Publications
& Patents

Abstract: Due to the limited computing resources of edge devices, traditional object detection algorithms struggle to meet the efficiency and accuracy requirements of autonomous driving. Consequently, designing a neural network model that balances hardware resource requirements, operating speed, and accuracy is crucial. To address this, by integrating algorithm with hardware characteristics, we propose a lightweight neural network architecture search algorithm based on in-place distillation and performance predictor (LNIP). Initially, we focus on optimizing the operators of the you only look once version 8 nano (YOLOv8n) and dynamically adjust its network structure. Then, we trained a super-network using a progressive shrinking strategy, the sandwich rule, and in-place distillation. Subsequently, we employed a Gaussian process to model the relationship between network architecture and accuracy, utilizing encoding methods and custom kernel function to develop high-performance predictor. Finally, during the search process, we introduce a reward function based on Pareto optimality to balance the performance of the model with hardware constraints. Building upon this foundation, we design an efficient search algorithm based on the performance predictor to progressively explore the optimal network structure tailored to hardware characteristics. We compared our lightweight network with state-of-the-art methods on the BDD100K, COCO, and PASCAL VOC datasets and deployed it on the Black Sesame A1000 and NVIDIA Xavier for comprehensive evaluation. On the NVIDIA Xavier, the lightweight network achieves a latency of 11.81 ms and an edge precision of 46.1%. These experimental results demonstrate that our method outperforms existing methods in balancing hardware constraints and model performance.

Keywords: Neural network architecture search; Performance predictor; Pareto optimality; In-place distillation; Object detection.

Abstract: The work develops alternating-projection algorithms in the range-Doppler and space-Doppler domains to suppress folded clutter and resolve range ambiguities in pulse-Doppler radar using pulse-agile waveforms.

Keywords: Pulse-Doppler radar; pulse agility; folded clutter; range ambiguity; alternating projection.

Abstract: This study jointly optimizes diverse phase-coded frequency-modulation waveforms and their mismatched filters, seeking lower range sidelobes while preserving useful pulse-compression performance.

Keywords: PCFM waveform; waveform diversity; joint design; mismatched filter; range sidelobes.

Abstract: A dynamic alternating-projection procedure is presented for efficient mismatched-filter design, targeting practical sidelobe control with a fast iterative computation.

Keywords: Mismatched filter; dynamic alternating projection; radar pulse compression; sidelobe suppression.

摘要：本文将动态规划用于捷变脉冲串的优化设计，以高效搜索满足波形设计要求的脉冲序列。

关键词：捷变脉冲串；动态规划；波形优化；雷达信号处理。

Abstract: The paper introduces a Doppler-tuned matched-filter formulation and a two-dimensional mismatched filter for suppressing range-Doppler sidelobes in pulse-diverse radar, with constraints on SNR loss and mainlobe width.

Keywords: Pulse diversity; range-Doppler sidelobes; two-dimensional mismatched filter; pulse-Doppler radar; SNR loss.

Abstract: This conference paper optimizes a pulse-agile sequence built from diverse nonlinear-FM waveforms to reduce range-folded clutter in radar processing.

Keywords: Pulse agility; nonlinear FM; folded clutter; sequence optimization; radar waveform diversity.

Abstract: The work designs a pulse-agile nonlinear-FM waveform family through spectrum modulation, providing waveform diversity for radar transmission and processing.

Keywords: Nonlinear FM; spectrum modulation; pulse agility; radar waveform design.

Abstract: This paper investigates a linear-FM pulse-train design in which pulse width is varied across pulses to provide waveform agility.

Keywords: Linear FM; pulse train; pulse-width agility; radar waveform design.

Abstract: This work considers the design and processing of frequency-hopped synthetic-bandwidth waveforms for radar self-speed measurement.

Keywords: Frequency hopping; synthetic bandwidth; self-speed measurement; radar waveform; signal processing.

Abstract: The ISO-MMF method designs unimodular sequences and associated mismatched filters with controllable correlation sidelobes over selected regions, using a closed-form solution for efficient implementation.

Keywords: Unimodular sequence; mismatched filter; correlation sidelobes; ISO-MMF; radar and sonar.

Abstract: A sub-aperture EVD method divides a full ISAR aperture into partially overlapping segments, estimates phase error in each segment, and combines them to improve autofocus efficiency.

Keywords: ISAR; phase autofocus; eigenvalue decomposition; sub-aperture; translational motion compensation.

Abstract: The paper uses the ADAM optimization algorithm to design unimodular phase-coded waveforms whose ambiguity functions suppress sidelobes within prescribed range-Doppler regions.

Keywords: Unimodular waveform; phase coding; ambiguity function; ADAM; range-Doppler sidelobes.

Abstract: This work proposes a weighted mismatched-filter approach for reducing the range sidelobes of radar pulse-compression outputs while managing the associated signal-to-noise and mainlobe trade-offs.

Keywords: Mismatched filter; weighted design; range sidelobes; pulse compression; radar waveform.

Abstract: The ISO-CA method creates window functions with controllable mainlobe width, peak sidelobe level, sidelobe fall-off rate, and steerable sidelobes, and is evaluated with simulations and measured data.

Keywords: Window function; spectral characteristics; mainlobe width; peak sidelobe level; steerable sidelobes.

Abstract: This paper develops an eigenvalue-decomposition approach for estimating and compensating ISAR phase errors, enabling efficient autofocus of range-Doppler images.

Keywords: ISAR; phase autofocus; eigenvalue decomposition; motion compensation; radar imaging.

Abstract: The paper modifies a single-eigenvector ISAR autofocus strategy to estimate phase error for translational-motion compensation and improve image focus.

Keywords: ISAR; autofocus; single eigenvector; phase-error estimation; radar imaging.