A novel data fusion scheme using grey model and extreme learning machine in wireless sensor networks

With the increasing presence and adoption of wireless sensor networks (WSNs), the demand of data acquisition and data fusion are becoming stronger and stronger. In WSN, sensor nodes periodically sense data and send them to the sink node. Since the network consists of plenty of low-cost sensor nodes with limited battery power and the sensed data usually are of high temporal redundancy, prediction- based data fusion has been put forward as an important issue to reduce the number of transmissions and save the energy of the sensor nodes. Considering the fact that the sensor node usually has limited capabilities of data processing and storage, a novel prediction-based data fusion scheme using grey model (GM) and optimally pruned extreme learning machine (OP-ELM) is proposed. The proposed data fusion scheme called GM-OP-ELM uses a dual prediction mechanism to keep the prediction data series at the sink node and sensor node synchronous. During the data fusion process, GM is introduced to initially predict the data of next period with a small number of data items, and an OPELM- based single-hidden layer feedforward network (SLFN) is used to make the initial predicted value approximate its true value with extremely fast speed. As a robust and fast neural network learning algorithm, OP-ELM can adaptively adjust the structure of the SLFN. Then, GM-OP-ELM can provide high prediction accuracy, low communication overhead, and good scalability. We evaluate the performance of GM-OP-ELM on three actual data sets that collected from 54 sensors deployed in the Intel Berkeley Research lab. Simulation results have shown that the proposed data fusion scheme can significantly reduce redundant transmissions and extend the lifetime of the whole network with low computational cost.     

Divide-and-conquer architecture based collaborative sensing for target monitoring in wireless sensor networks

Most surveillance applications in wireless sensor network (WSN) have stringent accuracy requirements in targets surveillance with maximized system lifetime, while large amount of continuous sensing data and limited resource in WSNs pose great challenges. So it is necessary to select appropriate sensors that can collaboratively work with each other in order to obtain balance between accuracy and system lifetime. However, because of sensing diversity and big data from WSN, most existing methods can not select appropriate sensors to cover all critical monitoring locations in large scale real deployments. Accordingly, an AdaBoost based algorithm is first proposed to identify valid sensors with contribution towards accuracy improvement, which can reduce computation and communication overhead by excluding invalid sensors. The valid sensors are combined and work in a collaborative way, which can obtain better performance than other ways. Then, because of independence of each monitoring location, a divide-and-conquer architecture based method (EasiSS) is proposed to select the most informative sensor clusters from the valid sensors for critical monitoring locations. EasiSS can obtain higher classification accuracy at different user requirement. Finally, according to the experiment on real data, we demonstrate that our proposed method can get a better performance of sensor selection, comparing with traditional methods.     

DyDAP: A dynamic data aggregation scheme for privacy aware wireless sensor networks

End-to-end data aggregation, without degrading sensing accuracy, is a very relevant issue in wireless sensor networks (WSN) that can prevent network congestion to occur. Moreover, privacy management requires that anonymity and data integrity are preserved in such networks. Unfortunately, no integrated solutions have been proposed so far, able to tackle both issues in a unified and general environment. To bridge this gap, in this paper we present an approach for dynamic secure end-to-end data aggregation with privacy function, named DyDAP. It has been designed starting from a UML model that encompasses the most important building blocks of a privacy-aware WSN, including aggregation policies. Furthermore, it introduces an original aggregation algorithm that, using a discrete-time control loop, is able to dynamically handle in-network data fusion to reduce the communication load. The performance of the proposed scheme has been verified using computer simulations, showing that DyDAP avoids network congestion and therefore improves WSN estimation accuracy while, at the same time, guaranteeing anonymity and data integrity.     

基于压缩感知的无线传感器网络动态采样方法

基于固定采样率的无线传感网（WSN）压缩感知（CS）在收集随时间变化的数据时难以获得满意的数据恢复精度。针对该问题,提出了一种基于数据预测和采样率反馈控制的动态采样方法。首先,汇聚节点通过分析当前采样时段与上一采样时段获取数据的线性度量指标,预测数据的变化趋势;然后,根据预测结果计算感知节点未来的采样率,并通过反馈控制机制对感知节点的采样过程进行动态调节。实验结果表明,相比基于目前广泛采用的基于固定采样率的无线传感网压缩感知数据收集方法,该方法能够有效提高压缩数据的恢复精度。     

改进灰色GM(1,m)模型在变压器故障预测中的应用

针对灰色模型在预测变压器故障时对波动数据序列的预测误差较大的问题,提出了一种灰色GM(1,m)预测模型改进方案:对原始数据序列进行处理,使其具有更好的指数规律,以满足预测模型对光滑性的要求;对处理过的原始数据序列进行灰关联度分析,以得到各变量之间的关系;优化预测模型的背景值并用其建模;采用等维新息模型预测数据。采用改进的灰色GM(1,m)模型预测某变压器油中7种特征气体的体积分数,所得预测数据的平均残差和后验相对误差均小于GM(1,1)模型和传统GM(1,m)的预测结果,表明其具有更好的预测精确度。     

An energy-efficient clustered distributed coding for large-scale wireless sensor networks

A wireless sensor network (WSN) usually consists of a large number of battery-powered low-cost sensors with limited data collection and processing capacity. In order to prolong the lifetime of the WSN with a target error performance, a novel clustered distributed coding framework, referred to as distributed multiple-sensor cooperative turbo coding (DMSCTC), is developed for a large-scale WSN with sensor grouped in cooperative cluster. In the proposed DMSCTC scheme, a simple forward error correction is employed at each sensor and a simple multi-sensor joint coding is adopted at the cluster head, while complicated joint iterative decoding is implemented only at the data collector. The proposed DMSCTC scheme achieves extra distributed coding gain and cooperative spatial diversity without introducing extra complexity burden on the sensors by transferring the complicated joint decoding process to the data collector. With the proposed scheme, the WSN can achieve the target error performance with less power consumption, thus prolonging its lifetime. The error performance and energy efficiency of the proposed DMSCTC scheme are analyzed, and followed by Monte Carlo simulations. Both analytical and simulation results show that the DMSCTC can substantially improve the energy efficiency of the clustered WSN.     

An efficient scheme of target classification and information fusion in wireless sensor networks

In this paper, an efficient target classification and fusion scheme for wireless sensor networks (WSNs) is proposed and evaluated. When a classification algorithm for WSN nodes is designed, parametric approaches such as Gaussian mixture model (GMM) should be more preferred to non-parametric ones due to the hard limitation in resources. The GMM algorithm not only shows good performances for target classification in WSNs but it also requires very small resources. Based on the classifier, a decision tree generated by the classification and regression tree algorithm is used to fuse the information from heterogeneous sensors. This node-level classification scheme provides a satisfactory classification rate, 94.10%, with little resources. Finally, a confidence-based fusion algorithm improves the overall accuracy by fusing the information among sensor nodes. Our experimental results show that the proposed group-level fusion algorithm improves the accuracy by an average of 4.17% accuracy with randomly selected nodes.     

Target tracking using Interactive Multiple Model for Wireless Sensor Network

Target tracking in a Wireless Sensor Network (WSN) environment is a challenging research problem. Interactive Multiple Model (IMM) is a popular scheme for accurate target tracking. The existing target tracking scheme used in WSN employs Kalman Filter (KF) which fails to track the target accurately due to non availability of target data at regular intervals and missing of packets. Though existing KF based tracking in WSN scheme detects the target, it fails to identify the target. To overcome these problems, this paper proposes a IMM based Target Tracking in WSN named ITTWSN that uses multiple models (velocity and acceleration) to handle both maneuvering and non maneuvering targets and multiple sensors to detect and identify the targets. The performance of the proposed ITTWSN is compared with the KF scheme and it is found that the accuracy of the proposed ITTWSN is better than the existing KF based approach.     

一种基于云计算的无线传感网体系结构

提出了一种基于云计算的无线传感网体系结构。整个传感网被划分为若干个分区; 云作为虚拟汇聚节点, 有多个汇聚触点, 每个触点负责一个分区中传感器的数据收集; 每个分区中的传感器组成本地传感网, 所有本地传感网通过云连接成一个整体; 数据在云中分布存储、并行处理。模拟实验表明, 新结构传感网的数据传输性能明显提升。该结构适用于以数据为中心的无线传感网。     

基于核学习方法的短时交通流量预测

基于核学习的强大非线性映射性能,针对短时交通流量预测,提出一类基于核学习方法的预测模型。核递推最小二乘(KRLS)基于近似线性依赖(approximate linear dependence,ALD) 技术可降低计算复杂度及存储量,是一种在线核学习方法,适用于较大规模数据集的学习;核偏最小二乘(KPLS)方法将输入变量投影在潜在变量上,利用输入与输出变量之间的协方差信息提取潜在特征;核极限学习机(KELM)方法用核函数表示未知的隐含层非线性特征映射,通过正则化最小二乘算法计算网络的输出权值,能以极快的学习速度获得良好的推广性。为验证所提方法的有效性,将KELM、KPLS、ALD-KRLS用于不同实测交通流数据中,在同等条件下,与现有方法进行比较。实验结果表明,不同核学习方法的预测精度和训练速度均有提高,体现了核学习方法在短时交通流量预测中的应用潜力。     

基于支持向量机的大样本回归算法比较研究

支持向量机的研究是当前人工智能领域的研究热点。基于支持向量机的大样本回归问题一直是一个非常具有挑战性的课题。最近,基于递归最小二乘算法,Engel等人提出了核递归最小二乘算法。文中基于块增量学习和逆学习过程,提出了自适应迭代回归算法。为了说明两种方法的性能,论文在训练速度、精度和支持向量数量等方面,对它们做了比较。模拟结果表明:核递归最小二乘算法所得到的支持向量个数比自适应迭代回归算法少,而训练时间比自适应迭代回归算法的训练时间长,训练和测试精度也比自适应迭代回归算法差。     

一种融合粗糙集与灰模的装备故障预测方法

结合粗糙集理论和灰色系统理论对不精确信息处理的优势,文中提出一种融合粗糙集理论与GM(1,1)灰色预测模型的故障预测方法,先运用粗糙集的属性约简算法对故障诊断决策表进行约简,推出最优诊断规则,再利用GM(1,1)灰色预测模型对约简决策表中的各条件属性测试值计算得到其预测值,从而代回约简的诊断决策表进行故障预测,最后在某型机载电台装备中以某一故障为例进行应用验证,结果表明故障预测效率和精度都较高,从而为提高装备的可靠性和维修性提供依据.     

Energy-efficient distributed wireless sensor network scheme for cluster detection

Detection of an environmental phenomenon, e.g. air pollution and oil spills, occurs when a group of sensors continuously produces similar readings (i.e. data streams) over a period of time. Thus, detection of environmental phenomena is basically a process of clustering the sensors' data streams, which commonly involves the processing of hundreds and maybe thousands of data streams in real time. Since the sensor network environment is wireless, energy conservation of the sensors would be the main concern. Thus in this paper, we propose an efficient and energy friendly distributed scheme to detect phenomena in a wireless sensor network (WSN). To achieve fast response, the proposed algorithms reduce the dimensionality of the streams. Then, each stream is represented by a point in a multi-dimensional grid. The algorithm uses a grid-based clustering technique to detect clusters of similar stream values. The processing of the algorithm is distributed among different elements of the WSN in a hierarchical topology for more energy efficiency. The paper shows the feasibility of the proposed fully distributed scheme by comparing it with three other WSN schemes in terms of clustering accuracy and energy consumption.     

无线传感器网络低能耗协作频谱感知方法

针对无线传感器网络能量受限问题,以双门限能量检测为基础,以最小化系统感知能耗为目标,提出了一种适合无线传感器网络的休眠与审核相结合的频谱感知方法。在获得目标检测性能条件下,该方法让部分传感器负责频谱感知工作,其他处于休眠状态,只让有准确感知结果的传感器发送1 bit感知结果到融合中心。理论分析和仿真结果表明,与传统的能量检测方法相比,该方法既有效减少了系统的感知能耗,又节约了带宽。     

JAID: An algorithm for data fusion and jamming avoidance on distributed sensor networks

Mobile Agent (MA) technology has been recently proposed in Wireless Sensor Networks (WSNs) literature to answer the scalability problem of client/server model in data fusion applications. In this paper, we describe the critical role MAs can play in the field of security and robustness of a WSN in addition to data fusion. The design objective of our Jamming Avoidance Itinerary Design (JAID) algorithm is twofold: (a) to calculate near-optimal routes for MAs that incrementally fuse the data as they visit the nodes; (b) in the face of jamming attacks against the WSN, to modify the itineraries of the MAs to bypass the jammed area(s) while not disrupting the efficient data dissemination from working sensors. If the number of jammed nodes is small, JAID only modifies the pre-jamming scheduled itineraries to increase the algorithm’s promptness. Otherwise, JAID re-constructs the agent itineraries excluding the jammed area(s). Another important feature of JAID is the suppression of data taken from sensors when the associated successive readings do not vary significantly. Data suppression also occurs when sensors’ readings are identical to those of their neighboring sensors. Simulation results confirm that JAID enables retrieval of information from the working sensors of partially jammed WSNs and verifies its performance gain over alternative approaches in data fusion tasks.     

支持时钟同步技术的无线传感网操作系统研究及内核实现

详细分析了无线传感网操作系统的硬件平台环境、自身特点和基本功能，提出了一种合理划分内核层与服务层功能的操作系统整体架构及内核结构．该内核结构采用数据融合加强对精度要求较高的时钟同步任务的支持．给出了相关内核服务例程和基本的数据结构的设计．试验数据表明，数据融合后时钟同步算法中的多跳误差累积得到明显降低．     

Multisensory fusion based virtual tool wear sensing for ubiquitous manufacturing

Pervasiveness of ubiquitous computing advances the manufacturing scheme into a ubiquitous manufacturing era which poses significant challenges on sensing technology and system reliability. To improve manufacturing system reliability, this paper presents a new virtual tool wear sensing technique based on multisensory data fusion and artificial intelligence model for tool condition monitoring. It infers the difficult-to-measure tool wear parameters (e.g. tool wear width) by fusing in-process multisensory data (e.g. force, vibration, etc.) with dimension reduction technique and support vector regression model. Different state-of-the-art dimension reduction techniques including kernel principal component analysis, locally linear embedding, isometric feature mapping, and minimum redundancy maximum relevant method have been investigated for feature fusion in a virtual sensing model, and the kernel principal component analysis performs best in terms of sensing accuracy. The effectiveness of the developed virtual tool wear sensing technique is experimentally validated in a set of machining tool run-to-failure tests on a computer numerical control milling machine. The results show that the estimated tool wear width through virtual sensing is comparable to that measured offline by a microscope instrument in terms of accuracy, moreover, in a more cost-effective manner.     

改进的自回归AR（p）预测模型在WSN中的应用

自回归AR（p）预测模型是无线传感网络（WSN）中一种减少数据传输次数和降低能量消耗的方法。针对AR（p）模型在建模过程中忽略了不同时期的历史数据对预测值的影响存在的差异,导致模型预测精度不高、网络通信频率受影响的问题,提出了一种改进的预测模型FAR（p）。在AR（p）模型中引入一种新的模糊隶属度函数,通过模糊隶属度函数对预测模型的每个历史数据赋予权值,实现历史数据“重近轻远”的预测效果,并经二次加权平均算法处理后重新构建预测模型。仿真结果表明,改进的预测模型有效地提高了模型预测精度,减少了传感网络中数据传输次数,降低了能量消耗。     

基于深度学习算法的无线传感器异常数据检测

提出了基于深度学习的异常数据检测的方法,精准检测到无线传感器异常数据并直观展现检测结果。基于无线传感器网络模型分簇原理,通过异常数据驱动的簇内数据融合机制,去除无线传感器网络中的无效数据,获取无线传感器网络有效数据融合结果。构建了具有4层隐含层的深度卷积神经网络,将预处理后的无线传感器网络数据作为模型输入,通过隐含层完成数据特征提取和映射后,由输出层输出异常数据检测结果。实验证明：该方法可有效融合不同类型数据,且网络节点平均能耗较低;包含4层隐含层的深度卷积神经网络平均分类精度高达98.44%,1000次迭代后隐含层的训练损失均趋于0,可实现无线传感器异常数据实时、直观、准确检测。     

Standalone noise and anomaly detection in wireless sensor networks: A novel time-series and adaptive Bayesian-network-based approach

Wireless sensor networks (WSNs) consist of small sensors with limited computational and communication capabilities. Reading data in WSN is not always reliable due to open environmental factors such as noise, weakly received signal strength, and intrusion attacks. The process of detecting highly noisy data is called anomaly or outlier detection. The challenging aspect of noise detection in WSN is related to the limited computational and communication capabilities of sensors. The purpose of this research is to design a local time-series-based data noise and anomaly detection approach for WSN. The proposed local outlier detection algorithm (LODA) is a decentralized noise detection algorithm that runs on each sensor node individually with three important features: reduction mechanism that eliminates the noneffective features, determination of the memory size of data histogram to accomplish the effective available memory, and classification for predicting noisy data. An adaptive Bayesian network is used as the classification algorithm for prediction and identification of outliers in each sensor node locally. Results of our approach are compared with four well-known algorithms using benchmark real-life datasets, which demonstrate that LODA can achieve higher (up to 89%) accuracy in the prediction of outliers in real sensory data.