Network Function Virtualization (NFV) enables the embedding of Virtualized Network Function (VNF) into commodity servers. A sequence of VNFs can be chained in a particular order to form a service chain (SC). This paper considers placing multiple SCs in a geo-distributed edge system owned by multiple service providers (SPs). For a pair of SC and SP, minimizing the placement cost while meeting a latency constraint is formulated as an integer programming problem. As SC clients and SPs are self-interested, we study the matching between SCs and SPs that respects individual's interests yet maximizes social welfare. The proposed matching approach excludes any blocking individual and block pair which may jeopardize the stability of the result. Simulation results show that the proposed approach performs well in terms of social welfare but is suboptimal concerning the number of placed SCs.

Frequent Itemsets(FI) mining is a popular and important first step in analyzing datasets across a broad range of applications. There are two main problems with the traditional approach for finding frequent itemsets. Firstly, it may often derive an undesirably huge set of frequent itemsets and association rules. Secondly, it is vulnerable to noise. There are two approaches which have been proposed to address these problems individually. The first problem is addressed by the approach Frequent Closed Itemsets(FCI), FCI removes all the redundant information from the result and makes sure there is no information loss. The second problem is addressed by the approach Approximate Frequent Itemsets(AFI), AFI could identify and fix the noises in the datasets. Each of these two concepts has its own limitations, however, the authors find that if FCI and AFI are put together, they could help each other to overcome the limitations and amplify the advantages. The new integrated approach is termed Noise-tolerant Frequent Closed Itemset(NFCI). The results of the experiments demonstrate the advantages of the new approach: (1) It is noise tolerant. (2) The number of itemsets generated would be dramatically reduced with almost no information loss except for the noise and the infrequent patterns. (3) Hence, it is both time and space efficient. (4) No redundant information is in the result.

With the rapid development of multimedia, violent video can be easily accessed in games, movies, websites, and so on. Identifying violent videos and rating violence extent is of great importance to media filtering and children protection. Many previous studies only address the problems of violence scene detection and violent action recognition, yet violence rating problem is still not solved. In this paper, we present a novel video-level rating prediction method to estimate violence extent automatically. It has two main characteristics: (1) a two-stream network is fine-tuned to construct effective representations of violent videos; (2) a violence rating prediction machine is designed to learn the strength relationship among different videos. Furthermore, we present a novel violent video dataset with a total of 1,930 human-involved violent videos designed for violence rating analysis. Each video is annotated with 6 fine-grained objective attributes, which are considered to be closely related to violence extent. The ground-truth of violence rating is given by pairwise comparison method. The dataset is evaluated in both stability and convergence. Experiment results on this dataset demonstrate the effectiveness of our method compared with the state-of-art classification methods.

Recently, the face hallucination method based on deep learning understands the mapping between low-resolution (LR) and high-resolution (HR) facial patterns by exploring the priors of facial structure. However, how to maintain the face structure consistency after the reconstruction of face images at different scales is still a challenging problem. In this letter, we propose a novel multi-scale structure prior learning (MSPL) for face hallucination. First, we propose a multi-scale structure prior block (MSPB). Considering the loss of high-frequency information in the LR space, we mainly process the input image in three different scale ascending dimensional spaces, and map the image to the high dimensional space to extract multi-scale structural prior information. Then the size of feature maps is recovered by downsampling, and finally the multi-scale information is fused to restore the feature channels. On this basis, we propose a local detail attention module (LDAM) to focus on the local texture information of faces. We conduct extensive face hallucination reconstruction experiments on a public face dataset (LFW) to verify the effectiveness of our method.

Due to the increased power density and lower thermal conductivity, 3D ICs are faced with heat dissipation and temperature problem seriously. TSV (Through-Silicon-Via) has been shown as an effective way to help heat removal, but they introduce several issues related with cost and reliability as well. Previous researches of TSV planning didn't pay much attention to the impact of leakage power, which will bring in error on estimation of temperature, TSV number and also critical path delay. The leakage-temperature-delay dependence can potentially negate the performance improvement of 3D designs. In this paper, we analyze the impact of leakage power on TSV planning and integrate leakage-temperature-delay dependence into thermal via planning of 3D ICs. A weighted via insertion approach, considering the influence on both module delay and wire delay, is proposed to achieve the best balance among temperature, via number and performance. Experiment results show that, with leakage power and resource constraint considered, temperature and the required via number can be quite different, and the weighted TSV insertion approach with iterative process can obtain the trade-off between different factors including thermal, power consumption, via number and performance.

Recently, local features computed using convolutional neural networks (CNNs) show good performance to image retrieval. The local convolutional features obtained by the CNNs (LC features) are designed to be translation invariant, however, they are inherently sensitive to rotation perturbations. This leads to miss-judgements in retrieval tasks. In this work, our objective is to enhance the robustness of LC features against image rotation. To do this, we conduct a thorough experimental evaluation of three candidate anti-rotation strategies (in-model data augmentation, in-model feature augmentation, and post-model feature augmentation), over two kinds of rotation attack (dataset attack and query attack). In the training procedure, we implement a data augmentation protocol and network augmentation method. In the test procedure, we develop a local transformed convolutional (LTC) feature extraction method, and evaluate it over different network configurations. We end up a series of good practices with steady quantitative supports, which lead to the best strategy for computing LC features with high rotation invariance in image retrieval.

In this paper, a multicast concept for Device-to-Device (D2D) communication underlaying a cellular infrastructure is investigated. To increase the overall capacity and improve resource utilization, a novel interference coordination scheme is proposed. The proposed scheme includes three steps. First, in order to mitigate the interference from D2D multicast transmission to cellular networks (CNs), a dynamic power control scheme is proposed that can determine the upper bound of D2D transmitter power based on the location of Base Station (BS) and areas of adjacent cells from the coverage area of D2D multicast group. Next, an interference limited area control scheme that reduces the interference from CNs to each D2D multicast receiver is proposed. The proposed scheme does not allow cellular equipment (CUE) located in the interference limited area to reuse the same resources as the D2D multicast group. Then two resource block (RB) allocation rules are proposed to select the appropriate RBs from a candidate RB set for D2D multicast group. From the simulation results, it is confirmed that the proposed schemes improve the performance of the hybrid system compared to the conventional ways.

In recent years, single face image super-resolution (SR) using deep neural networks have been well developed. However, most of the face images captured by the camera in a real scene are from different views of the same person, and the existing traditional multi-frame image SR requires alignment between images. Due to multi-view face images contain texture information from different views, which can be used as effective prior information, how to use this prior information from multi-views to reconstruct frontal face images is challenging. In order to effectively solve the above problems, we propose a novel face SR network based on multi-view face images, which focus on obtaining more texture information from multi-view face images to help the reconstruction of frontal face images. And in this network, we also propose a texture attention mechanism to transfer high-precision texture compensation information to the frontal face image to obtain better visual effects. We conduct subjective and objective evaluations, and the experimental results show the great potential of using multi-view face images SR. The comparison with other state-of-the-art deep learning SR methods proves that the proposed method has excellent performance.

To solve the problem of the self-deployment of heterogeneous directional wireless sensor networks in 3D space, this paper proposes a weighted Voronoi diagram-based self-deployment algorithm (3DV-HDDA) in 3D space. To improve the network coverage ratio of the monitoring area, the 3DV-HDDA algorithm uses the weighted Voronoi diagram to move the sensor nodes and introduces virtual boundary torque to rotate the sensor nodes, so that the sensor nodes can reach the optimal position. This work also includes an improvement algorithm (3DV-HDDA-I) based on the positions of the centralized sensor nodes. The difference between the 3DV-HDDA and the 3DV-HDDA-I algorithms is that in the latter the movement of the node is determined by both the weighted Voronoi graph and virtual force. Simulations show that compared to the virtual force algorithm and the unweighted Voronoi graph-based algorithm, the 3DV-HDDA and 3DV-HDDA-I algorithms effectively improve the network coverage ratio of the monitoring area. Compared to the virtual force algorithm, the 3DV-HDDA algorithm increases the coverage from 75.93% to 91.46% while the 3DV-HDDA-I algorithm increases coverage from 76.27% to 91.31%. When compared to the unweighted Voronoi graph-based algorithm, the 3DV-HDDA algorithm improves the coverage from 80.19% to 91.46% while the 3DV-HDDA-I algorithm improves the coverage from 72.25% to 91.31%. Further, the energy consumption of the proposed algorithms after 60 iterations is smaller than the energy consumption using a virtual force algorithm. Experimental results demonstrate the accuracy and effectiveness of the 3DV-HDDA and the 3DV-HDDA-I algorithms.

In this letter, we study the resource allocation for the downlink cooperative non-orthogonal multiple access (NOMA) systems based on the amplifying-and-forward protocol relay transmission. A joint power allocation and amplification gain selection scheme are proposed. Fractional programming and the iterative algorithm based on the Lagrangian multiplier are used to allocate the transmit power to maximize the energy efficiency (EE) of the systems. Simulation results show that the proposed scheme can achieve higher energy efficiency compared with the minimum power transmission (MPT-NOMA) scheme and the conventional OMA scheme.

For high definition (HD) videos, the 3D-High Efficiency Video Coding (3D-HEVC) reference algorithm incurs dramatically highly computation loads. Therefore, with the demands for the real-time processing of HD video, a hardware implementation is necessary. In this paper, a reconfigurable architecture is proposed that can support both median filtering preprocessing and mean filtering preprocessing to satisfy different scene depth maps. The architecture sends different instructions to the corresponding processing elements according to different scenarios. Mean filter is used to process near-range images, and median filter is used to process long-range images. The simulation results show that the designed architecture achieves an averaged PSNR of 34.55dB for the tested images. The hardware design for the proposed virtual view synthesis system operates at a maximum clock frequency of 160MHz on the BEE4 platform which is equipped with four Virtex-6 FF1759 LX550T Field-Programmable Gate Array (FPGA) for outputting 720p (1024×768) video at 124fps.

Pedestrian detection is a significant task in computer vision. In recent years, it is widely used in applications such as intelligent surveillance systems and automated driving systems. Although it has been exhaustively studied in the last decade, the occlusion handling issue still remains unsolved. One convincing idea is to first detect human body parts, and then utilize the parts information to estimate the pedestrians' existence. Many parts-based pedestrian detection approaches have been proposed based on this idea. However, in most of these approaches, the low-quality parts mining and the clumsy part detector combination is a bottleneck that limits the detection performance. To eliminate the bottleneck, we propose Discriminative Part CNN (DP-CNN). Our approach has two main contributions: (1) We propose a high-quality body parts mining method based on both convolutional layer features and body part subclasses. The mined part clusters are not only discriminative but also representative, and can help to construct powerful pedestrian detectors. (2) We propose a novel method to combine multiple part detectors. We convert the part detectors to a middle layer of a CNN and optimize the whole detection pipeline by fine-tuning that CNN. In experiments, it shows astonishing effectiveness of optimization and robustness of occlusion handling.

In 3D IC design, thermal issue is a critical challenge. To eliminate hotspots, physical layouts are always adjusted by some incremental changes, such as shifting or duplicating hot blocks. In this paper, we distinguish the thermal-aware incremental changes in three different categories: migrating computation, growing unit and moving hotspot blocks. However, these modifications may degrade the packing area as well as interconnect distribution greatly. In this paper, mixed integer linear programming (MILP) models are devised according to these different incremental changes so that multiple objectives can be optimized simultaneously. Furthermore, to avoid random incremental modification, which may be inefficient and need long runtime to converge, here potential gain is modeled for each candidate incremental change. Based on the potential gain, a novel thermal optimization flow to intelligently choose the best incremental operation is presented. Experimental results show that migrating computation, growing unit and moving hotspot can reduce max on-chip temperature by 7%, 13% and 15% respectively on MCNC/GSRC benchmarks. Still, experimental results also show that the thermal optimization flow can reduce max on-chip temperature by 14% to the initial packings generated by an existing 3D floorplanning tool CBA, and achieve better area and total wirelength improvement than individual operations do. The results with the initial packings from CBA_T (Thermal-aware CBA floorplanner) show that 13.5% temperature reduction can be obtained by our incremental optimization flow.

In this paper, we propose a patch-wise learning based approach to deal with the multiple-shot people re-identification task. In the proposed approach, re-identification is formulated as a patch-wise set-to-set matching problem, with each patch set being matched using a specifically learned Mahalanobis distance metric. The proposed approach has two advantages: (1) a patch-wise representation that moderates the ambiguousness of a non-rigid matching problem (of human body) to an approximate rigid one (of body parts); (2) a patch-wise learning algorithm that enables more constraints to be included in the learning process and results in distance metrics of high quality. We evaluate the proposed approach on popular benchmark datasets and confirm its competitive performance compared to the state-of-the-art methods.

In this letter, we consider the power allocation scheme with rate proportional fairness to maximize energy efficiency in the downlink the non-orthogonal multiple access (NOMA) systems. The optimization problem of energy efficiency is a non-convex optimization problem, and the fractional programming is used to transform the original problem into a series of optimization sub-problems. A two-layer iterative algorithm is proposed to solve these sub-problems, in which power allocation with the fixed energy efficiency is achieved in the inner layer, and the optimal energy efficiency of the system is obtained by the bisection method in the outer layer. Simulation results show the effectiveness of the proposed algorithm.

This letter studies the price-based power control algorithm for the spectrum sharing cognitive radio networks. The primary user (PU) profits from the secondary users (SUs) by pricing the interference power made by them. The SUs cooperate with each other to maximize their sum revenue with the signal-to-interference plus noise ratio (SINR) balancing condition. The interaction between the PU and the SUs is modeled as a Stackelberg game. Closed-form expressions of the optimal price for the PU and power allocation for the SUs are given. Simulation results show the proposed algorithm improves the revenue of both the PU and fairness of the SUs compared with the uniform pricing algorithm.

As the nodes of AWSN (Aerial Wireless Sensor Networks) fly around, the network topology changes frequently with high energy consumption and high cluster head mortality, and some sensor nodes may fly away from the original cluster and interrupt network communication. To ensure the normal communication of the network, this paper proposes an improved LEACH-M protocol for aerial wireless sensor networks. The protocol is improved based on the traditional LEACH-M protocol and MCR protocol. A Cluster head selection method based on maximum energy and an efficient solution for outlier nodes is proposed to ensure that cluster heads can be replaced prior to their death and ensure outlier nodes re-home quickly and efficiently. The experiments show that, compared with the LEACH-M protocol and MCR protocol, the improved LEACH-M protocol performance is significantly optimized, increasing network data transmission efficiency, improving energy utilization, and extending network lifetime.

In this paper, the effective Long Range (LoRa) based wireless sensor network is designed and implemented to provide the remote data sensing functions for the planned smart agricultural recycling rapid processing factory. The proposed wireless sensor network transmits the sensing data from various sensors, which measure the values of moisture, viscosity, pH, and electrical conductivity of agricultural organic wastes for the production and circulation of organic fertilizers. In the proposed wireless sensor network design, the LoRa transceiver module is used to provide data transmission functions at the sensor node, and the embedded platform by Raspberry Pi module is applied to support the gateway function. To design the cloud data server, the MySQL methodology is applied for the database management system with Apache software. The proposed wireless sensor network for data communication between the sensor node and the gateway supports a simple one-way data transmission scheme and three half-duplex two-way data communication schemes. By experiments, for the one-way data transmission scheme under the condition of sending one packet data every five seconds, the packet data loss rate approaches 0% when 1000 packet data is transmitted. For the proposed two-way data communication schemes, under the condition of sending one packet data every thirty seconds, the average packet data loss rates without and with the data-received confirmation at the gateway side can be 3.7% and 0%, respectively.

The aim of this paper is to propose effective attentional regions for fine-grained visual recognition. Based on the Spatial Transformers' capability of spatial manipulation within networks, we propose an extension model, the Attention-Guided Spatial Transformer Networks (AG-STNs). This model can guide the Spatial Transformers with hard-coded attentional regions at first. Then such guidance can be turned off, and the network model will adjust the region learning in terms of the location and scale. Such adjustment is conditioned to the classification loss so that it is actually optimized for better recognition results. With this model, we are able to successfully capture detailed attentional information. Also, the AG-STNs are able to capture attentional information in multiple levels, and different levels of attentional information are complementary to each other in our experiments. A fusion of them brings better results.

A novel superpixel segmentation approach driven by uniform mixture model with spatially constrained (UMMS) is proposed. Under this algorithm, each observation, i.e. pixel is first represented as a five-dimensional vector which consists of colour in CLELAB space and position information. And then, we define a new uniform distribution through adding pixel position, so that this distribution can describe each pixel in input image. Applied weighted 1-Norm to difference between pixels and mean to control the compactness of superpixel. In addition, an effective parameter estimation scheme is introduced to reduce computational complexity. Specifically, the invariant prior probability and parameter range restrict the locality of superpixels, and the robust mean optimization technique ensures the accuracy of superpixel boundaries. Finally, each defined uniform distribution is associated with a superpixel and the proposed UMMS successfully implements superpixel segmentation. The experiments on BSDS500 dataset verify that UMMS outperforms most of the state-of-the-art approaches in terms of segmentation accuracy, regularity, and rapidity.