Due to the flexibility and ease of deployment of Field Programmable Gate Arrays (FPGA), more and more studies have been conducted on developing and optimizing target detection algorithms based on Convolutional Neural Networks (CNN) models using FPGAs. Still, these studies focus on improving the performance of the core algorithm and optimizing hardware structure, with few studies focusing on the unified architecture design and corresponding optimization techniques for the algorithm model, resulting in inefficient overall model performance. The essential reason is that these studies do not address arithmetic power, speed, and resource consistency. In order to solve this problem, we propose a deep learning acceleration core architecture based on FPGAs, which is designed for target detection algorithms with CNN models, using multi-channel parallelization of CNN network models to improve the arithmetic power, using scheduling tasks and intensive computation pipelining to meet the algorithm’s data bandwidth requirements and unifying the speed and area of the orchestrated computation matrix to save hardware resources. The proposed framework achieves 14 Frames Per Second (FPS) inference performance of the TinyYolo model at 5 Giga Operations Per Second (GOPS) with 30% higher running clock frequency, 2–4 times higher arithmetic power, and 28% higher Digital Signal Processing (DSP) resource utilization efficiency using less than 25% of FPGA resource usage.

Getting higher precision positioning and velocity result with lower cost in urban environment is an important pursuit of modern intelligent transportation, autonomous driving, and other fields. While a low-cost inertial navigation system (INS) and global navigation satellite systems (GNSS) integrated system have the potential to achieve the pursuit, its performance can be significantly affected by the occasional GNSS outage and variable measurement noise in complex driving environments. Inspired by AI-type solutions for predicting vehicle behavior, and to get rid of their disadvantage of environmental sensitive, we abstract vehicle behaviors in urban environments as constraint models. A multiple constraint fused Kalman-based navigation (MCKN) method for INS/GNSS integrated navigation system is introduced by combining multiple model interaction principle with zero-velocity update (ZVU) error correction algorithm. This novel MCKN approach allows multiple zero-velocity constrained models to work in parallel. The optimal navigation results are obtained by calculating the weights of each model at every positioning epoch. Road test results in the Zhongguancun area of Beijing show that the MCKN method effectively limits the accumulation of low-cost INS errors. During 1-hour experiment, the positioning and velocity error of MCKN method in ±3σ interval were 1.72 m and 0.12 m/s respectively, maintaining a high-level performance.

With the widespread global application of BeiDou navigation, BeiDou B1C signaling based on Quadrature Multiplexed Binary Offset Carrier (QMBOC) modulation is expected to be extensively used in urban environments due to its wider signal bandwidth, smaller code pseudorange measurement errors, and stronger multipath capabilities. Despite offering higher positioning accuracy and secondary modulation characteristics of the BeiDou, B1C signals introduce the challenge of multiple peaks in the autocorrelation function. This leads to phase ambiguity during signal acquisition and tracking, resulting in positioning deviations of tens or even hundreds of meters. In urban environments, such deviations give rise to significant practical application issues. To address this problem, we have designed a multi-loop structure for the synchronous tracking of B1C signals and proposed a multi-peak phase-deblurring algorithm specifically tailored for the BeiDou B1C signal in urban environments. This algorithm considers the coupling relationship between the code and the carrier loops, and by matching the structural design of multiple loops, it achieves a precise and unambiguous phase estimation of the pseudocode, enabling the stable tracking of the entire loop for the BeiDou B1C signal. Simulation and actual testing demonstrate that the algorithm exhibits an error less than 0.03 for chip intervals when the signal-to-noise ratio is greater than −20 dB. Additionally, the accuracy can be improved by adjusting the set conditions, making it suitable for urban environments.

With the large-scale use of BeiDou navigation and 5G technology worldwide, integrating BeiDou navigation and communication has become a hot research topic in navigation and positioning technology. Low-cost, miniaturized, and susceptible mixed-signal receivers will become the future receiver technology development trend. However, the current receiver technology still faces the challenge of further improving the positioning service capability and communication quality, which includes the lack of practical analysis of the compatibility between signals and the lack of mixed-signal processing capability of the receiver baseband key technology. To address these problems, we start by analyzing the signal part of 5G out-of-band signals falling into the BD B1 signal band, conduct a detailed analysis of the mixed signal regime and frequency planning, and design a hybrid receiver architecture compatible with both signals, and propose an SC-PMF-FFT fast capture algorithm based on strong correlation, which takes advantage of the strong correlation of signals broadcast on the BD B1 frequency point from B1I to B1C, and reuses the structure of the CDMA system signal capture algorithm to complete the fast capture of 5G signals using an OFDM system. The experiments show that the method can capture the BeiDou B1 signal with a sensitivity of −154 dBm and a whole constellation capture time of no more than 40 ms with the inlet power of the 5G signal not exceeding −45 dBw.

Marx circuit based on avalanche transistors could be used to produce high power-nanosecond-Gaussian pulse. In this paper, theoretical calculations and software simulation of circuit parameters on the impact of the pulse waveform parameters are analyzed. The results show us that the charging voltage mainly affect amplitude and rise time of the pulse generated, the charging capacitor mainly impact the amplitude and fall time of the pulse, charging resistors could limit the maximum pulse repetition frequency. This paper provides a theoretical reference for the selection of components in the design of pulse source circuit and the researches of adjustable Gaussian pulse source.