协同作战中信息分发算法研究

大规模协同作战中的高效通信问题为当前信息战的研究重点之一,但由于单独作战单元无法了解全局的信息内容及需求方,这给通信资源有限的参战兵力之间的态势共享带来了困难。为此提出了一个分布式协作信息分发算法,在理论上借鉴了复杂网络的小世界特性,即单元节点可以通过邻节点在较少跳数内完成信息的传输。在思路上利用了信息相关性概念,作战单元利用信息之间的关联性及信息与节点之间的收发关系建立控制信息发送的路由表。并通过仿真实验验证了不同网络拓扑下算法性能的变化情况,结果证明分发算法能较好地支持信息共享。     

Nuclear deterrence and cyber warfare: coexistence or competition?

Nuclear deterrence and cyber war seem almost antithetical in their respective intellectual pedigrees. Nuclear weapons are unique in their ability to create mass destruction in a short time. Information or “cyber” weapons, at least for the most part, aim at sowing confusion or mass disruption instead of widespread physical destruction. Nevertheless, there are some intersections between cyber and nuclear matters, and these have the potential to become troublesome for the future of nuclear deterrence. For example, cyber attacks might complicate the management of a nuclear crisis. As well, information attacks on command-control and communications systems might lead to a mistaken nuclear launch based on false warnings, to erroneous interpretations of data or to panic on account of feared information blackout. It is not inconceivable that future nuclear strike planning will include a preliminary wave of cyber strikes or at least a more protracted “preparation of the battlefield” by roaming through enemy networks to plant malware or map vulnerabilities.     

Explaining Cohesion in an Insurgent Organization: The Case of the Mizo National Front

Despite the prevalence of violent uprisings against the states, existing literature has produced a generalized explanation of the causes and onset of such uprisings ignoring crucial aspects of (non-state) armed group cohesion, such as their capabilities to manage intra-group tensions and rivalries. In this paper, I examine recent research in the field of civil conflict to understand the importance of investigating underlying constraints which can cause serious impediments to the capabilities of a state during the course of a war. The paper then focuses on the importance of internal strategic resources, such as pre-war and wartime sociopolitical networks in which insurgents mobilize, recruit new cadres and maintain intra-group cooperation, to suggest how it can contain the consequences of counterinsurgency operations and other state strategies on the internal cohesion of an insurgent group. A detailed case study of the Mizo National Front (MNF) in India demonstrates the argument’s plausibility.     

稀疏卷积计算高效数据加载与输出缓存策略

针对现有神经网络加速器在处理稀疏神经网络时存在的数据加载效率低、乘加资源利用率低、输出缓存寻址逻辑复杂等问题,提出了稀疏卷积计算高效数据加载与输出缓存策略。将属于同一输入通道的非零输入特征图像数据和非零权重进行全对全乘累加运算,降低了非零数据配对难度,提高了乘加资源利用率;通过采用输入驻留计算,以及密集型循环加载特征图像数据,大幅减少了数据片外调取次数;优化了输出缓存设计,解决了现有方案中存在的输出缓存地址访问争用、存储拥塞等问题。实验表明,与采用类似架构的细粒度脉动加速器相比,在处理单元面积上减少了21.45%;在数据加载速度方面平均提高了117.71%在平均乘法器利用率方面提高了11.25%,达到89%。     

联合生成对抗网络和检测网络的SAR图像目标检测

针对合成孔径雷达图像目标检测中存在的样本获取困难且数量有限问题,提出了联合生成对抗网络和检测网络的学习模型。利用原始训练集对特别设计的超快区域卷积神经网络进行预训练;再通过基于注意力机制的深度学习生成对抗网络生成高质量合成样本,并输入检测网络进行预测;依据预测信息和概率等价类属标签分配策略为新生样本提供注释信息,并以一定占比对原始训练集进行扩充;利用扩充数据集对检测网络进行再训练。多组仿真实验证明,所提框架能够有效提升网络检测效率和性能。     

分布式异构集群中节点优先级调优算法

节点优先级常用于评价异构集群中节点的性能,因此节点优先级评价指标权重的选择非常重要。采用层次分析法(analytic hierarchy process, AHP)建立了节点优先级评价指标体系,计算得到各指标的初始权重,并使用BP神经网络对初始权重进行优化。训练时,BP网络输入为集群运行中采集的节点实时资源数据,输出为节点的优先级。分析网络训练完成后得到的权重矩阵可以获得各优先级评价指标的优化权重。实验表明,基于AHP和BP的节点优先级评价模型可以更加准确地分析节点性能。相比于Spark默认算法和权重未优化的对照算法,使用调优后的节点优先级可以有效提高集群性能。运行不同工作量的相同负载时,集群平均性能分别提高了16.64%和9.76%;处理相同工作量的不同负载时,集群的平均性能分别提高了12.49%和6.54%。     

Neural network modeling to evaluate the dynamic flow stress of high strength armor steels under high strain rate compression

An artificial neural network (ANN) constitutive model is developed for high strength armor steel tempered at 500 °C, 600 °C and 650 °C based on high strain rate data generated from split Hopkinson pressure bar (SHPB) experiments. A new neural network configuration consisting of both training and validation is effectively employed to predict flow stress. Tempering temperature, strain rate and strain are considered as inputs, whereas flow stress is taken as output of the neural network. A comparative study on Johnson–Cook (J–C) model and neural network model is performed. It was observed that the developed neural network model could predict flow stress under various strain rates and tempering temperatures. The experimental stress–strain data obtained from high strain rate compression tests using SHPB, over a range of tempering temperatures (500–650 °C), strains (0.05–0.2) and strain rates (1000–5500/s) are employed to formulate J–C model to predict the high strain rate deformation behavior of high strength armor steels. The J-C model and the back-propagation ANN model were developed to predict the high strain rate deformation behavior of high strength armor steel and their predictability is evaluated in terms of correlation coefficient (R) and average absolute relative error (AARE). R and AARE for the J–C model are found to be 0.7461 and 27.624%, respectively, while R and AARE for the ANN model are 0.9995 and 2.58%, respectively. It was observed that the predictions by ANN model are in consistence with the experimental data for all tempering temperatures.     

无线传感器网络DV-Hop节点定位改进算法

针对无线传感器网络中距离无关类节点定位算法定位误差较大的问题,提出了一种改进型DV-Hop节点定位算法。通过设置待定位节点到信标节点间的最小跳数门限,降低了定位累积误差;改进了平均每跳距离估计方法,并利用信标节点测量的位置误差作为修正值,对每跳距离的估计值进行修正;待定位节点仅选取与之较近的信标节点计算位置,降低了距离误差。仿真结果显示,在信标节点比例和网络节点总数相同的条件下,改进算法性能明显优于DV-Hop算法。     

THO-MAC:一种两跳优化的低延迟低功耗无线传感器网络介质访问控制协议

功耗与延迟是无线传感器网络介质访问控制协议设计首要考虑的两个问题。提出了一种新的传感器网络低延迟、低功耗、接收节点初始化异步介质访问控制协议——THO-MAC协议。通过准确预测接收节点的唤醒时间,THO-MAC协议调度发送节点侦听信道,从而减少发送节点空闲侦听能量浪费。THOMAC协议在发送节点两跳转发节点集中选择使报文两跳转发延迟最小的转发节点,从而降低报文传输延迟。使用NS2模拟器对THO-MAC协议进行了详细模拟。模拟结果显示,与RI-MAC和Any-MAC协议相比,THOMAC协议可以减少35.5%和18%的报文传输延迟,同时节省23.5%和15.5%的节点功耗。     

动态拓扑图的卫星网络可视化方法

针对高度动态变化的卫星网络,提出一种基于动态拓扑图的可视化方法。动态拓扑图可视化的难点在于如何保持动态可视化过程中的稳定性,从而使用户容易地感知到网络中所发生的拓扑变化。根据卫星网络的动态变化特点,构建连续的动态拓扑图模型;设计一种保持布局稳定性的策略,并基于力引导思想提出一种动态拓扑图布局算法;以Iridium系统为典型实例,验证本文的可视化方法的合理性和可行性。实验表明,该方法能够以清晰的可视化图像支持用户对卫星网络动态拓扑的感知和理解。