A5057561043- Wireless Signal Modulation Classification
- Advanced Neural Network Applications
- Advanced Memory and Neural Computing
- Advanced Photonic Communication Systems
- CCD and CMOS Imaging Sensors
- Advanced Computational Techniques and Applications
- Radar Systems and Signal Processing
- Robotics and Sensor-Based Localization
- Visual Attention and Saliency Detection
- Service-Oriented Architecture and Web Services
- Advanced Database Systems and Queries
- Physical Unclonable Functions (PUFs) and Hardware Security
- Neural Networks and Reservoir Computing
- Integrated Circuits and Semiconductor Failure Analysis
- Terahertz technology and applications
- Advanced Manufacturing and Logistics Optimization
- Neuroscience and Neural Engineering
- Advanced SAR Imaging Techniques
- Web Data Mining and Analysis
- Banking Systems and Strategies
- Industrial Technology and Control Systems
- Data Management and Algorithms
- Advanced Multi-Objective Optimization Algorithms
- Semantic Web and Ontologies
- Machine Learning and ELM
University of Electronic Science and Technology of China
2018-2025
Jiaxing University
2023-2024
Nanjing Normal University
2007
In wireless communication security and spectrum management, specific emitter identification (SEI) is a potential technology to identify individual emitters. Recently novel SEI algorithm based on complex-valued neural network (CVNN) has emerged, which exhibits powerful processing capabilities in complex domains. However, it also brings high computational complexity, makes difficult meet the power efficiency requirements of system. To address above challenge, Zynq-based platform with...
PUF can be used for IoT device authentication. This paper proposes a novel FPGA-based RO with improved uniqueness and reliability. Firstly, the proposed improves by LUT-based self-compare structure, which reduces delay bias from systematic variations without special constraint on place & route selection of challenge response pairs. Secondly, reliability adaptive counter time period tuning based real-time measured stability. Implemented Xilinx Spartan-6 FPGA, shows better than several...
This letter presents an energy-efficient reconfigurable AI-based object detection and tracking processor for smart drone/robot applications. Several techniques have been proposed to achieve high energy efficiency while supporting flexible tasks with online learning, including a architecture neural network (NN) engine, learning shared NN inference engine automatic label generation layer- stride-aware computing technique. Compared several state-of-the-art designs, the design achieves better...
As the deeply and broadly applying GIS, GIS has turned into main technology of management spatial information as well integrated platform carried with information, gradually melted mainstream IT. However, due to content source diversity data difference software, application is still a "black box" which leads difficulties functions sharing between applications. Hence, both study On interoperability solution on bottleneck applications have become most valuable research in domain. Based Web...
Smart robots (e.g. drones) for object detection & tracking demand embedded intelligent processors. Neural network (NN) processors have been designed to accelerate NN pattern recognition [1] [2]. However, these designs lack special processing engines such as bounding box (bbox) calculation and selection. Also, their architectures are general AI tasks resulting in redundancy/inefficiency performing tracking. An processor has proposed previously [3], but it only supports specific does not...
As a potential air control measure, RF-based surveillance is one of the most commonly used unmanned aerial vehicles (UAV) methods that exploits specific emitter identification (SEI) technology to identify captured RF signal from ground controllers UAVs. Recently many SEI algorithms based on deep convolution neural network (DCNN) have emerged. However, there lack implementation hardware. This paper proposes high-accuracy and power-efficient hardware accelerator using an algorithm-hardware...
While the Internet of Things (IoT) technology is booming and offers huge opportunities for information exchange, it also faces unprecedented security challenges. As an important complement to physical layer technologies IoT, radio frequency fingerprinting (RFF) great interest due its difficulty in counterfeiting. Recently, many machine learning (ML)-based RFF algorithms have emerged. In particular, deep (DL) has shown benefits automatically extracting complex subtle features from raw data...
With the development of wireless communication technology and increasingly complex electromagnetic environment, radio frequency fingerprinting (RFF) plays a vital role in improving security information systems. Recently, many RFF algorithms based on machine learning have emerged. However, most them focus accuracy identification but ignore computational cost. This paper proposes novel classifier composed convolutional neural network (CNN) backbone with two random forest branches, called...
Specific emitter identification (SEI) plays a crucial role in spectrum management applications. Recently many SEI algorithms based on deep convolution neural network (DCNN) have been proposed. However, the high computational complexity of DCNN leads to large power consumption. Also, there is lack dedicated hardware design. In this work, we proposed SNR-aware adaptive scalable algorithm DCNN, and implemented power-efficient accelerator it. The can adaptively reconfigures itself between 16-bit...
This paper proposes a new method to parse various application schemas of Geography Markup Language (GML) for understanding syntax and semantic their element type in order implement uniform interpretation the same GML instance data among diverse users. The proposed generates an Integrative Syntactic Semantic Schemas Database (IGSSSDB) from GML3.1 core corresponding schema. parses based on IGSSSDB, which is composed syntactic information, nesting information mapping rules schemas. Three kinds...
This paper proposed RHIG: R+ tree-based Holistic Index of GML after sufficiently extracting character spatial data and analyses traditional XML index technology. RHIG possesses following characters: Fully utilizing structure information in schema, encode both schema document document, which enable each element take relevant thus enhance efficiency attribute querying; support simultaneously indexing querying on non-spatial contained document; could be stored relational database to avoid...
Existing deep convolutional neural network (DCNN) processors are mainly designed for high-end applications such as autonomous vehicle, data center and smart phone where the design focus is performance, while intelligent sensing devices power efficiency more important. In addition, programmability important DCNN to support different DCNN. We have proposed a power-efficient programmable processor dedicated demonstrated it using FPGA. Several techniques been improve efficiency. Implemented on...