Search

Cooperative localization has emerged as an appealing technology since it can improve the localization performance without any infrastructure change compared with non-cooperative localization. However, some well-localized agents may not be willing to sacrifice additional power to improve the others’ localization accuracy. This paper proposes an incentive mechanism from an economic perspective for cooperative localization, whereby a pricing scheme is designed to lead each agent to the optimal state. A game-theoretic algorithm is proposed where each player (agent) can obtain the optimal budget strategy to minimize its individual utility. To make profits in the game, the relationship between the agent's network condition and its budget strategy is derived. Furthermore, a fairness-aware price allocation rule (PAR) is developed to distribute the budget among the reference agents proportional to each node's contribution. Analytical and numerical results show that agents with better network conditions are more likely to join the cooperation under the proposed incentive mechanism, leading to an improved localization performance.

In vehicle ad-hoc networks (VANETs), the full coverage of roadside units (RSUs) is not necessary with the assistance of dead reckoning (DR) for the RSU-based vehicle localisation. This study proposes an optimisation strategy of RSUs deployment, which seeks an optimal RSU layout ensuring the best localisation accuracy with a minimum number of RSUs. With the assistance of DR, first, the average geometric dilution of precision (GDOP) for a specific localisation region is derived through a non-linear recursive model. Then the RSUs deployment is formulated into an optimisation problem, and the objective is as a function of the average GDOP and deploying interval. Finally, the optimisation problem is solved by a centre particle swarm optimisation (CPSO) algorithm. The convergence and stability of CPSO are evaluated via simulations. Furthermore, simulations also show that the proposed strategy can optimise the localisation accuracy of RSUs deployment in the VANET scenario.

Information Systems (IS) researchers and practitioners have to deal with the concept of causality. IS researchers explicitly model and test causal relationships, like Information Technology (IT) in...

PURPOSE Medical image segmentation is critical for many medical image analysis applications. 3D Convolutional Neural Networks (CNNs) have been widely adopted in the segmentation of volumetric medical images. The recent development of channel-wise attention and spatial-wise attention achieve the state-of-the-art feature representation performance. However, these attention strategies have not explicitly modeled interdependencies among slices in 3D medical volumes. In this work, we propose a novel attention module called progressive attention module (PAM) to explicitly model the slice-wise importance for 3D medical image analysis. METHODS The proposed method is composed of three parts: Slice Attention (SA) block, Key-Slice-Selection (KSS) block, and Channel Attention (CA) block. Firstly, the SA is a novel attention block to explore the correlation among slices for 3D medical image segmentation. SA is designed to explicitly reweight the importance of each slice in the 3D medical image scan. Secondly, the KSS block, cooperating with the SA block, is designed to adaptively emphasize the critical slice features while suppressing the irrelevant slice features, which helps the model focus on the slices with rich structural and contextual information. Finally, the CA block receives the output of KSS as input for further feature recalibration. Our proposed PAM organically combines SA, KSS, and CA, progressively highlights the key slices containing rich information for the relevant tasks while suppressing those irrelevant slices. RESULTS To demonstrate the effectiveness of PAM, we embed it into 3D CNNs architectures and evaluate the segmentation performance on three public challenging datasets: BraTS 2018 dataset, MALC dataset, and HVSMR dataset. We achieve 80.34%, 88.98%, and 84.43% of the Dice similarity coefficient on these three datasets respectively. Experimental results show that the proposed PAM not only boosts the segmentation accuracy of the standard 3D CNNs methods consistently, but also outperforms the other attention mechanisms with slight extra costs. CONCLUSIONS We propose a new PAM to identify the most informative slices and recalibrate channel-wise feature responses for volumetric medical image segmentation. The proposed method is evaluated on three public datasets, and the results show improvements over other methods. This proposed technique can effectively assist physicians in many medical image analysis. It is also anticipated to be generalizable and transferable to assist physicians in a wider range of medical imaging applications to produce greater value and impact to health. This article is protected by copyright. All rights reserved.

To identify NLOS propagations for a high-resolution wireless localization system, a strong NLOS classifier is proposed via a machine learning algorithm named AdaBoost. Signal features are extracted to characterize the distinctions between LOS and NLOS, based on which a series of weak learners are primarily given. By training and re-sampling data iteratively, weak learners with the minimal classification error are selected and the weighted sum is boosted into a strong classifier. Simulations demonstrate that the proposed AdaBoost-based NLOS classifier can improve NLOS identification performance, leading to reduced positioning errors.

The Image Caption algorithm based on the transformer structure is different from the traditional codec model based on convolutional neural networks and recurrent neural networks. Our proposed algorithm uses a convolutional neural network to extract image features and completely relies on the self-attention mechanism to process image features and generate sentences describing image content. The experimental results verified the validation of our method using BLEU, CIDEr and other indicators on the MSCOCO data set.

The interest in the facial expression recognition (FER) is increasing day by day due to its practical and potential applications, such as human physiological interaction diagnosis and mental diseases detection. This area has received much attention from the research community in recent years and achieved remarkable results; however, a significant improvement is required in spatial problems. This research work presents a novel framework and proposes an effective and robust solution for FER under an unconstrained environment. Face detection is performed using the supervision of facial attributes. Faceness-Net is used for deep facial part responses for the detection of faces under severe unconstrained variations. In order to improve the generalization problems and avoid insufficient data regime, Deep Convolutional Graphical Adversarial Network (DC-GAN) is utilized. Due to the challenging environmental factors faced in the wild, a large number of noises disrupt feature extraction, thus making it hard to capture ground truth. We leverage different multimodal sensors with a camera that aids in data acquisition, by extracting the features more accurately and improve the overall performance of FER. These intelligent sensors are used to tackle the significant challenges like illumination variance, subject dependence, and head pose. Dual-enhanced capsule network is used which is able to handle the spatial problem. The traditional capsule networks are unable to sufficiently extract the features, as the distance varies greatly between facial features. Therefore, the proposed network is capable of spatial transformation due to action unit aware mechanism and thus forward most desiring features for dynamic routing between capsules. Squashing function is used for the classification function. We have elaborated the effectiveness of our method by validating the results on four popular and versatile databases that outperform all state-of-the-art methods.

In recent years, the research of high-precision positioning with wireless sensor networks has attracted a lot of attention, especially in the scenarios of UAV (unmanned aerial vehicles) assisted positioning. This paper proposes a new method of getting more accurate range information. In contrast to existing traditional works, we use multiple range information instead of a single distance information. First, a UAV transmits beacon packets to each sensor node at random positions and all nodes record RSSI (Received Signal Strength Indicator) vectors. We can estimate the distance between nodes by comparing the similarity of RSSI vectors. Second, we estimate the distance between two nodes by means of comparing their CSI (channel state Information) to UAV. Finally, we use Kalman filter to fuse the two-range information. And we can get more accurate range information for positioning. Simulations validate high localization accuracy of the proposed algorithm. Besides, the numbers of beacons transmitted by UAV and the energy consumption can be reduced in the simulation.

Low earth orbit (LEO) satellite networks can provide global coverage, low-latency services to users. The satellite networks are multi-hop networks, and the dynamic characteristics of the networks bring great challenges to design efficient routing algorithms. In this paper, we proposed an improved routing algorithm based on inter-satellite link(ISL) state information and it is on-board autonomous control. The routing algorithm is divided into three phases, namely topology establishment phase, routing calculation phase and link failure response phase. Satellites make routing decisions for each hop based on the link state information with neighbor satellites and the network topology. The simulation results show that our routing algorithm has significant advantages in delivery ratio and end-to-end delay.

This paper proposes an efficient supervised video summarization algorithm with self-attention based encoder-decoder network. Given an input video, we implement a Bi-GRU network to encode the contextual information of the video frames using self-attention mechanism, and a GRU network as the decoder, accompanying with a regression network to predict the importance score of every video frame. Experiments and analysis are conducted on the public benchmark datasets TvSum and SumMe, the results validate the superiority of our algorithm.

Although there are many simulation tools in various energy fields, due to the differences in technical characteristics in various energy fields, simulation techniques cannot be directly integrated to meet the development needs of the energy internet. Facing different application requirements, different modelling ideas and methods as well as solution algorithms are proposed in this chapter. For modelling, simulation and analysis of quasi-steady-state and long-term issues of the energy internet, a standard procedure is suggested for developing concepts or modifying existing systems, with underlying issues discussed. For the study of dynamics and transients, modelling methods of components and networks under different energy conditions are first discussed. The node equation in matrix form is used in the power grid, and the node pressure equation and branch flow equation are adopted in a thermal network and gas pipeline network. Then, considering the interweaving and interaction of long dynamic processes of wide time scale and phased evolution in the energy internet, this chapter proposes a three-layer multi-mode phased hybrid simulation framework to solve the dynamic and transient coupling and interactions between devices and adjacent networks of different energy types in the energy internet. Then, some existing simulation software suites and tools in energy engineering fields and academic circles are introduced in detail. Possible problems are discussed, including determining how to select software suites and tools with embedded models, combining different tools or developing and integrating new modules or tools. Finally, this chapter enumerates some typical component models in the energy internet and presents simulation results based on several cases. A brief analysis of a park energy internet case shows the correctness of the modelling and solution considerations mentioned above.

Precise segmentation of organs and tumors plays a crucial role in clinical applications. It is a challenging task due to the irregular shapes and various sizes of organs and tumors as well as the significant class imbalance between the anatomy of interest (AOI) and the background region. In addition, in most situation tumors and normal organs often overlap in medical images, but current approaches fail to delineate both tumors and organs accurately. To tackle such challenges, we propose a position-guided deformable UNet, namely PGD-UNet, which exploits the spatial deformation capabilities of deformable convolution to deal with the geometric transformation of both organs and tumors. Position information is explicitly encoded into the network to enhance the capabilities of deformation. Meanwhile, we introduce a new pooling module to preserve position information lost in conventional max-pooling operation. Besides, due to unclear boundaries between different structures as well as the subjectivity of annotations, labels are not necessarily accurate for medical image segmentation tasks. It may cause the overfitting of the trained network due to label noise. To address this issue, we formulate a novel loss function to suppress the influence of potential label noise on the training process. Our method was evaluated on two challenging segmentation tasks and achieved very promising segmentation accuracy in both tasks., Comment: Accepted by the 2020 International Joint Conference on Neural Networks (IJCNN 2020)

An attitude algorithm is used by a given magnetic and inertial measurement unit (MIMU) for weight fusion of the outputs of gyroscope and the outputs of the subunit constructed by accelerometer and magnetometer for attitude calculation of a rigid body. It has been proved that the weight assigned to the outputs of gyroscope determines the overall performances of attitude algorithm. A new weight adaptive regulation method based on Kalman filter and hidden Markov model is proposed, and then a new attitude algorithm is constructed by the integration of that method with recursive quaternion estimation algorithm. The authors’ attitude algorithm is compared with some other related attitude algorithms in their experiments. With the outputs of the sensors in MTi as the inputs of the new attitude algorithm, the maximum attitude estimation errors in response to three-dimensional random movements and the changes of their amplitudes are 2.85° and 6.82°, respectively, and the Allan variance of attitude drift in response to motionless conditions is 2.1×10−4°. Other attitude algorithms cannot simultaneously achieve similar dynamic and similar static performances comparing to their attitude algorithm. The results validate the overall performance improvement of their attitude algorithm.

The software-defined wireless sensor networks (SDWSNs) have been proposed recently to solve the energy limitation of sensor nodes and extend the lifetime of the wireless sensor networks by fast node reconstruction and dynamical resource allocation. In this study, the authors investigate an energy-efficient resource allocation algorithm in SDWSNs, in which radio resource allocation could be handled at central controllers with powerful storage and computation capacity. In this algorithm, the authors formulate an optimisation problem to minimise the energy consumption, under the individual constraint of quality of service. Then, the initial optimisation problem is transformed using semidefinite relaxation, to achieve centralised adaptive bandwidth and power allocation (CABPA). Additionally, two special cases are derived to reveal the performance of the CABPA. Furthermore, an OpenFlow-based scheme is proposed for information exchanging and updating to realise the centralised resource allocation. Meanwhile, a distributed scheme with limited information about the whole network is developed to serve as a performance benchmark for the CABPA in the SDWSN. Finally, the simulation results reveal that the proposed CABPA performs better than the other algorithms, and it balances the power and bandwidth utilisation.

In this paper, an overlapping coalition formation game is proposed to conduct joint interference management and resource allocation in device-to-device (D2D) communications. We aim at improving the sum rate of all D2D links in D2D communications underlying cellular network without sacrificing the quality of service of each cellular user (CU) and D2D link. First, based on the cross-tier interference level between CUs and D2D links, each D2D link selects multiple best resource blocks (RBs) to reuse. Then, by dividing the user equipments (i.e., CU and D2D links) which share the same RB into one coalition, an initial overlapping coalitional structure is formed. A cooperative game with overlapping coalitions is further proposed to maximize the system utility in terms of sum rate of all D2D links. Finally, the characteristic analysis of the proposed algorithm is provided to validate that the user equipments can cooperate to form a stable and most beneficial coalitional structure with overlapping coalitions. Simulation results show that the proposed overlapping coalition formation game achieves a better performance compared with other existing schemes.

In this study, the authors propose a semidefinite programming (SDP)-based localisation and tracking algorithm, which mitigates the non-line-of-sight (NLOS) error of range measurement and calibrates the accumulative error within the inertial sensing data. Both the range measurement in a mixed line-of-sight/NLOS environment and the step length estimated from inertial sensing information are approximated parametrically using Gaussian mixture modelling, and a maximum-likelihood estimator (MLE) is formulated to obtain the optimal position estimation. Since the Gaussian mixture models are non-linear functions of positions, the MLE is a non-convex problem, which global optimum is difficult to attain. Then, the non-convex MLE is transformed into an SDP-based localisation and tracking problem, relying on Jensen's inequality and semidefinite relaxation. Thus, a sub-optimal solution to the original MLE can be achieved. Moreover, the Cramer-Rao lower bound is also derived to serve as a performance indicator for localisation errors. The simulation and experimental results demonstrate the performance of the proposed algorithm. Compared with the existing algorithms, the proposed algorithm owns the best localisation accuracy, and can achieve a sub-metre level accuracy to a root mean square error of 0.46 m in the real deployments.

In this study, the authors propose an improved localisation algorithm in the software-defined sensor networks (SDSNs). This algorithm includes a node-selection strategy under the whole network power constraint, based on the software-defined networking (SDN) technique for providing the centralised control of the network. The analogous Cramer-Rao lower bound (A-CRLB) value is derived for each participating node, which represents a fundamental bound on the variance of the position estimator and is used to evaluate the contribution of each node to localisation accuracy. On the basis of A-CRLB values, the most helpful nodes for localisation are selected to maximise the sum of the nodes' contributory values to the localisation accuracy. With the global network knowledge provided by the SDN controller in the SDSN, the node-selection strategy is formulated into a 0-1 programming problem on the premise of power satisfaction of each node. Furthermore, the proposed node-selection -based localisation algorithm is applied to both noncooperative and cooperative localisation scenarios. Simulation results show that the proposed algorithms provide efficient and effective localisation schemes in SDSNs, and can improve the performance in terms of both the selection convergence speed and the localisation accuracy.

In this letter, the anchor scheduling scheme for localization in heterogeneous wireless sensor networks is studied. In order to minimize the number of actively participating anchors to prolong the network lifetime, we propose a centralized anchor scheduling scheme on the basis of the software-defined networking (SDN) paradigm. First, an expression evaluating the connectivity degree of an agent is derived and used to judge if this agent has desired number of connected anchors for its localization. Then, the state of each anchor is determined by the SDN controller through a flow table via sensor OpenFlow. Simulations show that the proposed anchor scheduling scheme reduces the number of active anchors and prolongs the network lifetime. It can also be shown that this scheme ensures the desired number of anchors for the localization, and can tradeoff the localization accuracy for energy by ensuring a better balance of energy consumption among minimum number of active anchors.

Considering the scenarios where the localization accuracy of the agent is required to meet a desired requirement rather than achieve the best result, it is not necessary for all nodes to participate in positioning the agent. In this paper, a reference node (RN) selection algorithm for wireless network localization under desired accuracy is proposed. A robust error bound (REB) is derived as the RN selection metric and the concept of equal-REB contour is given, based upon which the searching region (SR) for selecting RNs is defined. In REB, the measurement errors of distances are taken into consideration and modeled as a Gaussian noise whose variance is proportional to the square of the distance. The proposed RN selection strategy selects nodes from the SR instead of the whole network region iteratively until the localization accuracy meets the desired requirement. Simulations show that the RN-selection algorithm can select the RN sets providing better localization accuracy when using REB metric. Moreover, the improved performance in terms of power conservation of the proposed algorithm is evaluated through simulation results.

Since early detection of lung cancer is the most effective way to improve the survival rate of lung cancer patients, accurate detection of lung nodules is a key step in the diagnosis of lung cancer from CT images. In this paper, a deep learning-based approach is develop to detect lung nodules. First, a segmentation network is used to extract the suspected lung nodules, and then a classification network is used to perform the precise detection of the suspected lung nodules. On the one hand, according to the residual network and shortcut connection in ResNet, the gradient disappearance phenomenon and the more accurate acquisition of features can be effectively solved. We proposes a modified U-Net Block network, by applying the ResNet network to the U-Net network, improves the accuracy of segmentation of suspected lung nodules. On the other hand, multi-classification networks are combined to detect the suspected lung nodule areas, which effectively reduces the problem of false detection in single-class network, especially improving the accuracy of small nodule detection. Finally, superior performance was obtained in the evaluation of the lung nodule detection system using the scoring function, and the FROC score was 0.73.

The generic neural encoder-decoder framework for image captioning typically uses a convolution neural network to extract the image features and then uses a recurrent neural network to generate a sentence describing this image. The residual attention network is a model that achieves good results in image classification task, which is proved that this network is better than the classical convolution neural network in the ability of feature extraction. In this paper, we propose a combination of the residual attention network and the classical convolution to extract image spatial features, and then input this image spatial features to our visual attention module. At last, we use the decoder which consists of two long short-term memory(Two-LSTM) to generate a sentence describing the image. Our design scheme validates the results of BLEU-N in the MSCOCO dataset.

We propose a new reward function based on Deep Summarization Network (DSN), which is used to synthesize short video summaries to facilitate large-scale browsing of videos. The DSN uses the video summarization as a process of sequential decision making, predicting the probability of each video frame to indicate the likelihood that the video frame is selected, and then selecting the frame based on the probability distribution to form video summaries. By designing a new DSN reward function, the rewards for representative and diversity rewards are higher, and a large number of experiments are performed on the two benchmark datasets, demonstrating that our summary network is significantly better than existing unsupervised video summaries.

Simplicial complex provides a precise and tractable representation of the topology of wireless sensor networks. In this paper, a coverage optimization algorithm based on Rips complex is given for the purpose of energy conservation of wireless sensor networks. Considering an area of interest which is covered by sensor nodes completely and even superfluously, our algorithm is performed to turn off redundant sensor nodes effectively in the network while maintaining the coverage consistently. Simulation results show that this distributed algorithm can remove more than 70% internal sensor nodes, and complexity analysis for our algorithm is given.

In this paper, a node scheduling scheme for localization in heterogeneous software-defined wireless sensor networks (SD-WSNs) is proposed. An expression to evaluate the connectivity degree of the localized agent is derived, which is used to judge if the agent is connected with an expected number of anchors. The node scheduling scheme is designed on the basis of the software-defined networking (SDN) paradigm, and the state of each anchor is determined by the SDN controller through a flow table via sensor OpenFlow. In the proposed scheme, a timer for each anchor is calculated based on the Cramer-Rao lower bound (CRLB) value and the residual energy. Simulations show that the proposed node scheduling scheme can reduce the number of active nodes while ensuring an expected number of anchors for localization. It can also be shown that the scheme can reduce the energy consumption with only a slight decrease in positioning accuracy.

In this paper, we propose an energy-efficient power control algorithm (EPCA) using a game theory approach, which can be exploited for non-orthogonal multiple access (NOMA) in heterogeneous networks (HetNets). Firstly, we formulate the energy efficiency (EE) maximization problems for users in the macrocell and femtocell, respectively. Then, to reduce information exchanged between the base stations and users, we present a centralized implementation of the EPCA based on a noncooperative game. By studying the properties of the derived game, we develop convex optimization problems to deduce the unique Nash equilibrium (NE). Simulation results show that the proposed EPCA could converge to the equilibrium with higher system-level EE and spectrum efficiency.

In this paper, we propose a weighted cooperative localization algorithm with the ability to mitigate non-line-of-sight (NLOS) propagations in wireless networks. The link condition indicator (LCI) for each connection is calculated based on the amplitude and delay statistics of channel responses. We partition the ambiguity of link condition into \emph {N} levels according to the LCI values. With the distance- dependent LOS/NLOS probability suggested by the 3rd Generation Partnership Project (3GPP), the relationship between LOS/NLOS probability and the time-of-arrival (TOA) of inter-node signal transmission is derived. We incorporate this probability into \emph {N}-level LCI range regions and propose the \emph {N} probabilistic hard weight (\emph {N}-PHW) strategy for the cooperative localization, which penalizes the NLOS- induced positive biases by weighting the belief terms introduced by the conventional cooperative localization algorithm, the sum-product algorithm over a wireless network (SPAWN). Simulation results show that the proposed weighted algorithm significantly improves the localization performance in terms of localization accuracy, especially in serious NLOS environments.

In this paper, an indoor positioning algorithm based on visible light communication (VLC), combined with inertial measurement unit (IMU) is proposed to achieve the target of high-precision positioning in the indoor environment. A linearization method is proposed to estimate the quaternion errors and velocity errors with adaptive Kalman filtering (AKF). Then, the tilt angle is adopted in VLC received signal strength indicator (RSSI) propagation model to reduce the positioning errors, which is provided by IMU data. Furthermore, an indoor positioning algorithm based on VLC and IMU integration using weighted unscented Kalman filter (WUKF) is developed to obtain the final positioning solutions. Simulation results show that the proposed VLC and IMU integration algorithm can reduce the positioning solution errors to less than 0.23 m.

In this paper, we propose a localization algorithm in a visible light communication (VLC) system based on the light intensity. Firstly, by developing a VLC platform, the light intensity can be measured to derive a practical equation, which reveals the relationship between the received signal strength (RSS) and the distance from the transmitter to the receiver. However, the practical additive noise in such equation may not always follow the Gaussian distribution, thus, the noise is modeled as a Gaussian mixture model (GMM). Then, a maximum likelihood (ML) estimator is proposed to achieve the optimal localization results. Due to the nonlinearity and nonconvexity, the initial proposed estimator is relaxed into a semidefinite programming (SDP) problem which can be efficiently solved to obtain the globally optimal solution. Finally, experimental results show that the localization accuracy of proposed algorithm outperformances the existing algorithm.

Based on the field-oriented control (FOC) theory of the PMSM, the paper firstly analyzes the mathematical model of the permanent magnet linear synchronous motor (PMLSM), and then it designs the platform of the linear motor servo drive to accomplish the PMSLM operating characteristics, including both steady state and dynamic state. Comprehensive test results verify the rationality and feasibility of the theoretical analysis and simulation investigation.

In this paper, we propose a robust fingerprinting-based localization using directed graphical model. To overcome the influence caused by the jitter of received signal strength (RSS), the location of one mobile node can be estimated by fusing both the current matching result and the previous location estimation, using the Bayesian graphical model (BGM). Then, the localization problem is cast as a maximum a posteriori (MAP) estimator, which is also proved to coincide to maximum likelihood (ML) estimator. However, the initialized MAP estimator can be hardly solved with incomplete statistical property concerning the random vectors. To this end, we propose an adaptive smoothing algorithm (ASA) to attain the suboptimal solution of the original problem. Finally, the experimental results show that the proposed algorithm obtains a significant performance gain.

With the continuous development of information construction, the application of digital campus system is increasing, the network scale expanding gradually, access control, and information security problem increasingly prominent. The original decentralized independent certification, the independent authorization, the independent account management model has been unable to meet the current and future development requirements, to improve the efficiency of information utilization, office management level, realize the informatization management, used in the construction of digital campus information service platform, unified interface and the combination of the unified identity authentication system is particularly important. According to the characteristics of the development of digital campus, based on the unification of the CAS and LDAP authentication solution, and the overall architecture design, for the digital campus application system integration and lay a good foundation for further development.

In this paper, a three-dimensional (3-D) positioning system using fingerprinting and K-Nearest neighbor is proposed. The implementation of the proposed positioning system based on visible light communication (VLC) is discussed. On-off keying (OOK) modulation, Manchester coding and time division multiplexing (TDM) are utilized to obtain both lossless data and accurate received signal strength indications (RSSIs) without sacrificing the comfort of indoor illumination. The proposed 3-D positioning algorithm is conducted in three processes. Firstly, we quickly traverse all sampling points in the fingerprint database to find the nearest neighbor and choose candidate sampling points around the nearest neighbor. Secondly, we conduct an iterative search among candidate sampling points to find K-Nearest neighbors. Finally, we obtain estimated coordinates using weighted average method. Experimental result shows that, the proposed system possesses fast location and high accuracy.

This paper analyses the mathematical model of the regional multi-energy system and brings in Multifrontal Method and the nested dissection to solve the model. The model consists of linear and nonlinear algebraic equations and differential equations. The latter two kinds of equations can be finally turned into sparse linear equations. By applying Multifrontal Method, the linear matrix can be factorized in parallel instead of in serial and the consuming time can be reduced. The nested dissection can rearrange the elements of the sparse matrix to reach a high degree of parallelism and improve the performance of Multifrontal Method. Based on the 20-vertex system and the 536-vertex system which both have the similar structure to a typical regional multi-energy system, the numerical computation tests are carried out and the results indicate that the nested dissection goes well with Multifrontal Method while solving a regional multi-energy system.

The performance of an efficient and accurate face detection system depends on several issues: (1) distinctive representation for face patterns; (2) effective algorithm for feature selection and classifier learning; (3) suitable framework for rapid background removal. To address the first issue, we propose to represent face patterns with a set of heterogeneous and complementary feature descriptors including the Generalized Haar-like (GH) descriptor, Multi-Block Local Binary Patterns (MB-LBP) descriptor and Speeded-Up Robust Features (SURF) descriptor. To address the second issue, Particle Swarm Optimization (PSO) algorithm is incorporated into the Adaboost framework, replacing the exhaustive search used in original Adaboost for efficient feature selection. The utilization of heterogeneous feature descriptors enriches the diversity of feature types for Adaboost learning algorithm. As a result, classification performance of the boosted ensemble classifier also improves significantly. A three-stage hierarchical classifier structure is proposed to tackle the last issue. In particular, a new stage is added to detect candidate face regions more quickly by using a large size window with a large moving step. Nonlinear support vector machine (SVM) classifiers are used instead of decision stump classifiers in the last stage to remove those remaining complex non-face patterns that cannot be rejected in the previous two stages. Combining the abovementioned effective modules, we derive the proposed Hetero-PSO-Adaboost-SVM face detector that achieves superior detection accuracy while maintaining a low training and detection complexity. Extensive experiments demonstrate the robustness and efficiency of our system by comparing it with several popular state-of-the-art algorithms on our own test set as well as the widely used CMU+MIT frontal and CMU profile face dataset.

In this paper, we propose a modified cooperative localization algorithm based on the sum-product algorithm over a wireless network (SPAWN) in a mixed line-of-sight (LOS) and non-line-of-sight (NLOS) scenario. To alleviate the practical limitations of high computational complexity and increased network traffic, we design a node selection method for the cooperative localization algorithm. This method chooses the nodes with highest utilities which are indicated by the contribution each node makes and the maintenance cost it consumes. In addition, considering the positively biased distance errors caused by NLOS connections, we use a link condition indicator (LCI) to quantify the link quality of selected nodes corresponding to the LOS/NLOS conditions. By assigning different weights to the links according to this indicator, we penalize the NLOS measurements and mitigate the biases in the proposed cooperative localization algorithm. The simulation results demonstrate that this algorithm can achieve significant performance improvement compared with that without complexity reduction and NLOS mitigation.

In this paper, one centralized resource allocation algorithm is proposed to minimize energy consumption in software defined wireless sensor networks (SD-WSNs). The energy consumption problem is formulated as an optimization problem, given quality-of-service (QoS) constraint defined as Signal-to-Interference-plus-Noise Ratio (SINR). Then, the nonconvex optimization problem is relaxed into a semidefinite programming (SDP), which serves as a lower bound. To analyze the tightness of the lower bound, two special cases are introduced. Besides, one distributed approach is also developed to provide a performance benchmark. Furthermore, simulation results are revealed that the proposed centralized algorithm performances better with respect to the energy consumption and bandwidth utilization.

Ultra-wide bandwidth (UWB) and Inertial Navigation (IN) have been adopted in high precision localization and tracking systems. However, the ranging measurements influenced by non-line-of-sight (NLOS) path may degrade the localization accuracy. Besides, inertial measurement errors are within the sensing data and constantly accumulate over the time. To overcome the above problems, we investigate NLOS mitigation for UWB measurements, and calibration for IN estimations. Both ranging measurement and step length estimation are modeled using Gaussian mixture model (GMM), and one maximum likelihood (ML) estimator is developed. Then, the non-convexity estimator is relaxed into a semidefinite programming (SDP), which global minimum can be readily attained. Finally, both simulation and experimental results are provided to illustrate the validity and performance of our proposed SDP-based localization and tracking algorithm.

Schemes combining Ultra-wide bandwidth (UWB) ranging technology and Inertial Measurement Unit (IMU) have been proposed for high precision positioning and tracking. However, positioning accuracy can be significantly affected by the non-line-of-sight (NLOS) UWB ranging measurements and cumulative inertial sensing error. In this paper, we model the ranging measurement error and the step length as Gaussian Mixture Model (GMM), respectively. Then, we derived a Suboptimal Importance Density (SID) for particle filters, which could resolve the degeneracy of particles and sample impoverishment. Finally, experimental results illustrate the performance gain of the particle filters with the proposed SID.

Big data publishing anonymity technology is the main way to achieve privacy protection, it is a research hotspot of data publishing privacy protection. Applying anonymity protection to data is an effective way to protect the privacy of data subject in the process of micro data publishing. This paper analyzes the data publishing privacy protection technology in the anonymous models and main technologies, induces and summarizes the advantages and disadvantages of the anonymization algorithms and the existing problems. Finally, the development direction of privacy protection anonymity technology is prospected.

A novel design strategy of an unknown input observer (UIO) based on the radial basis function (RBF) neural network is proposed in this paper for the problem of intermittent fault estimation in the satellite attitude control system subject to the environment disturbance. The unknown input observer is employed to eliminate the interference of the space disturbance. At the meantime, due to the learning ability and nonlinear approximating ability of the radial basis function neural network, the observer is combined with the RBF neural network, so as to estimate the intermittent fault of the satellite attitude control system. Furthermore, a Lyapunov function is used for analyzing the stability of the proposed observer. Finally, the simulation on a closed-loop satellite attitude control system with the environment disturbance demonstrates the high performance of the proposed intermittent fault estimation method both for abrupt and soft faults in the form of additive dynamics.

Localization technologies in wireless sensor networks have been suffering from great energy consumption problem due to the energy-constrained characteristic of the networks. Existing power allocating solutions are mostly distributed for the lack of global network knowledge. In this paper, we investigate localization algorithm with the support of software-defined network (SDN) technique and propose a localization node selection algorithm based on the cramer-rao lower bound (CRLB). By making use of the global network knowledge provided by the SDN controller, we formulate the issues into a 0-1 programming problem on the premise of energy satisfaction. Simulation results show that significant improvement in localization performance can be achieved with our proposed SDN based algorithm.

Minimizing user-perceived latency is crucial for Content Distribution Networks (CDNs) hosting interactive services. Latency may increase for many reasons, such as interdomain routing changes and the CDN's own load-balancing policies. CDNs need greater visibility into the causes of latency increases, so they can adapt by directing traffic to different servers or paths. In this paper, we propose a tool for CDNs to diagnose large latency increases, based on passive measurements of performance, traffic, and routing. Separating the many causes from the effects is challenging. We propose a decision tree for classifying latency changes, and determine how to distinguish traffic shifts from increases in latency for existing servers, routers, and paths. Another challenge is that network operators group related clients to reduce measurement and control overhead, but the clients in a region may use multiple servers and paths during a measurement interval. We propose metrics that quantify the latency contributions across sets of servers and routers. Based on the design, we implement the LatLong tool for diagnosing large latency increases for CDN. We use LatLong to analyze a month of data from Google's CDN, and find that nearly 1% of the daily latency changes increase delay by more than 100 msec. Note that the latency increase of 100 msec is significant, since these are daily averages over groups of clients, and we only focus on latency-sensitive traffic for our study. More than 40% of these increases coincide with interdomain routing changes, and more than one-third involve a shift in traffic to different servers. This is the first work to diagnose latency problems in a large, operational CDN from purely passive measurements. Through case studies of individual events, we identify research challenges for managing wide-area latency for CDNs.

Many network-management problems in large backbone networks need the answer to a seemingly simple question: where does a given IP packet, entering the network at a particular place and time, leave the network to continue on its path to the destination? Answering this question at scale and in real time is challenging for several reasons: (i) a destination IP address could match several IP prefixes, (ii) the longest-matching prefix may change over time, (iii) the number of IP prefixes and routing protocol messages is very large, and (iv) network-management applications often require answers to this question for a large number of destination IP addresses in real time. In this paper, we present an efficient algorithm for tracking prefix-match changes for ranges of IP addresses. We then present the design, implementation, and evaluation of the Route Oracle tool that answers queries about routing changes on behalf of network management applications. Our design of Route Oracle includes several performance optimizations, such as pre-processing of BGP update messages, and parallelization of query processing. Experiments with BGP measurement data from a large ISP backbone demonstrate that our system answers queries in real time and at scale.

Conventional wisdom has held that routing protocols cannot achieve both scalability and high availability. Despite scaling relatively well, today's Internet routing system does not react quickly to changing network conditions (e.g., link failures or excessive congestion). Overlay networks, on the other hand, can respond quickly to changing network conditions, but their reliance on aggressive probing does not scale to large topologies. The paper presents a layered routing architecture called UFO (Underlay Fused with Overlays), which achieves the best of both worlds by having the "underlay" provide explicit notification about network conditions to help improve the efficiency and scalability of routing overlays.

The paper illustrates to integrate UWB technology with inertial navigation, and introduces the design of the hybrid indoor positioning system including server, gateway and sensor nodes. A linear regression matrix model is proposed to modify inertial navigation path. Least squares method is used to determine equations relating UWB trajectories to inertial navigation trajectories. Experimental results are shown to compare error distributions of hybrid algorithm and pure UWB solution.

This paper introduces an indoor positioning system which is built with heterogeneous sensors, i.e., inertial sensors and ultra wide band (UWB) ranging modules. It takes a hybrid approach with Inertial Navigation and UWB ranging method to explore their cooperative efforts. Besides, a novel hybrid algorithm for single-anchor positioning and calibration is proposed, which combines real-time inertial sensing data and UWB ranging measurements from single anchor. Additionally, the proposed hybrid algorithm is based on MMSE of user displacement, providing a significant deviation reducing for tracking. Finally, one experimentation is conducted to show that the hybrid algorithm can improve the accuracy of positioning by 47.2% compared to pure inertial solution, meanwhile its mean of the location errors is less than 1m. With the popularity of wearable devices with inertial sensors and wireless communication chips, such hybrid approach could be very promising for indoor positioning system.

Channel measurements and path-loss modelling for typical indoor environments at 28-GHz band, in both line-of-sight and non-line-of-sight scenarios, are conducted. An automatic measurement system is designed and used, with multiple combinations of steerable transmit and receive antennas at both azimuth and elevation planes. Based on the measured data, the received signal is combined in plural, in azimuth and elevation plane, at both transmit and receive sides. Therefore the influence caused by the radiation pattern could be offset. Then, the path-loss is modeled as the product of a distance dependent term, and the standard deviation of shadow is also discussed.

Context cue is important in object classification. By embedding the depth context cue of image attributes into kernel descriptors, we propose a new set of depth image descriptors called depth context kernel descriptors (DCKD) for RGB-D based object classification. The motivation of DCKD is to use the depth consistency of image attributes defined within a neighboring region to improve the robustness of descriptor matching in the kernel space. Moreover, a novel joint spatial-depth pooling (JSDP) scheme, which further partitions image sub-regions using the depth cue and pools features in both 2D image plane and the depth direction, is developed to take full advantage of the available depth information. By embedding DCKD and JSDP into the standard object classification pipeline, we achieve superior performance to state-of-the-art methods on RGB-D benchmarks for object classification and scene recognition.