The recent emergence of deep learning for characterizing complex patterns in remote sensing imagery reveals its high potential to address some classic challenges this domain, e.g. scene classification. Typical models require extremely large datasets with rich contents train a multilayer structure order capture the essential features scenes. Compared benchmark used popular frameworks, however, volumes available are particularly limited, which have restricted methods from achieving full...

Vehicular edge computing (VEC) is a new paradigm that has great potential to enhance the capability of vehicle terminals (VTs) support resource-hungry in-vehicle applications with low latency and high energy efficiency. In this article, we investigate an important computation offloading scheduling problem in typical VEC scenario, where VT traveling along expressway intends schedule its tasks waiting queue minimize long-term cost terms tradeoff between task consumption. Due diverse...

Automatic modulation recognition (AMR) plays a vital role in modern communication systems. This letter proposes novel three-stream deep learning framework to extract the features from individual and combined in-phase/quadrature (I/Q) symbols of modulated data. The proposed integrates one-dimensional (1D) convolutional, two-dimensional (2D) convolutional long short-term memory (LSTM) layers more effectively time space perspective. Experiments on benchmark dataset show has efficient...

Mobile Edge Computing (MEC) is a new computing paradigm with great potential to enhance the performance of user equipment (UE) by offloading resource-hungry computation tasks lightweight and ubiquitously deployed MEC servers. In this paper, we investigate problem decision resource allocation among multiple users served one base station achieve optimal system-wide utility, which defined as trade-off between task latency energy consumption. Mobility in process considered optimization. We prove...

Unmanned aerial vehicles (UAVs) play an invaluable role in information collection and data fusion. Because of their mobility the complexity deployed environments, constant position awareness collision avoidance are essential. UAVs may encounter and/or cause danger if Global Positioning System (GPS) signal is weak or unavailable. This paper tackles problem positioning on outdoor (wildness) search scenarios by using received strength (RSS) from onboard communication module. Colored noise found...

The application of small civilian unmanned aerial vehicles (UAVs) has attracted great interest for disaster sensing. However, the limited computational capability and low energy resource UAVs present a significant challenge to real-time data processing, networking policy making, which are vital importance many related applications such as oil-spill detection flooding. In order address challenges imposed by sheer volume captured data, particularly video intermittent network resources,...

Automatic modulation recognition (AMR) is a promising technology for intelligent communication receivers to detect signal schemes. Recently, the emerging deep learning (DL) research has facilitated high-performance DL-AMR approaches. However, most models only focus on accuracy, leading huge model sizes and high computational complexity, while some lightweight low-complexity struggle meet accuracy requirements. This letter proposes an efficient based phase parameter estimation transformation,...

Network function virtualization (NFV) has been considered as a promising technology for future Internet to increase the network flexibility, accelerate service innovation, and reduce Capital Expenditures Operational costs through migrating functions from dedicated devices commodity hardware. Recent studies reveal that although this migration of brings operation unprecedented flexibility controllability, NFV-based architecture suffers serious performance degradation compared with traditional...

This article proposes a controlling framework for multiple unmanned aerial vehicles (UAVs) to integrate the modes of formation flight and swarm deployment over fixed switching topologies. Formation strategies enable UAVs enjoy key collective benefits including reduced energy consumption, but shape each UAV's freedom are significantly restrained. Swarm thus proposed maximize following simple yet powerful rules. investigates integration switch between these two strategies, considering...

The ever-expanding scale of cloud datacenters necessitates automated resource provisioning to best meet the requirements low latency and high energy-efficiency. However, due dynamic system states various user demands, efficient allocation in faces huge challenges. Most existing solutions for cannot effectively handle environments because they depend on prior knowledge a system, which may lead excessive energy consumption degraded Quality-of-Service (QoS). To address this problem, we propose...

Most feature-based stereo visual odometry (SVO) approaches estimate the motion of mobile robots by matching and tracking point features along a sequence images. However, in dynamic scenes mainly comprising moving pedestrians, vehicles, so on, there are insufficient robust static to enable accurate estimation, causing failures when reconstructing robotic motion. In this article, we proposed DynPL-SVO, complete SVO method that integrated united cost functions containing information between...

Multimodal learning has shown great potentials in numerous scenes and attracts increasing interest recently. However, it often encounters the problem of missing modality data thus suffers severe performance degradation practice. To this end, we propose a general framework called MMANet to assist incomplete multimodal learning. It consists three components: deployment network used for inference, teacher transferring comprehensive information network, regularization guiding balance weak...

This paper proposes the full interference cancellation (FIC) algorithm to cancel interrelay (IRI) in two-path cooperative system. Arising from simultaneous data transmission source and relay nodes, IRI may significantly decrease performance if it is not carefully handled. Compared with existing partial scheme, FIC approach more robust yet as complex. Numerical results are also given verify proposed scheme.

This paper presents a new, practical infrared video based surveillance system, consisting of resolution-enhanced, automatic target detection/recognition (ATD/R) system that is widely applicable in civilian and military applications. To deal with the issue small numbers pixel on developed ATD/R as are encountered long range imagery, super-resolution method employed to increase signature resolution optimise baseline quality inputs for object recognition. tackle challenge detecting extremely...

We develop an automatic oil spill segmentation method in terms of f-divergence minimization. exploit for measuring the disagreement between distributions ground-truth and generated segmentations. To render tractable optimization, we minimize tight lower bound by adversarial training a regressor generator, which are structured different forms deep neural networks separately. The generator aims at producing accurate segmentation, while characterizes discriminative with respect to true It is...

Considering that low-cost and resource-cons- trained sensors coupled inherently could be vulnerable to growing numbers of intrusion threats, industrial Internet-of-Things (IIoT) systems are faced with severe security concerns. Data sharing for building high-performance detection models is also prohibited due the sensitivity, privacy, high value IIoT data. This article presents an anomaly-based system federated learning privacy-preserving machine in future networks. To tackle urgent issue...

This article proposes an autoencoder-based method to enhance the information interaction between in-phase/quadrature (I/Q) channels of input data for automatic modulation recognition (AMR). The proposed utilizes autoencoder built by fully-connected layers correlate features I/Q and obtain feature from intermediate layer, which is concatenated together with original as model inputs. To accommodate new dimensions, a modification scheme existing representative deep learning based AMR (DL-AMR)...

Multi-UAV systems rely on the communication network to exchange mission-critical data for their coordination and deployment, while delays could cause significant challenges both tasks. The impact of becomes even more severe if delay, structure formation are all time-varying, a common challenge faced by real-world multiUAV systems. To address this challenge, we consider time-varying that exist in multiple channels caused transmitting information internal UAVs themselves obtaining processing...

Landslides often contain boulders on their surface or within the landslide body. Embedding sensors inside a may help monitor its movement and dynamics. In this study, smart were tested for tracking movements of cobble, estimating magnitude mode in dedicated laboratory experiments. The cobble was embedded with sensor equipped accelerometers, gyroscopes, magnetometers. experiments consisted letting travel down an inclined plane. By changing angle plane, showed different modes such as rolling...