Search

After nearly a decade of practice and exploration, the construction of smart cities in China still exposes many practical difficulties that are hard to break through. Driven by the longterm prospect of smart society, the construction of smart urban agglomeration provides a new idea for breaking the bottleneck of construction and promoting the phased achievement of smart society. Based on the internal logic of holistic governance theory, this study makes a theoretical interpretation of the connotation and characteristics of smart urban agglomerations, and puts forward that smart urban agglomerations have four characteristics: combination of goals and concepts, integration of regional governance, complementary application of technologies and coordination of industrial functions. At the same time, combining with the policy planning of China's major urban agglomerations, this study analyzes the key policy points of China's smart urban agglomerations development from three aspects of technology foundation, resource integration and application system. Therefore, we conclude that as a transitional development model between smart city and smart society, smart urban agglomeration is not only a further extension of the new smart city in the spatial scope, but also a forward-looking thinking to integrate the fragmented and distributed smart city.

The Zynq®-7000 SoC family integrates the software programmability of an ARM®-based processor with the hardware programmability of an FPGA, enabling key analytics and hardware acceleration while integrating CPU, DSP, ASSP, and mixed signal functionality on a single device. Consisting of single-core Zynq-7000S and dual-core Zynq-7000 devices, the Zynq-7000 family is the best price to performance-per-watt, fully scalable SoC platform for your unique application requirements.

With the acceleration of information and the rapid development of economy, the management of various industries in the society will tend to be intelligent, information and standardization. There are various new technologies speed up the pace of security management. In this paper, we adopted sensing layer data acquisition technology of Internet of things based on RFID to design an intelligent security patrolling system of the police robots. Meanwhile, we put forward RFID core technology and optimized cryptography technology, which are used to solve the reliability and integrity verification requirements of data acquisition by nodes of the sensing layer of Internet of things.

Matrix factorization (MF) has been one of the powerful machine learning techniques for collaborative flittering, and it is also widely extended to improve the quality for various tasks. For recommendation tasks, it is noting that a single user or item is actually shown to be sparsely correlated with latent factors extracted by MF, which has not been developed in existing works. Thus, we are focusing on levering sparse representation, as a successful feature learning schema for high dimensional data, into latent factor model. We propose a Sparse LAtent Model (SLAM) based on the ideas of sparse representation and matrix factorization. In SLAM, the item and user representation vectors in the latent space are expected to be sparse, induced by the l 1 -regularization on those vectors. Besides, we extend a dual graph Lapalacian regularization term to simultaneously integrate both user network and item network knowledge. Also, an iterative optimization method is presented to solve the new learning problem. The experiments on real datasets show that SLAM can predict the user–item ratings better than the state-of-the-art matrix factorization based methods.

With the rapid development of social media, people get increasingly involved in public affairs through social networks, e.g., microblogs. Although social network plays a significant role in public affairs, some online government communities lack sufficient response, triggering an negative environment for policy decision process. In order to foster a healthy online government community, understanding the mechanism and drivers of user interaction becomes critical, especially for user to user interaction. We build a theoretical model to explain the formation of user to user network and understanding the underlying mechanism of why a user would be replied by others in government microblogging. We collect the user information, posted microblogs and the replies from government microblogs in 10 different areas to construct user replying networks. By using exponential random graph model (GRGM), we delve into the mechanism of how the network structures and node attributes affect the formation of user replying networks. Our experimental results validate that the mutual, homophily and emotional effect are the main drivers of users replying behavior.

Deep convolutional network has achieved great success in visual recognition of static images, while it is not so advantageous as traditional methods in action recognition in videos. However, in recent years, researchers have proposed a series of human action recognition networks for videos, and proposed some efficient and effective methods for action recognition. This paper will enumerate the current main action recognition network model. First we will introduce Two-Stream Convolutional Network, Temporal Segment Networks and C3D Network, and then we will introduce an efficient method for action recognition. The purpose is to introduce the characteristics of different network models, and provide some valuable reference for the research of action recognition.

With the improvement of the complexity of electronic system, clock frequency and density of Printed Circuit Board (PCB) increase higher and higher. In the design of high-speed PCB. Integrity of signal is a top priority for designers. Many factors such as electromagnetic compatibility (EMC), PCB electrical parameters, component layout, high-speed signal line cabling will affect the performance of the final system. How to reasonably avoid the impact of these factors and what issues should be noted become the key to our successful design of high-speed PCB and to ensure the integrity of electronic system signals.

Application orchestration technology has always been one of the research directions of Devops. The traditional application orchestration technology has some problems such as slow deployment and scaling speed. Therefore, this paper proposes a Docker-based hybrid cloud application orchestration scheme addresses these issues with traditional application orchestration techniques. The goal is to solve the problem of existing application orchestration techniques from the root by borrowing Docker’s quick-start feature and the universal characteristics of hybrid cloud technology. Firstly, this paper proposes a four-layer architecture including the base layer, core layer, gateway layer and interaction layer, and designs the application orchestration module, dynamic scaling module and system monitoring module in the core layer. Secondly, the implementation scheme is tested, and the experiment shows that the scheme shows good deployment speed, speed of expansion and high reliability.

Social recommendation techniques have been developed to employ user's social connections for both rating prediction and Top-N recommendation. However, they are mostly using social network enhanced matrix factorization (MF) where the objective is to minimize the prediction error of rating scores, which makes it impractical and unsuccessful for Top-N recommendation. This paper thus focuses on developing more effective methods to utilize social network information for Top-N recommendation. Social network regularized Sparse LInear Model (SocSLIM) with its extensions incorporating local learning (LocSocSLIM) to improve efficiency are proposed. SocSLIM learns sparse coefficient matrix for users by solving a sparse representation problem over user-item rating/purchase matrix and user-user social network's adjacency matrix at the same time by sharing coefficient matrix. The coefficient matrix is used to predict the recommendation scores, which are then combined with a proposed item based Distance regularized Sparse LInear Model (DSLIM) to generate recommendations for the users. The experimental results demonstrate that SocSLIM effectively uses the social information to outperform the state-of-the-art methods by at least 12%. Moreover, the local weight learning extension LocSocSLIM significantly improves the efficiency up to 10 times as compared to SocSLIM as the original SLIM while achieving the close performance guarantees.

Although social media plays a significant role in public affairs, some online government communities still lack sufficient response, triggering a negative environment for the policy decision process. In order to foster a healthy online public community, understanding the mechanism and drivers of user interaction becomes critical, especially for the user-to-user interaction. We build a theoretical model to explain the formation of a user-to-user interaction network and understand the underlying mechanism why a user would be replied to by others in government microblogs. We collect a dataset from government microblogs on Sina Weibo to construct attributed-user reply networks containing 2461 users and 3937 replies. By using the exponential random graph model (ERGM), we test how the network structures and attributes, especially about the interest similarity and emotion expression affecting the formation of user reply networks. Our experimental results provide important empirical evidence that user interaction in online government microblogs is highly reciprocal and transitive. Interest-based homophily is a significant predictor rather than gender homophily. Moreover, users with high social influence and with more extreme emotional signals are more likely to get replies from others. However, these users with high social influence and activeness, in turn, will not be likely to reply to others.

Today is the era of information explosion, efficient methods for massive data processing are becoming a hotspot. MapReduce is a popular parallel programming model and widely used in large-scale data parallel computing. This paper proposes a parallel programming model based on heterogeneous multi-core embedded processor. On the base of ordinary computer completing computing, we make optimization and improvement, and use the method of combining embedded dual-core ARM processors with multiple high performance FPGAs to complete complex processing. Xilinx Zynq device is used as the hardware platform. By selecting Sobel image processing, histogram, matrix multiply and illumination enhancement cases, we test the parallel programming model. The results show that the speedup is 84.62x versus CPU-based implementation. In addition, it is proven that this parallel programming model is suitable for the CPU-FPGA heterogeneous system which can be used as cluster cloud computing node. Meanwhile, we make a comparison of our model with other references and verify the efficient of the programming model in this paper.

The system’s self-protection mechanism immediately stops the motor when motor system in the event of a malfunction, so it is difficult to collect the fault data when monitoring the motor status. Under the premise of only collecting motor’s health data, using SVDD algorithm to train health data and building non-health data sets based on practical experience in this paper. Based on BP neural network, a random self-adapting particle swarm optimization algorithm (RSAPSO) is used to substitute the original gradient descent method in BP network, training speed and accuracy of BP network training is improved. Three commonly used test functions were used to test the performance of the improved PSO, and the improved particle swarm optimization is compared with the standard particle swarm optimization, particle swarm optimization with compression factor and adaptive particle swarm optimization. In this paper, three asynchronous motor Y225S-4 output shaft vibration acceleration signal in healthy state as a case to test the effectiveness of the algorithm, results show that in the case of only health data, the new algorithm based on single classification has better performance and can effectively monitor the working state of the motor.

This study proposes a novel method of optimal path planning in stochastic constraint network scenarios. We present a dynamic stochastic grid network model containing semienclosed narrow and long constraint information according to the unstructured environment of an underground or mine tunnel. This novel environment modeling (stochastic constraint grid network) computes the most likely global path in terms of a defined minimum traffic cost for a roadheader in such unstructured environments. Designing high-dimensional constraint vector and traffic cost in nodes and arcs based on two- and three-dimensional terrain elevation data in a grid network, this study considers the walking and space constraints of a roadheader to construct the network topology for the traffic cost value weights. The improved algorithm of variation self-adapting particle swarm optimization is proposed to optimize the regional path. The experimental results both in the simulation and in the actual test model settings illustrate the performance of the described approach, where a hybrid, centralized-distributed modeling method with path planning capabilities is used.

Convolution neural network has important applications in the field of image recognition and retrieval, face recognition and object detection in deep learning. In the training of convolution neural network, 2D convolution, spatial pooling, linear mapping and other operations of forward propagation will have a huge computational complexity. In this paper, an effective optimization technique is proposed to map the convolutional neural network to the digital processor DSP. These technologies include: fixed-point conversion, data reorganization, weight deployment and LUT (look-up table). These technologies enable us to optimize the use of resources on the C66x DSP. The experiment is carried out on Texas Instruments C6678 development board, and the optimization technique proposed in this paper can be applied to multiple open-source network topologies.

In convolution neural networks, digital multiplication operation is the arithmetic operation of the most space-consuming and power consumption. This paper trains convolutional neural network with three different data formats (float point, fixed point and dynamic fixed point) on two different datasets (MNIST, CIFAR-10). For each data set and each data format, the paper assesses the impact of the multiplication accuracy to the error rate at the end of the training. The results show that the network error rate which is trained with low accuracy fixed point has small difference with the network training error rate which is trained with floating point, and this phenomenon shows that the use of low precision can fully meet the training requirements in the process of training the network.

Detection of moving objects is a kind of segmentation techniques based on regional characteristic such as color, gray, texture, which is the key technology in analyzing and processing of video image. A real-time motion detection method based on improved Gaussian Mixture Model is presented in this paper which is optimized and structure adjusted from Gaussian Mixture Model. Gaussian Mixture Model has been widely used in complex background scene modeling, especially in some occasions with small repetitive motion, such as shaking of the leaves, a rotating fan, bushes, the sea waves, rain, snow, etc.

This paper presents a comparative analysis of 60 municipal smart city plans drawn from countries around the world, with the goal of enumerating the specific policies and programs that are included under the general rubric of “smart city” initiatives. The objective is to identify the combinations of projects that are most often deployed together, and thus to define “archetypes” or “models” in smart city development. We follow an inductive method and conduct content analysis of 60 smart city plans, with each plan coded for the presence or absence of activity in 25 program categories. Cluster analysis identified four different models: an essential services model, smart transportation model, broad spectrum model, and a business ecosystem model.

Yang, J.; Tang, Z.; Wang, Z.; Wang, X.; Huang, Q., and Wu, M., 2018. Improved fault petri nets of unmanned seafloor drilling based on traces transition. In: Liu, Z.L. and Mi, C. (eds.), Advances in Sustainable Port and Ocean Engineering. Journal of Coastal Research, Special Issue No. 83, pp. 479–485. Coconut Creek (Florida), ISSN 0749-0208. For this kind of complex system equipment, parts fault detection of single large difficulty and low efficiency from the whole system considering many factors can lead to failure, and these factors affect each other, and the fault also exists between some logical relations, because of the condition monitoring and fault diagnosis accurately and effectively is very difficult. Based on the analysis of the traditional Petri Net application in the field of fault diagnosis, this paper gives fault diagnosis modeling expression of different classification problem & a kind of theory, and introduces transition token matrix representation of traces transition in order to ...

The processing efficiency of Intelligent Transportation System has become an issue which is attracting more and more attention in order to combat vehicle crimes. The vehicle color recognition plays an important role in identifying, searching, improving and enhancing vehicle Intelligent Transportation System. However, area identification or location and surface high light detection are needed before vehicle color recognition. Convolution neural network training algorithm is taken as the best choice to conduct vehicle color recognition for its parameters are less and computation speed is faster. Neural network algorithm is of high parallelism, and its algorithm can even be further optimized if CUDA acceleration is selected. This research results have reference value for improving the processing efficiency of Intelligent Transportation System.

Vehicle recognition is the important content of intelligent transportation system, there have been many researches on vehicle recognition, and the technology of vehicle recognition based on CPU and DSP cannot meet the needs of the present. This article is about the study of Vehicle recognition and how to realize the GPU algorithm on the CUDA transplantation, make the algorithm parallel, thus speeding up the computation efficiency of vehicle recognition. This thesis is based on the Jeston TK1 development board as the experimental object, achieving high efficiency of GPU image processing.

High Performance computing (HPC) are becoming not only more complex but also challenging in terms of speedup and scalability. As the size of compute intensive problems increases, Intel MIC architecture comes true. In this paper, PCG method based on Intel MIC architecture is employed to solve large scale linear equations. The numerical results show that PCG method based on Intel MIC architecture has a considerable speedup and scalability.

Congestion estimation is an important issue for design automation of the VLSI layout. Fast congestion estimation provides an efficient means to adjust the placement and wire planning. A probabilistic model of interconnections enables designers to quickly predict routing congestion. We propose a powerful and fast estimation approach that allows wires to have bounded-length detours to bypass congestions. The method is more realistic and precise than the previous work. The experimental results demonstrate the effectiveness of the method on routing benchmarks.

The high-resolution micro-computed tomography (micro-CT) system has now become an important tool for biological research. The micro-CT system enables a non-invasive inspection to screen anatomical changes in small animals. The promising advantages include high-spatial resolution, high sensitivity to bone and lung, short scan time and cost-effectiveness. The dose received by the small animal might be a critical concern in the research. In this article, the choice of the components, fundamental physical problems, the image reconstruction algorithm and the representative applications of micro-CT are summarized. Some results from our research group are also presented to show high-resolution images obtained by the micro-CT system.

Congestion estimation plays an important role in the physical layout of VLSI design. This paper presents a new probabilistic estimation model that improves the previous estimators by relaxing the constraint on detours in a route. The model is more general and realistic for it gives the flexibility for the wires to have wider usage area to bypass the congestion regions and blockages. Given a routing grid and a set of nets to be routed, the model predicts the routing density on each edge of the grid. The routing density provides direct congestion estimation. We compare our estimation results to the actual routing results. Experimental results show the effectiveness of our estimator.

Depth estimation becomes the key technology to resolve the communications of the stereo vision. We can get the real-time depth map based on hardware, which cannot implement complicated algorithm as software, because there are some restrictions in the hardware structure. Eventually, some wrong stereo matching will inevitably exist in the process of depth estimation by hardware, such as FPGA. In order to solve the problem a postprocessing function is designed in this paper. After matching cost unique test, the both left-right and right-left consistency check solutions are implemented, respectively; then, the cavities in depth maps can be filled by right depth values on the basis of right-left consistency check solution. The results in the experiments have shown that the depth map extraction and postprocessing function can be implemented in real time in the same system; what is more, the quality of the depth maps is satisfactory.

New types of search engines are emerging to overcome defects of traditional search engines. What the next search engine should be? It is an efficient way to classify and compare current search engines to find possible answers. This paper first defines search pattern as the combination of index structure, user profile, and interaction mechanism, which can model a search engine and help us classify search engines more essentially. Then, current search engines are classified and compared from the view of search pattern to find what the next search engine should be. Finally, a new search pattern for next generation search engine is proposed by synthesizing the advantages of different search patterns, which can help developing new search engines and may be the possible direction of the next search engine.

In this paper, median filter algorithm and motion vector compensation algorithm is researched. We design and implement the algorithms on the FPGA board in detail. According to the tests and experiments, we verify the practicability of the algorithms on FPGA and video surveillance system. The algorithms will effectively increase the efficiency of late video signal processing and help the application of video enhance.

Sobel is one of the best edge detection algorithms. However, it often loses some textural information when complex image is processed. This paper mainly discusses how to improve the shortcomings of Sobel and optimize the filtering, templates, and noise algorithms for edge detection. Beside the optimization, the paper presents the implementation of edge detection on Davinci platform. The algorithm code has been optimized which based on the TMS320DM6446 dual-core DSP. Multiplication operations have been split into the shift and add operations through integration algorithm engine which use the DSP’s pipelining operations. After the test experiment on Davinci platform, it can work efficiently. This method can not only provide the possibility of the real-time image processing, but also eliminate the impact of noise, and enhance the detection accuracy.

This paper introduces the Canny edge detection algorithm and the implementation of the Canny edge-detection algorithm on an DSP. Canny edge detection significantly preserves the important structural properties in an image, while reducing the amount of data and filters out useless information. The Canny Edge Detection Algorithm is used to illustrate the creation of a "real life" analytics algorithm. It conveniently illustrates the different memory management models possible, the architecture choices to make and also creates an easily visualized demo application for display. This section describes the codec architecture, build process and integration into an application on the DM6446.

This article introduces the implementation of an adaptive edge-detection filter on a DSP (dm642) using a combination of hardware and software components, which can detect textural features of digital image, makes it non-linear enhancement and suppresses the noise to a certain extent. As a benchmark, preliminary results are presented for this system and evaluated with respect to the different edge detector. And how to use the advantages of the DM642 processor to improve the efficiency of edge-detection is mainly discussed in this article. In order to improve the performance of edge-detection algorithm on DM642, a series of rapid image processing algorithms are optimized and proposed. The emphasis of the experiments is put on the feasibility of distributed high-performance processing from both hardware and software aspects, which may be easily applied to other larger scale or more hard real-time intelligent information processing. Experiments show that it can detect edges with high accuracy, more details and fine distinction.

This article introduces one adaptive binary arithmetic coding system based on context. H.264/AVC is a new video coding standard developed by ITU-T VCEG and ISO/IEC MPEG. It includes many new technology features, such as Multi-mode prediction, Flexible Macro block Ordering (FMO), integer transform, Universal Variable Length Codes (UVLC), Context-based Adaptive Binary Arithmetic Coding (CABAC), laying code, error restriction machine, etc. This paper introduced the Context-based Adaptive Binary Arithmetic Coding (CABAC) method presented in H.264/AVC. And it makes a comparison between CABAC and UVLC. Context-based Adaptive Binary Arithmetic Coding (CABAC) is a new entropy coding method presented in H.264/AVC that is highly efficient in video coding. In the method, the probability of current symbol will be estimated by using a well designed context model, which is adaptive and can approach to the statistic characteristic. Then an arithmetic coding mechanism will largely reduce the redundancy including inter-symbol redundancy. Compared with UVLC method in the prior standard, CABAC is complicated but efficiently reduce the bit rate.

This article deals with several kernel technologies applied in vehicle information service.The technology referring to GPS (Global Positioning System),GPRS (General Packet Radio Service),GIS (Geographical Information System),and Computer Network etc,as we have studied,is believed to be able to construct a Vehicle Information Service System which faces the public.Synthesizing the technologies including GPS,GPRS,GIS,image capture and compress based on ARM9 processor and embedded system to establish a intelligent vehicle service model,which emphasizes U-V-S (User-Vehicle-Server) interaction.It is a integrated application in embedded system and its application software runs on relatively real-time and multi-mission operating systems.

This paper study softcopy display algorithms of digital radiographic images acquired using a prototype flat panel detector. The processing is manipulated in two steps. Firstly, a look up table is applied to map the gray value of original image into proper visual scope. Then, the mapped image is decomposed into a serial subband images with a multiscale pyramid structure. The contrast of subband images are enhanced by using nonlinear functions. Both performance and computational simplicity make our algorithms attractive

Interconnect congestion estimation plays an important role in the physical design of integrated circuits. This paper presents a novel probabilistic approach to predicting wiring space in two-dimensional arrays. We propose a hierarchical estimation method to derive approximated upper bounds for wiring space. We use the net density distribution for predicting the routing congestion. Experimental results demonstrate the promising performance of the new approach.