Withthe development of platform technology and sensor technology, unmanned platforms are widely used in underwater operations, and the functional requirements are gradually increasing withthe application scenarios, ...
Withthe development of platform technology and sensor technology, unmanned platforms are widely used in underwater operations, and the functional requirements are gradually increasing withthe application scenarios, which also put forward higher requirements for computing performance. In this paper, a real-time signalprocessing heterogeneous cooperative system based on DSP and FPGA is designed. DSP is used as the main processor to schedule multiple subordinate FPGA processors to complete heterogeneous multitask high-speed signalprocessing tasks. In order to verify the feasibility, this paper implements the real-time signalprocessing application of underwater unmanned platform imaging algorithm based on this architecture. the experiment shows that the DSP-FPGA heterogeneous collaborative processing framework can effectively implement the imaging algorithm, complete the online real-time signalprocessing and output, meet the requirements of low delay, high-speed parallel real-time signalprocessing of the unmanned platform, and verify the reliability of the system in the actual lake test.
Most of existing studies on neural network pruning only consider memory-based pruning strategies. However pruning for computational workload is often more important in hardware deployments due to a greater focus on mo...
详细信息
To solve the problem of passive localization in Unmanned Aerial Vehicles (UAV) formation, a UAV passive localization model based on formation azimuth optimization algorithm is proposed. By solving the equation based o...
To solve the problem of passive localization in Unmanned Aerial Vehicles (UAV) formation, a UAV passive localization model based on formation azimuth optimization algorithm is proposed. By solving the equation based on the formation characteristics, the model can achieve relatively accurate positioning and adjustment of the UAV with position deviation in the formation. this paper demonstrates the specific model and adjustment steps of the method in the scenario of 15 UAVs cone-shaped formation and 10 UAVs circular formation, and the simulation and error analysis of the model are carried out. According to the analysis, the positioning accuracy is obviously improved after optimization, which proves that the method is feasible.
this article studies big data technology, analyzes the relationship between various factors of urban road traffic and the occurrence of traffic accidents, and establishes a prediction model based on this. By collectin...
this article studies big data technology, analyzes the relationship between various factors of urban road traffic and the occurrence of traffic accidents, and establishes a prediction model based on this. By collecting and processing a large amount of traffic data, including traffic flow, speed, weather, and other information, and combining historical traffic accident records, a prediction model based on machine learning was constructed. the model can monitor road traffic conditions in real time and predict the probability of traffic accidents. In addition, we have also verified and evaluated the model, proving its accuracy and practicality. this research contributes to improving the efficiency of traffic management and road safety, and has important practical significance.
Aiming at the problem that the detection of direct sequence spread spectrum (DSSS) signal under the jamming signals with different bandwidths, a detection method for DSSS signal based on decision tree is proposed in t...
Aiming at the problem that the detection of direct sequence spread spectrum (DSSS) signal under the jamming signals with different bandwidths, a detection method for DSSS signal based on decision tree is proposed in this paper. First, features are extracted as detection statistics based on cepstrum and time-domain cross-correlation, respectively. And then the decision tree is used to make the detection decision by comparing the detection statistics withthe predefined thresholds. After that, the optimal thresholds are determined through a global search algorithm, and the optimal thresholds are used for DSSS signal detection. the simulation results demonstrate that when the interference to signal ratio (ISR) is -1dB, the probability of detection reaches 95%.
To improve the user experience in a multi-user multi-edge server system, this paper propose a task offloading algorithm based on soft Actor-Critic (SAC) and Federated Learning (FL), named FLSAC. this algorithm fully c...
详细信息
ISBN:
(数字)9798350376548
ISBN:
(纸本)9798350376555
To improve the user experience in a multi-user multi-edge server system, this paper propose a task offloading algorithm based on soft Actor-Critic (SAC) and Federated Learning (FL), named FLSAC. this algorithm fully considers the requirements of delay-sensitive tasks, the resource utilization of edge servers, energy and memory limitations. FLSAC adopts a continuous offloading strategy and sets the weighted average of all users' delay, total energy consumption, and task offloading rate as the user cost. Withthe goal of minimizing user costs, each task's data offloading percentage is used as the learning target to better meet the needs of delay-sensitive tasks. Experimental results show that compared with Twin Delayed Deep Deterministic Policy Gradient (TD3) and Proximal Policy Optimization (PPO), FLSAC's user total cost decreases by 13.3% and 38.6%, respectively, with good stability and convergence. At the same time, edge computing networks have good scalability, alleviating the problem of insufficient generalization caused by data scarcity in some edge computing servers in practical use. Finally, this paper analyze the time complexity of FLSAC.
the P300-based brain-machine interface can achieve efficient communication between human brain and computer, and help physically disabled people achieve autonomous control of external devices. In this paper, we propos...
the P300-based brain-machine interface can achieve efficient communication between human brain and computer, and help physically disabled people achieve autonomous control of external devices. In this paper, we propose a convolutional neural network-based signal detection method, which includes signal pre-processing, construction and training of convolutional neural network, and testing and performance evaluation of the model. this paper conducts experimental validation on dataset II of the 3rd BCI competition and compares it with some traditional machine learning methods and deep learning methods. the test accuracy of the experimental results reaches 0.85 and the AUC reaches 0.93, which indicates that the method has significantly improved in classification accuracy and model performance, and has high practicality and application prospects.
Ultra Wide Band radar life detection technology has been widely used in life detection in various scenes because of its high life detection accuracy, but it is difficult to extract vital signs in the process of echo s...
Ultra Wide Band radar life detection technology has been widely used in life detection in various scenes because of its high life detection accuracy, but it is difficult to extract vital signs in the process of echo signalprocessing because of scene ***, a research on the processing method of UWB radar echo signal is *** mean band-pass filter is used to remove and enhance the background noise of vital signs in UWB echo signal, and the vital signs signal is located by combining energy moment calculation. In order to improve the accuracy of vital sign signal extraction, and according to the linear characteristics of vital sign signals, an adaptive EWT algorithm is proposed by combining Pearson correlation coefficient with high computational efficiency and EWT algorithm with strict mathematical basis and no modal aliasing and endpoint effect problem, then the located vital sign signals are further filtered and extracted by this ***, the effectiveness of the proposed method is verified by experiments.
Contraposing to the problems in image features, grayscale tone performance and artistic effect expression of artistic style rendering technology in simulating handwritten signature to draw character portrait, this pap...
详细信息
this paper presents the design of a simple portable large pulse signal generator, which is used to drive high-brightness LED luminescence and provide analog strong light source for photoelectric detector, so as to che...
this paper presents the design of a simple portable large pulse signal generator, which is used to drive high-brightness LED luminescence and provide analog strong light source for photoelectric detector, so as to check the working state of system in pulse radiation measurement. this design is based on programmable CPLD device and two-stage drive circuit to generate single pulse signal with adjustable pulse width and amplitude manually. the prototype withthe advantages of small volume, light weight, large output signal amplitude and low cost, which has been successfully applied in strong pulse radiation measurement for large-scale equipment such as accelerator.
暂无评论