Search

Convolutional neural networks (CNNs) based super-resolution (SR) methods have achieved remarkable progress in recent years. Most of these methods assume that the degradation is known and fixed, such as bicubic downsampling. However, the performance of CNNs-based methods suffers from a severe drop when the actual degradation mismatch the training one. This paper proposes a prior correction network (PCNet) to solve the blind SR problem, which makes CNNs-based super-resolvers trained in a fixed blur kernel but applied to other unknown blur kernels. The PCNet consists of a kernel estimation network, a correction filter module, and a correction refinement network. The kernel estimation network aims to estimate unknown blur kernel from the input low-resolution (LR) image. The correction filter module then transfers the estimated degraded domains to adapt to specific degradations (e.g., bicubic downsampling). Finally, the correction refinement network adjusts the corrected LR image to eliminate the influence of blur kernel mismatch or misestimate. Experimental results on diverse datasets show that the proposed PCNet, combined with existing CNNs-based SR methods, outperforms other state-of-the-art algorithms for blind SR. Code is available at: \url{ https://github.com/caoxiang104/PCNet }.

Recently, depth map super-resolution (DSR) has obtained remarkable performance with the development of convolutional neural networks (CNNs). High-resolution (HR) depth map can be inferred from a low-resolution (LR) one with the guidance of its corresponding HR intensity image. However, most of the existing CNNs-based methods unilaterally transfer structures information of guidance image to the input depth map, which ignores the corresponding relations between the depth map and the intensity map. In this paper, we propose a novel dual auto-encoder attention network (DAEANet) for DSR. The proposed DAEANet includes two auto-encoder networks, where guidance auto-encoder network (GAENet) and target auto-encoder network (TAENet) aim to extract feature information from intensity image and depth map. Specifically, all auto-encoder networks are similar and trained simultaneously to ensure structural consistency. Furthermore, to preserve the structure information in the process of training, the attention mechanism is employed to our DAEANet. Extensive experiments on several popular benchmarks show that the proposed DAEANet outperforms existing state-of-the-art algorithms.

In the research of multi-robot systems, multi-AUV (multiple autonomous underwater vehicles) cooperative target hunting is a hot issue. In order to improve the target hunting efficiency of multi-AUV, a multi-AUV hunting algorithm based on dynamic prediction for the trajectory of the moving target is proposed in this article. Firstly, with moving of the target, sample points are updated dynamically to predict the possible position of a target in a short period time by using the fitting of a polynomial, and the safe domain of the moving target, which is a denied area for the hunting AUVs, is built to avoid the target's escape when it detects AUVs. Secondly, the method of negotiation is adopted to allocate appropriate desired hunting points for each AUV. Finally, the AUVs arrive at desired hunting points rapidly through deep reinforcement learning (DRL) algorithm to achieve hunting the moving target. The simulations show that hunting AUVs can surround the moving target of which the trajectory is unknown rapidly and accurately by the algorithm in the 3D environment with complex obstacles and results obtained is satisfactory.

In order to accomplish the target hunting by multi-AUV in 3-D underwater environments, the AUVs need to cooperate in the process of pursuing targets. To improve the efficiency of target hunting and...

Due to the complexity of underwater environment, the navigation of autonomous underwater vehicle (AUV) is affected by current. Although some traditional methods can plan the non-collision path for AUV to reach the destination, it cannot overcome the influence of current on AUV navigation. In order to solve this problem, a velocity vector synthesis method combining belief function (BF) method is proposed. Among them, the BF method is used to dynamically plan the non-collision path. The velocity vector synthesis method is used to adjust the AUVs movement direction to offset the effect of time-varying current. The simulation results show that the proposed algorithm is effective and can overcome the problem of ‘current influence’. At the same time, compared with other methods, the proposed algorithm can guide AUV to achieve more efficient and adaptive path planning.

The autonomous underwater vehicle (AUV) is widely used to search for unknown targets in the complex underwater environment. Due to the unpredictability of the underwater environment, this paper combines the traditional frontier exploration method with deep reinforcement learning (DRL) to enable the AUV to explore the unknown underwater environment autonomously. In this paper, a grid map of the search environment is built by the grid method. The designed asynchronous advantage actor-critic (A3C) network structure is used in the traditional frontier exploration method for target search tasks. This network structure enables the AUV to learn from its own experience and generate search strategies for the various unknown environment. At the same time, DRL and dual-stream Q-learning algorithms are used for AUV navigation to further optimize the search path. The simulations and experiments in an unknown underwater environment with different layouts show that the proposed algorithm can accomplish target search tasks with a high success rate, and it can adapt to different environments. In addition, compared to other search methods, the frontier exploration algorithm based on DRL can search a wider environment faster, which results in a higher search efficiency and reduced search time.

Intelligent Transportation Systems (ITS) have been an important research area focused on improving the efficiency of future transportation systems. In an ITS, vehicular networks are used to collect and transmit data from various points of interest. In these vehicular networks, the Roadside Unit (RSU) is a key infrastructure component facilitating communications among all elements of the transportation system.Inspired by the concept of edge computing, we believe RSUs can store data from points of interest and perform certain pre-processing tasks before further uploading the data to the Cloud. It is important to carefully deploy RSUs in vehicular networks in order to ensure full coverage of data generators. However, RSUs have limited data storage capacities and restricted communication ranges. In this paper, we jointly consider these two aspects with a goal of deploying the minimal number of RSUs while still covering all points of interest. Performance evaluation demonstrates our solution improves coverage performance under these constraints.

Edge computing is a novel computing paradigm that can improve the performance of many data intensive applications. Edge servers offer computing and data storage services closer to the locations of data sources. Data storage is an important issue in edge computing. Data need to be carefully allocated and stored in the system. In this paper, we investigate where to store data from data sources. We consider the scenario that data sources upload their data to edge servers or to the cloud. We show the impact of data storage and point out three goals, i.e., maximizing total data access frequencies supported by the edge server, minimizing total data uploading time, and maximizing the number of data sources uploading data to the edge server. We explore the tradeoff among these goals and propose different solutions. Performance evaluation shows these three goals can be achieved by our solutions respectively.

Recently spiking neural networks (SNNs), the third-generation of neural networks has shown remarkable capabilities of energy-efficient computing, which is a promising alternative for deep neural networks (DNNs) with high energy consumption. SNNs have reached competitive results compared to DNNs in relatively simple tasks and small datasets such as image classification and MNIST/CIFAR, while few studies on more challenging vision tasks on complex datasets. In this paper, we focus on extending deep SNNs to object tracking, a more advanced vision task with embedded applications and energy-saving requirements, and present a spike-based Siamese network called SiamSNN. Specifically, we propose an optimized hybrid similarity estimation method to exploit temporal information in the SNNs, and introduce a novel two-status coding scheme to optimize the temporal distribution of output spike trains for further improvements. SiamSNN is the first deep SNN tracker that achieves short latency and low precision loss on the visual object tracking benchmarks OTB2013/2015, VOT2016/2018, and GOT-10k. Moreover, SiamSNN achieves notably low energy consumption and real-time on Neuromorphic chip TrueNorth.

Currently, the “2019-CoV-2” has been raging across the world for months, causing massive death, huge panic, chaos, and immeasurable economic loss. Such emerging epidemic viruses come again and again over years, leading to similar destructive consequences. Air-borne transmission among humans is the main reason for the rapid spreading of the virus. Blocking the air-borne transmission should be a significant measure to suppress the spreading of the pandemic. Considering the hospital is the most “dangerous” place to occur massive cross-infection among patients as emerging virus usually comes in a disguised way, an air distribution optimization of a general three-bed hospital ward in China is carried out in this paper. Using the Eulerian-Lagrangian method, sneeze process from patients who are assumed to be the virus carrier, which is responsible for a common event to trigger cross-infection, is simulated. The trajectory of the released toxic particle and the probability of approaching others in the same ward are calculated. Two evaluation parameter, total maximum time (TMT) and overall particle concentration (OPC) to reflect the particle mobility and probability to cause cross-infection respectively, are developed to evaluate the proposed ten air distributions in this paper. A relatively optimized air distribution proposal with the lowest TMT and OPC is distinguished through a three-stage actively analysis. Results show that a bottom-in and top-out air distribution proposal is recommended to minimize the cross-infection., Highlights • A ward-based sneeze process from a patient with epidemic disease is modeled. • Air distribution (AD) affects the flow pattern of the sneeze-induced virus particles. • Ten AD proposals are developed to evaluate the their probability of cross-infection. • Two parameters are proposed to judge which proposals is better quantitatively. • The direct probability of cross-infection can be decreased by 95% via optimization.

Data centers are important platforms and resources for many data intensive applications. A data center typically consists of many servers, which store a huge amount of data and perform computing tasks. Initial data needs to be transferred to the data center for users to access. It is usually necessary to replicate data for better reliability and fault tolerance. In this paper, we consider that the data center loads data from external storage devices. We investigate how to allocate and replicate data. We particularly emphasize the impact of data replication and present three goals, i.e., maximizing total data access frequencies, maximizing total replicas, and balancing the number of replicas. We propose different solutions to explore the tradeoff. Performance evaluation shows these solutions can handle various goals well and efficiently replicate data.

Multiple autonomous underwater vehicles (multi-AUV) target search is the important element to realise underwater rescue, underwater detection. To improve the efficiency of the multi-AUV target sear...

If the associated differential inter-system biases (DISBs) are priori known, only one common reference satellite is sufficient, which is called the inter-system model. The inter-system model can help to maximize the redundancy of the positioning model, and thus can improve the positioning performance, especially in harsh environment. However, in practice use not all receivers can be calibrated with DISBs in advance. In this paper, taking combined GPS and BDS pseudorange positioning as the example, we compare three positioning models and their positioning performance. One is traditional intra-system model, and the other two belongs to the inter-system models, i. e. the model with calibration of DISB and the model with real-time estimation of DISB parameter. Positioning performance using the three models is evaluated with simulated obstructed environments. It will be shown that besides the model with calibration of DISB, the model with real-time estimation of DISB parameter can also effectively improve positioning accuracy and reliability compared with the traditional intra-system model, especially for the severely obstructed environment with only a few satellites observed. When no more than 7 satellites visible, the positioning accuracies in each directions can be improved by no less than 15 %. The proposed model can be used alternatively when no priori DISB calibration is available.

For target search and tracking in unknown underwater environment, an integrated algorithm for a cooperative team of multiple autonomous underwater vehicles (Multi-AUV) is proposed by combining the Glasius bio-inspired neural network (GBNN) and bio-inspired cascaded tracking control approach to improve search efficiency and reduce tracking errors. Among them, the GBNN is mainly used to control a multi-AUV team in search of each targets. Once any target is found, the bio-inspired cascaded tracking control approach is applied to track it in case that it may escape. This integrated algorithm deals with various situations such as search for static or dynamic targets, and tracking of different trajectory in underwater environments with obstacles. The simulation results show that this integrated algorithm is of high efficiency and adaptability.

For natural language processing problems, the short text classification is still a research hot topic, with obviously problem in the features sparse, high-dimensional text data and feature representation. In order to express text directly, a simple but new variation which employs one-hot with low-dimension was proposed. In this paper, a Densenet-based model was proposed to short text classification. Furthermore, the feature diversity and reuse were implemented by the concat and average shuffle operation between Resnet and Densenet for enlarging short text feature selection. Finally, some benchmarks were introduced to evaluate the Falcon. From our experimental results, the Falcon method obtained significant improvements in the state-of-art models on most of them in all respects, especially in the first experiment of error rate. To sum up, the Falcon is an efficient and economical model, whilst requiring less computation to achieve high performance.

The control design of target search and hunting using multi-robot remains a challenge in recent years. In this paper, we propose a control algorithm of multi-robot for target search and hunting inspired by potential field-based particle swarm optimization (PPSO). Firstly, a potential field function is established according to the initial positions of the obstacles, un-search area and targets. Then, the fitness function of PSO's (particle swarm optimization) is determined by the potential function of the work area. Lastly, multi-robot start performing target search and hunting missions driven by the proposed PPSO algorithm. Simulation results demonstrate that the PPSO algorithm is applicable and feasible for multi-robot cooperation to search and hunting targets. Compared with other commonly used methods for control of multi-robot, simulation results indicate that the PPSO algorithm has more stability and higher efficiency.

To improve the efficiency of multiple Autonomous Underwater Vehicles (multi-AUV) cooperative target search in a Three-Dimensional (3D) underwater workspace, an integrated algorithm is proposed by combining a Self-Organising Map (SOM), neural network and Glasius Bioinspired Neural Network (GBNN). With this integrated algorithm, the 3D underwater workspace is first divided into subspaces dependent on the abilities of the AUV team members. After that, tasks are allocated to each subspace for an AUV by SOM. Finally, AUVs move to the assigned subspace in the shortest way and start their search task by GBNN. This integrated algorithm, by avoiding overlapping search paths and raising the coverage rate, can reduce energy consumption of the whole multi-AUV system. The simulation results show that the proposed algorithm is capable of guiding multi-AUV to achieve a multiple target search task with higher efficiency and adaptability compared with a more traditional bioinspired neural network algorithm.

Isomerization of surface-exposed aspartic acid (Asp) in the complementarity-determining regions of therapeutic proteins could potentially impact their target binding affinity because of the sensitive location, and often requires complex analytical tactics to understand its effect on structure-function and stability. Inaccurate quantitation of Asp-isomerized variants, especially the succinimide intermediate, presents major challenge in understanding Asp degradation kinetics, its stability, and consequently establishing a robust control strategy. As a practical solution to this problem, a comprehensive analytical tool kit has been developed, which provides a solution to fully characterize and accurately quantify the Asp-related product variants. The toolkit offers a combination of 2 steps, an ion-exchange chromatography method to separate and enrich the isomerized variants in the folded structure for structure-function evaluation and a novel focused peptide mapping method to quantify the individual complementarity-determining region isomerization components including the unmodified Asp, succinimide, and isoaspartate. This novel procedure allowed an accurate quantification of each Asp-related variant and a comprehensive assessment of the functional impact of Asp isomerization, which ultimately helped to establish an appropriate control strategy for this critical quality attribute.

Data centers have been prevalent resources for many years in data intensive applications. Typically, there are multiple servers in a data center to store data and perform computing. One of the last steps of provisioning a data center is to populate the servers with data.In this paper, we consider the scenario that data are transferred from external storage devices. We investigate data placement strategy with the consideration of two competing objectives for data transfer, reducing completion time and enhancing spatial locality. We propose a heuristic solution to jointly consider these two objectives and explore their tradeoff. Performance evaluation shows that with an adjustable balance ratio, a flexible data transfer performance can be achieved.

Most researches on EEG emotion recognition are based on multi-electrode EEG signal acquisition equipment. There are few studies on single-electrode EEG equipment and the classification accuracy obtained is not ideal. In order to further improve the classification accuracy, the paper proposes an EEG emotion recognition method based on the combination of multiple features in the time-frequency domain. Firstly, the wavelet transform is used to decompose the EEG signals under the three types of emotion labels and extracts the time-frequency information of four frequency bands, including alpha, low beta, high beta, and gamma, and then uses different sizes of sliding time windows to calculate separately their statistical characteristics in the time-frequency domain. Finally, the Long Short-Term Memory (LSTM) network with time-frequency domain characteristics is used to extract the sequence information of the deep features and combine the output results of the softmax classifier. The results of experiments show that the average accuracy of the proposed method is 93.09% and 98.36% on emotion food and emotion state dataset, respectively. Compared with traditional machine learning methods and other deep learning methods, the time-frequency LSTM model named TF-LSTM in this paper has better EEG generalization ability and classification performance, and also provides a new feasible scheme for emotion recognition research based on single-electrode EEG signal.

Target search in 3-D underwater environments is a challenge in multiple autonomous underwater vehicles (multi-AUVs) exploration. This paper focuses on an effective strategy for multi-AUV target search in the 3-D underwater environments with obstacles. First, the Dempster-Shafer theory of evidence is applied to extract information of environment from the sonar data to build a grid map of the underwater environments. Second, a topologically organized bioinspired neurodynamics model based on the grid map is constructed to represent the dynamic environment. The target globally attracts the AUVs through the dynamic neural activity landscape of the model, while the obstacles locally push the AUVs away to avoid collision. Finally, the AUVs plan their search path to the targets autonomously by a steepest gradient descent rule. The proposed algorithm deals with various situations, such as static targets search, dynamic targets search, and one or several AUVs break down in the 3-D underwater environments with obstacles. The simulation results show that the proposed algorithm is capable of guiding multi-AUV to achieve search task of multiple targets with higher efficiency and adaptability compared with other algorithms.

An integrated multiple autonomous underwater vehicles (multi-AUV) dynamic task assignment and path planning algorithm is proposed for three-dimensional underwater workspace with ocean current. The ...

NoSQL databases are flexible and efficient for many data intensive applications, and the key-value store is one of them. In recent years, a new Ethernet accessed disk drive called the &ldquo, Kinetic Drive&rdquo, was developed by Seagate. This new Kinetic Drive is specially designed for key-value stores. Users can directly access data with a Kinetic Drive via its IP address without going through a storage server/layer. With this new innovation, the storage stack and architectures of key-value store systems have been greatly changed. In this paper, we propose a novel global key-value store system based on Kinetic Drives. We explore data management issues including data access, key indexing, data backup, and recovery. We offer scalable solutions with small storage overhead. The performance evaluation shows that our location-aware design and backup approach can reduce the average distance traveled for data access requests.

Various faults of photovoltaic (PV) modules inevitably occur in the work process, since PV modules are installed in hostile situation. To obtain the types of failure, a novel fault diagnosis method based on back propagation (BP) neural network with Levenberg-Marquardt (L-M) algorithm for PV modules is proposed. Through the in-depth analysis the output of PV modules under normal and fault conditions, the input variables of the diagnosis model are acquired. The high-speed and real-time fault diagnosis model for PV modules is first designed based on TMS320VC5402 DSP and long-distance wireless fault diagnosis is realized by Zigbee technology. The simulation and experimental results show that the fault diagnosis method for PV modules based on BP network with L-M algorithm can effectively detect four types of fault for PV modules such as open circuit, short circuit, partial shading and abnormal degradation. The numerical results verify the effectiveness and correctness of the proposed method, which can provide a great educational benefit of PV operation technology.

Meeting latency SLOs (Service Level Objectives) in a networked storage environment is essential while challenging. In this environment, a storage request has to go through client I/O stacks, dynamically changing networks, and the storage system attached to a server. Its response also has to traverse all the way back to the client. Along this long I/O path, any of these components can become congested. The behavior of one component may affect the performance of the others. Isolated control on each component is not effective to meet latency SLOs of storage requests. In this paper, we propose and implement JoiNS, a system trying to guarantee latency SLO for applications that access data on a remote networked storage. JoiNS carefully considers all the components along the I/O path and controls them in a coordinated fashion. JoiNS has both global network and storage visibilities with a logically centralized controller which keeps monitoring the status of each involved component. JoiNS coordinates these components and adjusts the priority of I/O packets in each component based on the latency SLO, network and storage status, time estimation, and characteristics of each I/O request. We integrate Software Defined Network(SDN) into our system to coordinate with storage. Our evaluation shows JoiNS can achieve up to 6X speedup in this networked storage environment with various loads of background traffic.

The path planning problem for autonomous underwater vehicle (AUV) is to generate real-time collision-free trajectories from beginning to end in an underwater environment. Although in the past few decades the study of AUV path planning has obtained many achievements, but many methods only consider the optimal path of AUV, while ignoring the can't get too close to the obstacles in the practical underwater environment. Another problem is that the generated path may not meet the limitation that the length of the path should be minimized when the method is applied in point-to-point path planning. In order to solve these problems, a potential field bio-inspired neural network (PBNN) method is proposed. The bio-inspired neural network approach can be used to plan real-time collision free paths. The potential field introduced by the proposed approach can plan a real-time "comfortable" trajectory by overcoming the "too close" shortcomings. The proposed approach of introducing attractive potential field in point to point path planning can optimize the path. Simulation studies verify that the proposed method can safely and effectively navigate of an AUV.

The intervisibility judgment algorithm is one of the most important algorithms in military simulation system. In view of the problem that the existing intervisibility algorithms are inefficient and do not consider the influence of the curvature of the earth. This paper improves the existing intervisibility algorithm through calculating the slope of the connection line of the sight point and target point before selecting sampling points. According to the different slopes, different sampling strategies are adopted, and the slope determination method is used to replace the original altitude data determination method, in addition, the influence of the curvature of the earth is also incorporated into the intervisibility judgment algorithm. The improved algorithm proposed in this paper avoids complicated interpolation calculation process, and the sampling process is more flexible. The experimental results show that the improved intervisibility algorithm can effectively improve the operating efficiency, shorten the system runtime, and improve the accuracy of the intervisibility judgment.

As a carrier and transmission channel of information, the communication system is a sub-part of the combat simulation system. The modeling of communication systems is an important part of the combat simulation process. In view of the deficiencies in the existing modeling and simulation of communication systems, the model architecture of military communication systems in combat simulation is presented in this paper. Its composition and function are described. A general modeling framework for oriented-combat communication simulation system is proposed and the corresponding implementation of the modeling framework is given. The general modeling framework can effectively support the construction and application of communication systems in combat simulation, and has good scalability and flexibility.

In this paper, target hunting aims to detect target and surround the detected target using multi-AUV (Multiple Autonomous Underwater Vehicles) in a given area. The main challenge in multi-AUV target hunting is the design of AUV's motion path and coordination mechanism. To conduct the cooperative target hunting by multi-AUV in underwater environment, an integrated algorithm based on improved potential field (IPF) is proposed. Firstly, a potential field function is established according to the information of underwater environment. Then, the dispersion degree, the homodromous degree and district-difference degree are introduced to increase the cooperation of the multi-AUV system. Lastly, the target hunting is solved by embeds the three kinds of degree into the potential field function. The simulation results show that the proposed approach is applicable and feasible for multi-AUV cooperative target hunting.

High-resolution (HR) depth map can be better inferred from a low-resolution (LR) one with the guidance of an additional HR texture map of the same scene. Recently, deep neural networks with large receptive fields are shown to benefit applications such as image completion. Our insight is that super resolution (SR) is similar to image completion, where only parts of the depth values are precisely known. However, large receptive fields in general will increase the depth and the number of parameters of the network, which may cause degradation and large memory consumption. To solve these problems, we adapt the convolutional neural pyramid (CNP) structure by introducing residual block and linear interpolation layer, and adopt the CNP in the joint super-resolution framework. We call this convolutional neural model joint residual pyramid (JRP). Our JRP model consists of three sub-networks, two convolutional neural residual pyramids concatenated by a normal convolutional neural network. The convolutional neural residual pyramids extract information from large receptive fields of the depth map and guidance map, while the convolutional neural network effectively transfers useful structures of the guidance image to the depth image. Experimental results show that our model outperforms existing state-of-the-art algorithms not only on data pairs of RGB/depth images, but also on other data pairs like color/saliency and color-scribbles/colorized images.

With the rise of cloud storage and many data intensive applications, there is an unprecedented growth in the volume of unstructured data. In response, key-value object storage is becoming more popular for the ease with which it can store, manage, and retrieve large amounts of this data. Seagate recently launched Kinetic direct-access-over-Ethernet hard drives which incorporate a LevelDB key-value store inside each drive. In this work, we evaluate these drives using micro as well as macro benchmarks to help understand the performance limits, trade-offs, and implications of replacing traditional hard drives with Kinetic drives in data centers and high performance systems. We perform in-depth throughput and latency benchmarking of these Kinetic drives (each acting as a tiny independent server) from a client machine connected to them via Ethernet. We compare these results to a SATA-based and a faster SAS-based traditional server running LevelDB. Our sample Kinetic drives are CPU-bound, but they still average sequential write throughput of 63 MB/sec and sequential read throughput of 78 MB/sec for 1 MB value sizes. They also demonstrate unique Kinetic features including direct disk-to-disk data transfer. Our macro benchmarking using the Yahoo Cloud Serving Benchmark (YCSB) shows that mid-range LevelDB servers outperform the Kinetic drives for several workloads; however, this is not always the case. For larger value sizes, even these first generation sample Kinetic drives outperform a full server for several different workloads.

In this paper, the problem of spatial signature estimation using a uniform linear array (ULA) with unknown sensor gain and phase errors is considered. As is well known, the directions-of-arrival (DOAs) can only be determined within an unknown rotational angle in this array model. However, the phase ambiguity has no impact on the identification of the spatial signature. Two auto-calibration methods are presented for spatial signature estimation. In our methods, the rotational DOAs and model error parameters are firstly obtained, and the spatial signature is subsequently calculated. The first method extracts two subarrays from the ULA to construct an estimator, and the elements of the array can be used several times in one subarray. The other fully exploits multiple invariances in the interior of the sensor array, and a multidimensional nonlinear problem is formulated. A Gauss–Newton iterative algorithm is applied for solving it. The first method can provide excellent initial inputs for the second one. The effectiveness of the proposed algorithms is demonstrated by several simulation results.

Ocean currents impose a negative effect on Autonomous Underwater Vehicle (AUV) underwater target searches, which lengthens the search paths and consumes more energy and team effort. To solve this problem, an integrated algorithm is proposed to realise multi-AUV cooperative search in dynamic underwater environments with ocean currents. The proposed integrated algorithm combines the Biological Inspired Neurodynamics Model (BINM) and Velocity Synthesis (VS) method. Firstly, the BINM guides a team of AUVs to achieve target search in underwater environments; BINM search requires no specimen learning information and is thus easier to apply to practice, but the search path is longer because of the influence of ocean current. Next the VS algorithm offsets the effect of ocean current, and it is applied to optimise the search path for each AUV. Lastly, to demonstrate the effectiveness of the proposed integrated approach, simulation results are given in this paper. It is proved that this integrated algorithm can plan shorter search paths and thus the energy consumption is lower compared with BINM.

In this paper, we propose a FPGA based intelligent desk lamp system with a camera, two servo motors and LEDs. The proposed lamp system can automatically trace a moving book with a red bookmark by pan and tilt control to provide sufficient illumination. Thus, the user can read book with enough light for eye caring. The architecture of the system includes a vision system and a servo-motor control system. The image processing module inside the system can obtain the book’s position data. According to the book’s position data, two servo motors automatically adjust the pan and tilt angles of the desk lamp. The adjustable ranges for pan and tilt angles are 40° and 30°, respectively. Additionally, an App is designed and allows the user to remotely control the lamp by using a smartphone. All the processing modules are implemented on Altera DE0-Nano kit with Cyclonetex® IV EP4CE22F17C6N FPGA with Verilog HDL.

Criminals are using every means available to launder the profits from their illegal activities into ostensibly "legitimate" assets. Meanwhile, most commercial anti-money laundering systems are still rule-based, which cannot adapt to the ever-changing tricks. Although some machine learning methods have been proposed, they are mainly focused on the perspective of abnormal behavior for single accounts. Considering money laundering activities are often involved in gang criminals, these methods are still not intelligent enough to crack down on criminal gangs all-sidedly. In this paper, a systematic solution is presented to find suspicious money laundering gangs. A temporal-directed Louvain algorithm has been proposed to detect communities according to relevant anti-money laundering patterns. All processes are implemented and optimized on Spark platform. This solution can greatly improve the efficiency of anti-money laundering work for financial regulation agencies.

Nowadays, NoSQL databases become more and more prevalent. As an important category of NoSQL databases, key-value store (KVS) is widely used in many Big Data applications by providing flexible and efficient performance. In a KVS, a key-value pair can be easily accessed by giving a key. Recently, a new Ethernet accessed disk drive called "Kinetic Drive" was developed by Seagate. This drive can be directly accessed by applications via IP addresses without going through a dedicated storage server. Given the keys from a user, Kinetic Drive can perform key-value pair operations by itself. It can reduce the management complexity, especially in large-scale deployment. Given a set of Kinetic Drives in a data center, it is important to manage the key-value pairs and store them in an organized way. In this paper, we investigate how to allocate data on a large-scale key-value store system using Kinetic Drives. We show the tradeoff among various design factors. We investigate key indexing schemes and allocate data on drives accordingly. We propose efficient approaches to migrate data among drives. We carry out performance evaluation that shows our approaches can handle various key distributions well.

The research and application of effective scheduling methods play an important role in manufacturing enterprises to improve production efficiency and reduce production costs. Based on the characteristics of the capability maturity of processors, this paper establishes a process task scheduling strategy that is conducive to guarantee delivery time and reduce labor costs according to the actual production environment of complex equipment manufacturing enterprises. A scheduling two-level planning model with delay time and labor cost as indicators is constructed, and the model is solved by genetic algorithm. The planning scheduling method based on order priority and technician effective working time constraints is studied. Finally, an example is given to verify the feasibility and effectiveness of the proposed method.

As one of the challenging tasks of multiple autonomous underwater vehicles systems, the realization of target hunting is the great significance. The multiple autonomous underwater vehicle target hunting is studied in this article. In some research, because the hunting members cannot reach the hunting point at the same time, the hunting time is long or the target escapes. To improve the efficiency of the target hunting, the leader–follower formation algorithm is introduced. Firstly, the task is assigned based on the distance between the autonomous underwater vehicle and the target. Then, the autonomous underwater vehicles with the same task are formed based on leader–follower mode, and the formation is kept to track the target. In the final capture phase, multiple autonomous underwater vehicle system use angle matching algorithm to round up target. The simulation results show that the proposed algorithm can effectively accomplish the target hunting task, save the hunting time, and avoid the target escape. Compared with the bioinspired neural network algorithm, the proposed algorithm shows better performance.

For achieving fast accesses the cloud's big data through mobile smart node, an extreme high data rate for the forwarding link under high-mobility wireless communications is necessary. 3GPP thus specifies LTE/LTE-Advanced specifications as the 4G system to cooperate with the existing UMTS system. To balance loads among different communication interfaces in the hybrid cellular system is one of the most important issues that should be addressed for achieving efficient radio resource allocations. In a shared packet service, the 3GPP UMTS adopts the VSF-OFCDM interface to allocate orthogonal codes of an OVSF code tree in two-dimension spreading of the time and frequency domains. However, UMTS suffers from high packet loss rate and high bandwidth waste rate. In 4G, although the LTE/LTE-Advanced interface offers a high data rate, it suffers from unbalanced loads and moderate reward. This paper thus proposes an adaptive radio interface selection for balancing loads between the UMTS and LTE/LTE-Advanced interfaces according to various interference and mobility environments. Numerical results indicate that the proposed approach outperforms other approaches in fairness, FRL, utilization and call blocking.

In this paper, we performed an exploration on the design and evaluation of pen tail gesture, an interaction method that allows the use of pen tail movement to initiate interactions. Based on our interviews with some designers and researchers who regularly used pen-based tools, we conducted three experiments to establish baseline criteria to distinguish intentional pen tail gestures from incidental pen tail movements, and to understand the basic movement behaviors in pen tail gestures. We developed designs and recognition methods of pen gestures, and implemented three application prototypes based on them. Our research can inspire some new designs of pen-based tools and enrich the design repertoire of pen-based user interfaces.

The universal mobile telecommunications system UMTS and long term evolution LTE/LTE-advanced specifications have been proposed to offer high data rate for the forwarding link under high-mobility wireless communications. The keys include supporting multi-modes of various coding schemes e.g., VSF-OFCDM, OFDM, OFDMA, multiple-input multiple-output, relay networks, and so on. To balance loads among different communication interfaces is one of the most important issues that should be addressed for achieving efficient radio resource allocations. In a shared packet service, the 3GPP UMTS adopts the VSF-OFCDM interface to allocate orthogonal codes of an OVSF code tree in two-dimension 2D spreading at the time and frequency domains. However, it easily leads to high packet loss rate and high bandwidth waste rate, because it does not consider interference and the adaptive modulation and coding scheme simultaneously. Conversely, although the LTE/LTE-advanced interface offers a high data rate, it suffers from unbalanced loads and moderate reward. This paper thus proposes an adaptive radio resource allocation for balancing loads between the UMTS and LTE/LTE-advanced interfaces according to various interference and mobility environments. In addition, an adaptive multi-code allocation is proposed for the UMTS to minimize the bandwidth waste rate while guaranteeing quality of service. Numerical results indicate that the proposed approach outperforms other approaches in fractional reward loss and system utilization. Copyright © 2013 John Wiley & Sons, Ltd.

Aiming at the need for autonomous learning in reinforcement learning (RL), a quantitative emotion-based motivation model is proposed by introducing psychological emotional factors as the intrinsic motivation. The curiosity is used to promote or hold back agents' exploration of unknown states, the happiness index is used to determine the current state-action's happiness level, the control power is used to indicate agents' control ability over its surrounding environment, and together to adjust agents' learning preferences and behavioral patterns. To combine intrinsic emotional motivations with classic RL, two methods are proposed. The first method is to use the intrinsic emotional motivations to explore unknown environment and learn the environment transitioning model ahead of time, while the second method is to combine intrinsic emotional motivations with external rewards as the ultimate joint reward function, directly to drive agents' learning. As the result shows, in the simulation experiments in the rat foraging in maze scenario, both methods have achieved relatively good performance, compared with classic RL purely driven by external rewards.

Chinese Sign Language (CSL) subword recognition based on surface electromyography (sEMG), accelerometer (ACC), and gyroscope (GYRO) sensors was explored in this paper. In order to fuse effectively the information of these three kinds of sensors, the classification abilities of sEMG, ACC, GYRO, and their combinations in three common sign components (one or two handed, hand orientation, and hand amplitude) were evaluated first and then an optimized tree-structure classification framework was proposed for CSL subword recognition. Eight subjects participated in this study and recognition experiments under different testing conditions were implemented on a target set consisting of 150 CSL subwords. The proposed optimized tree-structure classification framework based on sEMG, ACC, and GYRO obtained the best performance among seven different testing conditions with single sensor, paired-sensor fusion, and three-sensor fusion, and the overall recognition accuracies of 94.31% and 87.02% were obtained for 150 CSL subwords in a user-specific test and user-independent test, respectively. Our study could lay a basis for the implementation of large-vocabulary sign language recognition system based on sEMG, ACC, and GYRO sensors.

Nowadays, MapReduce has become very popular in many applications, such as high performance computing. It typically consists of map, shuffle and reduce phases. As an important one among these three phases, data shuffling usually accounts for a large portion of the entire running time of MapReduce jobs. MapReduce was originally designed in scale-out architecture with inexpensive commodity machines. However, in recent years, scale-up computing architecture for MapReduce jobs has been developed. Some studies indicate that in certain cases, a powerful scale-up machine can outperform a scale-out cluster with multiple machines. With multi-processor, multi-core design connected via NUMAlink and large shared memories, NUMA architecture provides a powerful scale-up computing capability. Compared with Ethernet connection and TCP/IP network, NUMAlink has a much faster data transfer speed which can greatly expedite the data shuffling of MapReduce jobs. The impact of NUMAlink on data shuffling in NUMA scale-up architecture has not been fully investigated in previous work. In this paper, we ignore the computing power (i.e., map and reduce phases) of MapReduce, but focus on the optimization of data shuffling phase in MapReduce framework in NUMA machine. We concentrate on the various bandwidth capacities of NUMAlink(s) among different memory locations to fully utilize the network. We investigate the NUMAlink topology using SGI UV 2000 as an example and propose a topology-aware reducer placement algorithm to speed up the data shuffling phase. In addition, we extend our approach to a larger computing environment with multiple NUMA machines, and design a reducer placement scheme to expedite the inter-NUMA machine data shuffling. Experiments results show that data shuffling time can be greatly reduced in NUMA architecture with our solution.

Network Security Situation Awareness (NSSA) helps security analysts to be aware of the actual security situation of their networks. But its adaptive control model and method remain a new research field that needs to be explored. In this paper, we presented a novel NSSA adaptive control model based on cognitive network. The model adopted circle cognitive structure composed of Monitor, Awareness, Decision and Execution (MADE). In the MADE cognitive circle, we described a fusion awareness method based on PSO-DS theory. The decision and execution strategies were also discussed briefly. The experiments prove that our model realizes the adaptive regulation of NSSA effectively and it also can be applied into other fields in order to develop automatic tools and devices.