulation of gas‐liquid flow in packed beds for offshore floating applications using a swell simulation hexapod

A laboratory‐scale packed column was positioned on a six degree of freedom swell simulation hexapod to emulate the hydrodynamics of packed bed scrubbers/reactors onboard offshore floating systems. The bed was instrumented with wire mesh capacitance sensors to measure liquid saturation and velocity fields, flow regime transition, liquid maldistribution, and tracer radial and axial dispersion patterns while robot was subject to sinusoidal translation (sway, heave) and rotation (roll, roll + pitch, yaw) motions at different frequencies. Three metrics were defined to analyze the deviations induced by the various column motions, namely, coefficient of variation and degree of uniformity for liquid saturation fluctuating fields, and effective Péclet number. Nontilting oscillations led to frequency‐independent maldistribution while tilting motions induced swirl/zigzag secondary circulation and prompted nonuniform maldistribution oscillations that deteriorated with decreasing frequencies. Regardless of excited degree of freedom, a qualitative loss of plug‐flow character was observed compared with static vertical beds which worsened as frequencies decreased. © 2015 American Institute of Chemical Engineers AIChE J, 61: 2354–2367, 2015     

Mobile Cloud Middleware

Mobile Cloud Computing (MCC) is arising as a prominent research area that is seeking to bring the massive advantages of the cloud to the constrained smartphones. Mobile devices are looking towards cloud-aware techniques, driven by their growing interest to provide ubiquitous PC-like functionality to mobile users. These functionalities mainly target at increasing storage and computational capabilities. Smartphones may integrate those functionalities from different cloud levels, in a service oriented manner within the mobile applications, so that a mobile task can be delegated by direct invocation of a service. However, developing these kind of mobile cloud applications requires to integrate and consider multiple aspects of the clouds, such as resource-intensive processing, programmatically provisioning of resources (Web APIs) and cloud intercommunication. To overcome these issues, we have developed a Mobile Cloud Middleware (MCM) framework, which addresses the issues of interoperability across multiple clouds, asynchronous delegation of mobile tasks and dynamic allocation of cloud infrastructure. MCM also fosters the integration and orchestration of mobile tasks delegated with minimal data transfer. A prototype of MCM is developed and several applications are demonstrated in different domains. To verify the scalability of MCM, load tests are also performed on the hybrid cloud resources. The detailed performance analysis of the middleware framework shows that MCM improves the quality of service for mobiles and helps in maintaining soft-real time responses for mobile cloud applications.     

移动边缘计算中的时延和能耗均衡优化算法

为了提升移动边缘计算（MEC）网络中的任务卸载效用,提出了一种基于任务卸载增益最大化的时延和能耗均衡优化算法.通过分析通信资源和计算资源对时延和能耗这2种性能指标的制约关系,将原问题分解为联合发射功率子信道分配子问题和MEC计算频率分配子问题.通过Karush-Kuhn-Tucker条件,导出了最优的MEC计算频率闭式解.此外,提出了一种基于二分法的发射功率分配算法和基于匈牙利二部图匹配的子信道分配算法.仿真结果表明,提出的算法相比传统算法可以显著提升用户的任务卸载效用.     

无人机任务卸载与充电协同优化

在野外恶劣环境应用中，可以使用具有灵活性和便捷性的无人机（UAV），通过无线数据传输辅助携带用户任务到边缘服务器。然而，UAV飞行平台难以提供长时间的任务卸载服务，大大限制了其应用前景。本文研究了在移动边缘计算环境中，如何有效整合UAV的任务卸载和充电调度。首先，构建了一个新的应用模型，该模型协同处理UAV的任务卸载调度和自身充电需求，并在UAV辅助任务卸载应用场景中加入了若干个无线充电平台。其次，考虑了用户任务的价值和UAV的充电需求，以在时延敏感和能量约束的条件下优化UAV辅助用户设备进行任务卸载的收益。最后，采用深度强化学习算法，对深度Q网络（DQN）进行调优后形成Fixed DQN算法，以有效处理模型中的大规模状态动作搜索空间问题。本文以UAV仅作为任务载体并考虑其自主充电需求为前提，通过在一个半径为3000 m、含有11个节点的区域验证Fixed DQN算法的可行性；并在不同用户节点数量、充电节点数量及服务时间条件下，通过与蚁群算法、遗传算法和DQN算法的对比实验评估其性能。实验结果表明：本文提出的Fixed DQN算法在所有测试条件下均显著优于蚁群算法、遗传算法和DQN算法，特别是在节点数量增加和服务时间延长的情景中；此外，Fixed DQN算法相对于DQN算法的性能提升突显了深度强化学习在参数调优方面的有效性。研究结果证实了Fixed DQN算法在解决UAV任务卸载和充电调度问题中的高效性和调参策略的重要性。     

面向车辆边缘计算的任务合作卸载

随着物联网技术和人工智能技术的飞速发展，车辆边缘计算越来越引起人们的注意。车辆如何有效地利用车辆周边的各种通信、计算和缓存资源，结合边缘计算系统模型将计算任务迁移到离车辆更近的路边单元，已经成为目前车联网研究的热点。由于车辆应用设备有限的计算资源，车辆用户的任务计算需求无法满足，需要充分利用车辆周边的计算资源来计算任务。本文研究了车辆边缘计算中任务的合作卸载机制，以最小化车辆任务的计算时延。首先，设计了任务合作卸载的三层系统架构，考虑了车辆周边停泊车辆的计算资源以及路边单元的计算资源，组成云服务器层、停泊车辆合作集群层和路边单元合作集群层的三层架构。通过路边单元合作集群和停泊车辆合作集群的合作卸载，充分利用系统的空闲计算资源，进一步提高了系统的资源利用率。然后，基于k-聚类算法的思想提出了路边单元合作集群划分算法对路边单元进行合作集群的划分，并采用块连续上界最小化的分布式迭代优化方法设计了任务合作卸载算法,对终端车辆用户的任务进行卸载计算。最后，通过将本文算法和其他算法方案的进行实验对比，仿真结果表明，本文算法在系统时延和系统吞吐量方面具有更好的性能表现，可以降低23%的系统时延，并且能提升28%的系统吞吐量。     

MEC计算卸载与资源分配联合智能优化方案

移动边缘计算(MEC)中的分布式基站部署、有限的服务器资源和动态变化的终端用户使得计算卸载方案的设计极具挑战。鉴于深度强化学习在处理动态复杂问题方面的优势,设计了最优的计算卸载和资源分配策略,目的是最小化系统能耗。首先考虑了云边端协同的网络框架;然后将联合计算卸载和资源分配问题定义为一个马尔可夫决策过程,提出一种基于多智能体深度确定性策略梯度的学习算法,以最小化系统能耗。仿真结果表明,该算法在降低系统能耗方面的表现明显优于深度确定性策略梯度算法和全部卸载策略。     

Optimization and Design of Cloud-Edge-End Collaboration Computing for Autonomous Robot Control Using 5G and Beyond

Robots have important applications in industrial production, transportation, environmental monitoring and other fields, and multi-robot collaboration is a research hotspot in recent years. Multi-robot autonomous collaborative tasks are limited by communication, and there are problems such as poor resource allocation balance, slow response of the system to dynamic changes in the environment, and limited collaborative operation capabilities. The combination of 5G and beyond communication and edge computing can effectively reduce the transmission delay of task offloading and improve task processing efficiency. First, this paper designs a robot autonomous collaborative computing architecture based on 5G and beyond and mobile edge computing(MEC). Then, the robot cooperative computing optimization problem is studied according to the task characteristics of the robot swarm. Then, a reinforcement learning task offloading scheme based on Q-learning is further proposed, so that the overall energy consumption and delay of the robot cluster can be minimized. Finally, simulation experiments demonstrate that the method has significant performance advantages.     

Computation Partitioning in Mobile Cloud Computing： A Survey

Mobile devices are increasingly interacting with clouds,and mobile cloud computing has emerged as a new paradigm.An central topic in mobile cloud computing is computation partitioning,which involves partitioning the execution of applications between the mobile side and cloud side so that execution cost is minimized.This paper discusses computation partitioning in mobile cloud computing.We first present the background and system models of mobile cloud computation partitioning systems.We then describe and compare state-of-the-art mobile computation partitioning in terms of application modeling,profiling,optimization,and implementation.We point out the main research issues and directions and summarize our own works.     

Task distribution offloading algorithm of vehicle edge network based on DQN

In order to achieve the best balance between latency,computational rate and energy consumption,for a edge access network of IoV,a distribution offloading algorithm based on deep Q network (DQN) was considered.Firstly,these tasks of different vehicles were prioritized according to the analytic hierarchy process (AHP),so as to give different weights to the task processing rate to establish a relationship model.Secondly,by introducing edge computing based on DQN,the task offloading model was established by making weighted sum of task processing rate as optimization goal,which realized the long-term utility of strategies for offloading decisions.The performance evaluation results show that,compared with the Q-learning algorithm,the average task processing delay of the proposed method can effectively improve the task offload efficiency.