Search

Attribute learning has improved the performance in scene understanding and scene recognition. However, there are many attributes described by words or short texts in a static scene and traffic crowd scene. If there are two similar scenes, the semantic relationship topology structures of corresponding attribute groups of the two scenes are also homogeneity. But it is difficult to learn a semantic relation topology projection across semantic text data and visual data. To solve the problem, we construct approximate homeomorphism mapping based on the scene attributes semantic relational regularization. Hence, we propose a novel attribute semantic topological relationship regularization based scene attribute semantic learning(ARSL) method for scene semantic understanding. We establish a transport and travel scene recognition model based on attribute semantic features which are achieved by the proposed ARSL algorithm. In order to verify the proposed method, the experiments are implemented on the static transport-travel scene dataset and dynamic transport-travel crowds scene dataset respectively. The static transport-travel scene dataset is constructed by the SUN Attribute dataset including images and texts. However, the dynamic transport-travel crowds scene dataset is constructed through the WWW Crowd dataset including videos and texts, and the dynamic transport-travel crowd scene dataset is named as the WWW Crowd-Sub dataset. The performances of the proposed method are improved by 38.48% and 17.51% on the SUN Attribute dataset and WWW Crowd-Sub dataset respectively. The experimental results on the SUN Attribute dataset and WWW Crowd-Sub dataset demonstrate that the proposed approach has superior performance compared to state of the art. It can be demonstrated that the performance of the proposed ARSL method is effective against static transport-travel scene and dynamic transport-travel crowd scene.

In order to improve the transmission rate performance of precoding channel, this paper combines wireless power supply communication network and MU-MIMO to build WPCN-MU-MIMO system. Based on the energy transmission slot allocation characteristics of wireless power supply communication network technology, the precoding is optimized by using slot allocation. Due to the limitation of antenna array dimension and traditional power supply equipment, in order to solve the problem of precoding performance limitation caused by power supply, this paper considers that the time slot and power are linear in wireless power supply communication network technology, and the power and signal-to-noise ratio and channel transmission rate are also linear in MU-MIMO system, so WPCN-MU is combined with wireless power supply communication network technology and MU-MIMO system- The channel transmission rate performance and time slot conversion of MIMO system are convex functions. Considering that the channel capacity and time slot of WPCNMU-MIMO system are concave functions, it is proved that the optimization of WPCN-MU-MIMO precoding belongs to the optimal result in a certain time slot interval. Simulation results verify the correctness of channel transmission rate optimization through slot allocation in WPCN-MU-MIMO system.

With the rapid advancement of Internet of Things (IoT) and social networking application, the clouds is moving towards the network edges. It is foreseeable that more and more data will be processed in edge, and research organizations estimate that over 90% of the data will be stored and processed locally. This paper focus on the resource allocation for the multi-job multi-edge environments. We formulate the allocation problem as the concurrent job scheduling problem (CJSP), which is shown to be NP-complete. We propose the Weight Balance (WB) Algorithm to solve a special case of CJSP and we show that WB is optimal under some conditions. We then expand WB to solve the general CJSP. Extensive simulations demonstrate that the performance of our algorithm at small user and edge scale is almost as good as the optimal algorithm.

StakPak insulated gate bipolar transistor (IGBT) modules, which have complex internal structures and double-sided heat dissipation characteristics, represent a complex physical environment with multi-physical field coupling under normal working conditions. This makes it difficult to provide an accurate evaluation of the important reliability parameters of the devices contained in these modules, such as the device thermal resistance and junction temperature. In this paper, ANSYS software is used to model and perform a finite element analysis of single-chip StakPak IGBT modules. The heat flux ratio between the collector and emitter of the IGBT chip is calculated via a simulation; the overall power consumption of the module is then divided into two parts, and the thermal resistance values of the two electrodes, the overall temperature and the stress distribution of the IGBT chip and the StakPak device are simulated and calculated. The relationship between the heat dissipation ratio of the two electrodes and the thermal resistance is studied with respect to pressure, temperature and other factors.

Deep Neural Network (DNN) has been applied widely nowadays, making remarkable achievements in a wide variety of research fields. With the improvement of the accuracy requirements for the inference results, the topology of DNN tends to be more and more complex, evolving from chain topology to directed acyclic graph (DAG) topology, which leads to the huge amount of computation. For those end devices which have limited computing resources, the delay of running DNN models independently may be intolerable. As a solution, edge computing can make use of all available devices in the edge computing environments comprehensively to run DNN inference tasks, so as to achieve the purpose of acceleration. In this case, how to split DNN inference task into several small tasks and assign them to different edge devices is the central issue. This paper proposes a load-balancing algorithm to split DNN with DAG topology adaptively according to the environment. Extensive experimental results show the the propose adaptive algorithm can effectively accelerate the inference speed.

Edge computing is one of the emerging technologies aiming to enable timely computation at the network edge. With virtualization technologies, the role of the traditional edge providers is separated into two: edge infrastructure providers (EIPs), who manage the physical edge infrastructure, and edge service providers (ESPs), who purchase slices of physical resources (e.g., CPU, bandwidth, memory space, disk storage) from EIPs and then cache service entities to offer their own value-added services to end users. These value-added services are also called virtual network function or VNF. As we know, edge computing environments are dynamic, and the requirements of edge service for computing resources usually fluctuate over time. Thus, when the demand of a VNF cannot be satisfied, we need to design the strategies for migrating the VNF so as to meet its demand and retain the network performance. In this paper, we concentrate on migrating VNFs efficiently (MV), such that the migration can meet the bandwidth requirement for data transmission. We prove that MV is NP-complete. We present several exact and heuristic solutions to tackle it. Extensive simulations demonstrate that the proposed heuristics are efficient and effective.

Satellite component recognition has always been a hot topic in the field of orbital services. However, it is very challenging to segment the components such as satellite body, solar panel, and antenna in pixel-level accurately due to the poor illumination condition and the scarce image for spaceborne observation. Based on the Mask R-CNN, this paper proposes a lightweight instance segmentation model for satellite component segmentation and recognition. It improves residual module by using deep separable convolution, replacing nonlinear activation function with linear one after deep separable convolution and deleting the dimensionality reduction convolution layer in residual module. Also, the training datasets consist of the synthetic images generated by the 3D max software and the C-DCGAN based image generation method through several known satellite CAD models. The simulation experiments are carried out and the results show that the proposed method can effectively recognize the typical satellite components and achieve better performance than the compared model in aspects of accuracy, model parameters, and model size.

Characterisation of virtual objects and their properties is made more efficient through a grasping interface device with force feedback. The requirement for such dexterity has produced complex haptic interface structures that impose limitations on device transparency and fidelity. This study examines the effectiveness of a cable-driven pinch-grasp haptic interface to accurately generate forces to a user’s fingertips during object interaction in a virtual environment. Usability experiments were completed and comprised of size and stiffness discrimination tasks. Both studies compared grasping virtual objects under two modality configurations with the pinch-grasp interface. One configuration encompassed the combination of force and vision feedback, and the other maintained solely vision interaction with the device. Task completion times and correct identification were determined. Results demonstrate that the device improved task completion times in the two experiments when both force and vision feedback modalities were active. The percentage of correct identification was also greater in both experiments with force and vision feedback. These results motivate the use of a bidirectional cable actuation system for the development of multipoint dexterous haptic interface devices.

The huge amount of data contained in the point cloud restrains its applications in practice. To compress the point cloud, the spatial correlation in the point cloud is explored, where adjacent points are searched and used for prediction. In this paper, an adjacent points searching method based on the Mor-ton code to find proper adjacent points is proposed, based on which a weighted prediction is performed. The proposed method locates the adjacent points quickly and accurately using the Morton code of the point cloud coordinate. Experimental results show that the proposed method significantly reduces the computational complexity while maintaining similar performance as the state-of-the-art.

Considering that the power grid construction project has many sites, wide coverage and high level of safety risk, the multi-rotor unmanned aerial vehicle (UAV) patrol system is proposed as a new technique for safety and quality management in construction. Because the multi-rotor UAV patrol system is small, lightweight, highly efficient, intelligent and optional perspectives, safety risks and quality problems can be exposed more quickly. UAV cameras, like the eyes of regulators, play an important role in safety supervision and quality management. It can solve the problems caused by low camera pixel, long shooting distance, single shooting Angle and backlight, and can also find hidden defects and quality hidden troubles that are difficult to be found in naked eye inspection and acceptance at ordinary times. It can effectively avoid blind spots in inspection and ensure project quality.

This paper proposes a novel method for online junction temperature measurement of silicon carbide metaloxide-semiconductor field-effect transistors (SiC MOSFETs). The measurement method is using the miller plateau voltage (V GS ) and source leakage current (I DS ) at transient inrush currents during turn-on process that based on the temperaturesensitive electrical parameters measurement method (TSEP). The comparison of infrared temperature measurement results shows that this method can provide more accurate junction temperature results and also can be used for transient junction temperature measurement in SiC MOSFET devices. It solves the problem that the traditional temperature-sensitive electrical parameter measurement method is difficult to measure the transient temperature rise. Finally, the results show that the use of V GS and I DS can provide more accurate junction temperature results which can be used for transient junction temperatures in SiC MOSEFT devices.

Softwareized network, represented by software-defined network technology and network function virtualization technology, is considered to be an important emerging network paradigm to solve the challenges of current networks. To further guarantee the service quality of SFC and ensure the end-to-end delay within the specified threshold, it is necessary to conduct dynamic optimization and adjustment of SFC to realize elastic expansion and contraction and ensure the reliability and stability of network services. At present, the research on dynamic optimization of SFC is in the initial stage, and it focuses on the expansion and contraction of a single VNF. This paper proposes a scheduling optimization method of SFC based on the time delay threshold. The method determines the reasonable time of dynamic optimization by monitoring the end-to-end delay of the SFC. Then, the SFC-specific network function nodes are expanded and contracted to ensure the end-to-end delay within the specified range and meet the user's requirements on service quality.

In view of the increasing development of automated vehicles (AVs) technologies, it will be likely that road traffic is made up of a mixture of human-driven vehicles (HVs) and AVs in the coming years. To support traffic operation and management, this study proposed a hybrid approach integrating Wiedemann car-following model and cellular automation (CA) to reconstruct the trajectories of fully-sampled traffic flow on freeways. First, Wiedemann car-following model is applied to classify the vehicle driving states into following and closing. Then, human-driven vehicles (HVs) are inserted between the leading and following AVs based on the vehicle’s behavior within the following AV’s detection range. Next, the trajectories of inserted HVs are reconstructed by resorting to CA with four update rules set to determine vehicles’ acceleration, deceleration, randomization and position. Last, the proposed hybrid approach is performed under different traffic densities and AVs penetration rates. Results show that the proposed method for trajectory reconstruction performs satisfactorily on freeways even at low penetration rates of AVs.

In this paper we propose accelerating live rendering of flame using generative adversarial neural networks. The proposed method targets entertainment and simulation-based training industries whose demands for high fidelity and high frame rate increases steadily. The proposed approach takes image frames rendered with low voxel resolution (8 × 8 × 8 voxels at 90 FPS) and produces image frames equivalent to imagery produced from high voxel resolution (64 × 64 × 64 voxels) typically rendered at 3 FPS. The error was evaluated using the structural similarity image metric (SSIM). The average error between generated image frames and the ground truth recorded 92:7%±4:6%.

This paper presents a suitability evaluation approach to assess whether the LTE-based wireless private network satisfies the communication needs of smart grid applications. This approach can obtain different evaluation indexes including reliability, delay and throughput when the business attributes are used as the initial inputs by analyzing the MAC and PHY layer models and coverage characteristics. Then combined with the business analysis model, the applicability of wireless private network can be evaluated according to different power business needs. In order to verify the use of the proposed evaluation method, a simulation analysis of the suitability of the LTE 1.8GHz wireless private network is conducted for the businesses of power distribution and utilization.

In recent years, coverage-based greybox fuzzing has proven itself to be one of the most effective techniques for finding security bugs in practice. Particularly, American Fuzzy Lop (AFL for short) is deemed to be a great success in fuzzing relatively simple test inputs. Unfortunately, when it meets structured test inputs such as XML and JavaScript, those grammar-blind trimming and mutation strategies in AFL hinder the effectiveness and efficiency. To this end, we propose a grammar-aware coverage-based greybox fuzzing approach to fuzz programs that process structured inputs. Given the grammar (which is often publicly available) of test inputs, we introduce a grammar-aware trimming strategy to trim test inputs at the tree level using the abstract syntax trees (ASTs) of parsed test inputs. Further, we introduce two grammar-aware mutation strategies (i.e., enhanced dictionary-based mutation and tree-based mutation). Specifically, tree-based mutation works via replacing subtrees using the ASTs of parsed test inputs. Equipped with grammar-awareness, our approach can carry the fuzzing exploration into width and depth. We implemented our approach as an extension to AFL, named Superion; and evaluated the effectiveness of Superion on real-life large-scale programs (a XML engine libplist and three JavaScript engines WebKit, Jerryscript and ChakraCore). Our results have demonstrated that Superion can improve the code coverage (i.e., 16.7% and 8.8% in line and function coverage) and bug-finding capability (i.e., 31 new bugs, among which we discovered 21 new vulnerabilities with 16 CVEs assigned and 3.2K USD bug bounty rewards received) over AFL and jsfunfuzz. We also demonstrated the effectiveness of our grammar-aware trimming and mutation.