脉冲噪声环境下高斯稀疏信源贝叶斯压缩感知重构

 大多数现有的压缩感知重构算法对脉冲噪声不具有鲁棒性,在脉冲噪声环境下,重构性能急剧下降,使得整个重构系统崩溃.针对此问题,本文提出了一种脉冲噪声环境下的稀疏重构算法BINSR算法,其基于贝叶斯理论,可以有效地估计出信号的支撑集和脉冲噪声中脉冲的位置,并且根据压缩感知观测序列的democracy特性,利用最小均方误差MMSE估计量,有效地估计出原信号.在此基础上,本文结合鲁棒统计学,提出自适应的ABINSR算法,使其不再依赖于信号以及噪声的统计参数.实验结果表明,BINSR算法在脉冲噪声环境下可以有效地恢复出稀疏信号,很大程度上改善了脉冲噪声环境下算法的重构性能.ABINSR算法不仅对脉冲噪声具有鲁棒性,而且可以在高斯白噪声环境下实现有效的信号重构.     

P2P流媒体网络中基于博弈理论的带宽请求分配策略

基于对等网(P2P)技术的视频流媒体系统以其较低的服务器开销和良好的可扩展性被广泛应用。但由于在这类系统中普遍存在节点的资源及视频播放位置的异构性,使得节点负载不均衡的问题极易发生,进而严重影响到节点的视频播放质量。该文重点研究面向节点负载均衡的节点带宽请求分配策略,将服务请求节点竞争服务提供节点带宽的行为建模为一个非协作博弈,通过寻求该博弈问题的Nash均衡解的方法确定优化的带宽请求分配(GBRA)策略。通过和典型的带宽请求分配策略进行实验对比,数值结果表明该文提出的GBRA 策略能有效改善P2P流媒体网络节点负载不均的问题并降低节点获取所需视频数据的平均延迟。

     

一种互操作协议性能评估方法

在异构网络环境下。互操作协议对应用系统具有重要意义，互操作协议的性能对应用系统的性能也起着举足轻重的作用．为确定应用系统性能瓶颈，必须评估互操作协议性能．本文提出了一种互操作协议性能评估方法．该方法针对实际应用系统中的互操作协议，提出符合实际的评估指标，对协议性能进行定性分析和定量分析．在定性分析中得出安全性、容错能力、可扩展性等性能指标．在定量分析中，建立仿真实验模型与排队论数学模型相结合的混合模型。得出网络带宽占用、响应时间与吞吐量等性能指标．综合这些指标，我们给出互操作协议性能评估的结果．     

科技项目申请书关键词提取方法

关键词提取在文本相似度计算得到应用。传统的关键词提取方法忽略文本中的未登录词以及缺乏对词语语义的理解。针对科技项目申请书,研究提出一种基于未登录词识别与语义的关键词提取方法。应用Lucene和统计相融合的方法进行分词,并识别未登录词作为申请书关键词的一部分;依据社会网络理论构建词语语义相似度网络,并计算词语关联度提取申请书其他关键词。实验结果表明,与传统的关键词提取方法相比,新方法能提取更准确的关键词,有更好的科技项目相似性检查效果。     

基于改进反余切函数的近似L0范数的重构算法

重构算法是压缩感知理论应用于实际的关键。其中 ,近似L0范数算法是通过选取一个 连续的平滑函数近似逼近L0范数,从而将离散的最小化L0范数问题转为平滑函数求最优值问 题。针对现有算法精度不高的缺点,采用了一种逼近程度更高的改进反余切函数族来近 似L0范数,并采用修正阻尼牛顿法求解。该算法结合了近似零范数算法的高收敛性和牛顿法 的高效寻优,获得了精度较高的重构信号。仿真实验表明,在不同压缩比下,该算法在一维 信号重建和二维图像重建的过程中,较SL0算法、NSL0算法和AL0算法的信噪比和重构精度都 有了较大的提升,提高了同等条件下信号恢复的质量,有效地改善了重建效果。     

基于排队论的路由选择策略研究

针对局域网路由器配置的实际问题,利用排队论相关知识,比较不同方案下的信息传输的平均时间,优化路由设计。将信息传输过程中的路由器分别建模为开的Jackson网络、串联排队网络和取消中间路由的排队网络,比较三种传输网络方式下的信息传输时间,得到最佳的路由选择方案。利用matlab通过蒙特卡洛对传输过程进行仿真,仿真结果验证了理论分析的正确性。     

Analysis and modeling of task scheduling in wireless sensor network based on divisible load theory

In this paper, we propose a novel task scheduling algorithm (Divisible Task scheduling Algorithm for Wireless sensor networks (DTAW)) based on divisible load theory in heterogeneous wireless sensor networks to complete the tasks within the shortest possible time and reduce the sensors' energy‐consuming. In DTAW, the tasks are distributed to the wireless sensor network by the (SINK) on the basis of the processing and communication capacity of each sensor. After receiving the subtasks, the intracluster sensors carry out its tasks simultaneously and send the results to cluster head sequentially. By removing communication interference between each sensor, reduced task completion time and improved network resource utilization are achieved. Each cluster head simultaneously finishes sending fused data to the SINK after fusing the data obtained from intracluster sensors. In this way, the overlap between the task performing and communication phase would be much better. Simulation results are presented to demonstrate the impacts of different network parameters on the makespan and energy consumption. The results show that the algorithm enables to reasonably distribute tasks to each sensor and then effectively reduces the time‐consuming and energy‐consuming. Copyright © 2012 John Wiley & Sons, Ltd.     

基于朗伯模型参数估计的VLC室内定位优化算法

为实现高精度室内定位,本文设计了一种可见光 通信(VLC)室内定位系统,并通过 结合优化的朗伯模型、码分多址技术(CDMA)、三边定位算法而有效提升了定位精度和系统 扩展性。首先,每个发光二极管(LED)的ID信息经过直接序列调制后加载到LED驱动电路上 ,LED发出带有自身ID信息的灯光信号。在接收端通过光电探测器(PD)接收灯光信号,并 根据扩频码的正交性恢复出ID信息及接收信号强度(RSS),以此提高信道容量并增强系统 抗干扰能力。然后,根据朗伯光源模型,由三边定位算法得出待定位点的定位估计坐标。为 进一步提高精度,引入k最近邻(KNN)思想,采集适当的指纹点并由指纹点信息对每盏灯在 定位估计坐标处的朗伯光源模型参数进行估计,由优化后的朗伯模型计算出精度更高的定位 坐标。在1m×1m×1.35 m的空间区域中,进行本VLC室内定位系统 的实验测试。结果表明,提 出的高精度VLC室内定位系统的平均定位误差降低至2cm左右,其定位精度相比于传统三边 定 位算法提升了30%。此外,该系统方案所采用基于指纹点信息优化朗 伯模型参数的方法具备良好的实用扩展性,可实现广阔的应用场景。     

Two‐sided matching framework for optimal user association in 5G multi‐RAT UDNs

Recently, academic and industrial research communities are paying more explicit attention to the 5G multiple radio access technology ultra‐dense networks (5G multi‐RAT UDNs) for boosting network capacity, especially in UD urban zones. To this aim, in this paper, we intend to tackle the user association problem in 5G multi‐RAT UDNs. By considering the decoupled uplink/downlink access (DUDA), we divide our user association problem into two distinct subproblems representing, respectively, the uplink and the downlink channels. Next, we formulated each one as a nonlinear optimization problem with binary variables. Then, to solve them, we were restricted by the hard complexity, as well as the hard feasibility of centralized user association schemes. Thus, to resolve our user association problem in a reasonable time and distributed manner, we formulated each subproblem as a many‐to‐one matching game based on matching theory. Next, we provide two fully distributed association algorithms to compute the uplink and downlink stable matching among user equipments (UEs) and base stations (BSs). Simulation results corroborate our theoretical model and show the effectiveness and improvement of our achieved results in terms of the overall network performance, quality of service (QoS), and energy efficiency (EE) of UEs.     

Transition‐Metal Dichalcogenide NiTe2: An Ambient‐Stable Material for Catalysis and Nanoelectronics

By means of theory and experiments, the application capability of nickel ditelluride (NiTe₂) transition‐metal dichalcogenide in catalysis and nanoelectronics is assessed. The Te surface termination forms a TeO₂ skin in an oxygen environment. In ambient atmosphere, passivation is achieved in less than 30 min with the TeO₂ skin having a thickness of about 7 Å. NiTe₂ shows outstanding tolerance to CO exposure and stability in water environment, with subsequent good performance in both hydrogen and oxygen evolution reactions. NiTe₂‐based devices consistently demonstrate superb ambient stability over a timescale as long as one month. Specifically, NiTe₂ has been implemented in a device that exhibits both superior performance and environmental stability at frequencies above 40 GHz, with possible applications as a receiver beyond the cutoff frequency of a nanotransistor.