Video-guided Machine Translation (VMT) seeks to enhance translation quality by incorporating contextual information derived from paired short video clips.However, many VMT samples are text-sufficient; even when visual information is needed, only minimal cues are required.Aiming to tackle these issues, we propose a novel framework DART (Disambiguation-Aware Reasoning for Videoguided Machine Translation).Reinforcement learning is used to incorporate multimodal large language models' multimodal reasoning into VMT.The model dynamically switches between text-only processing and multimodal integration, contingent on the necessity of visual disambiguation.Furthermore, we present TVRF (Translation-oriented Video Relevance Filtering), a systematic pipeline for constructing training data based on multimodal relevance to translation.This pipeline filters samples where video information is translationrelevant, mitigating training collapse caused by video-irrelevant data in conventional VMT.Experimental results show that our approach improves multimodal information utilization in VMT, yielding gains in both translation quality and computational efficiency.* Equal corresponding authors.Reasoning: Okay, I need to translate the input sentence ... So the answer is "哦！".(809 tokens in total) DA T Existing LMRMs … "It's hitting his yacht."Translation: 哦！(Oh!) Reasoning: Okay, so I need to translate the sentence ..

Video-guided Machine Translation (VMT) aims to improve translation quality by integrating contextual information from paired short video clips.Mainstream VMT approaches typically incorporate multimodal information by uniformly sampling frames from the input videos.However, this paradigm frequently incurs significant computational overhead and introduces redundant multimodal content, which degrades both efficiency and translation quality.To tackle these challenges, we propose SHIFT (Selected Helpful Informative Frame for Translation).It is a lightweight, plug-andplay framework designed for VMT with Multimodal Large Language Models (MLLMs).SHIFT adaptively selects a single informative key frame when visual context is necessary; otherwise, it relies solely on textual input.This process is guided by a dedicated clustering module and a selector module.Experimental results demonstrate that SHIFT enhances the performance of MLLMs on the VMT task while simultaneously reducing computational cost, without sacrificing generalization ability.

This paper presents a robust sparse direct position determination (DPD) method for multiple emitters using distributed sensor arrays in the presence of unknown gain-phase errors. The proposed method tackles the problem under a block sparse Bayesian learning (BSBL) framework, which incorporates perturbed steering vector factorization to separate the position parameter from the gain-phase errors, making dictionary completely known without learning. This paper devises a customized hyperparameter update rule for the proposed DPD model within the foundation of the BSBL-EM method, allowing for varying block parameters instead of constraining them to be consistent. The position estimates of emitters are determined by calculating the mean value of the posterior distribution of the reconstructed waveforms. Simulations demonstrate the superior performance of the developed BSBL direct localization method over its state-of-the-art rivals, which exhibits enhanced localization accuracy and robustness against gain-phase errors.

Direct position determination (DPD) has received much attention in emitter localization, owing to its better accuracy than conventional two-step positioning. Most of the existing DPD algorithms are developed for circular signals (CS) by using uniform linear arrays (ULAs). However, these algorithms may ignore other characters of the signals, e.g., noncircularity. The use of ULAs limits the accuracy of source localization and the number of sources that can be estimated. In this article, a weighted$\ell_{0}$-norm sparse reconstruction algorithm for noncircular signals (NCS) is developed for DPD with a designed sparse array in motion. First, a sparse array configuration named extended nested array (ENA) is devised for NCS, which consists of three subarrays. Theoretical analysis proves that the designed array can obtain higher degrees of freedom (DOFs) effectively, and reduce the mutual coupling effects between antennas. Then, a weighted$\ell_{0}$-norm sparse reconstruction algorithm is developed to improve the accuracy of DPD. Finally, simulation results are provided to demonstrate the superiority of the proposed algorithm with the designed sparse array. Our scheme can provide better localization performance than the state-of-the-art methods.

Abstract Polarisation‐sensitive array (PSA) has earned extensive attention in satellite communication owing to excellent anti‐interference performance. Nevertheless, the configuration of traditional PSA where each polarised antenna requires multiple radio frequency (RF) front‐ends makes it much more costly in terms of hardware and software for system design. In this paper, resorting to RF switches, a novel reconfigurable PSA optimisation technique is developed that utilises fewer RF front‐ends, which can considerably decrease computational expenditure and achieve high anti‐jamming performance. An RF switch switching (RFSS) scheme is devised to guide the implementation of PSA reconfiguration. To accurately demonstrate the effect of the array configuration on interference rejection performance, the polarisation‐spatial subspace correlation coefficient (PSC) is presented. Then the relationship between the optimal signal to interference plus noise ratio (SINR) and the PSC based on adaptive processing is formulated. Aiming to acquire the optimal reconstructed PSA outputting the maximum SINR, the mathematical model of the PSA reconfiguration problem is established. Subsequently, two optimisation methods are provided to address the problem efficiently, thereby gaining the optimal configuration of the reconstructed PSA and implementing the reconfiguration by RF switches. Numerical and experimental simulations verify the correctness and reliability of the developed scheme and approaches.

Synthetic Aperture Radar (SAR) has the characteristics of all-time, all-weather, strong penetrating power and high spatial resolution. The monitoring of flood using remote sensing based on the SAR data is of great significance to scientifically manage and analyze the flood disasters. The flood disaster in China's Yi River Basin in 2020 was selected to realize the monitoring of floods and disaster assessment based on Sentinel-1 B Sar data. First, the Sentinel-1 Dual-Polarized Water Index (SDWI) was adopted to enhance the water body information, and the object-oriented threshold method was used to extract the water area. Next, high spatial resolution images were obtained to verify the accuracy of water extraction. Finally, combined with land classification data, the assessment of disaster-affected areas was carried out. The results shows that the producer precision, user precision, and overall precision of water extraction in the disaster-affected area are all higher than 95%, which indicate that the stability and reliability of this method.

Abstract In high‐speed mobile scenarios, due to the high‐speed relative motion between transmitter and receiver, the high Doppler frequency shift interferes with the inter‐subcarrier orthogonality in orthogonal frequency‐division multiplexing (OFDM) systems. Therefore its performance is significantly degraded. Recently, orthogonal time–frequency space (OTFS) is considered as an effective alternative scheme to OFDM for time‐varying channels. As with OFDM‐massive multiple input multiple output (MIMO) systems, downlink channel estimation is necessary for OTFS‐massive MIMO systems to improve the spectral efficiency in frequency‐division duplex (FDD) mode without channel reciprocity. Here, first the cyclic prefix ‐OFDM‐based massive MIMO OTFS system channel with antenna directivity pattern is analyzed, and transform the burst sparsity in the angle domain into block sparsity by using non‐uniform Fourier transform (NUFT). Furthermore, to solve the problem that the pilot overhead grows linearly with the number of antennas, we propose a three‐dimensional (3D) dynamic support detect (DSD) algorithm. Compared with the traditional OMP algorithm, and the 3D‐ structured orthogonal matching pursuit algorithm, simulation results demonstrate the proposed DSD algorithm has higher channel estimation accuracy, and lower pilot overhead.

Abstract A sparse reconstruction algorithm for the localisation of mixed near‐field and far‐field sources (MFNS) based on four‐order statistics is proposed in this study. First, utilising the structural characteristics of a uniform symmetric linear array, a fourth‐order cumulant (FOC) matrix is constructed, which decouples the angular information from the range parameters. Based on the sparse representation framework, a weighted l 1 ‐norm minimisation algorithm is developed to obtain the direction of arrivals (DOAs) of the MFNS. However, the existing selection strategy of the tuning factor is not adaptive to different observation scenarios. So a closed‐form expression of the tuning factor based on the FOC estimation error is presented. Then, another FOC matrix is constructed, which includes both the DOA and range information of the MFNS. With the DOA estimates, the two‐dimensional spatial dictionary can be reduced into a one‐dimensional dictionary, which only depends on the range parameters. Using the similar sparse reconstruction method, the range estimates of the MFNS can be obtained, and the types of the sources can be distinguished according to their range parameters. According to numerical simulations, the estimation performance of the proposed algorithm approaches the CRB in the high signal‐to‐noise ratio region, which successfully circumvents the saturation problem due to the fixed tuning factor.

In this paper, a novel efficient algorithm is developed for direction of arrival (DOA) estimation of coherent signals under the direction-dependent mutual coupling (DDMC) based on weighted subspace fitting. DOAs are determined by applying the least square fitting between signal space and the modified array manifold at first. Subsequently, we put forward an approach to calculate the DDMC matrices and the complex fading coefficients by utilizing the estimated DOAs. Without any iterations, the proposed algorithm can identify the angular information of the coherent signals in a single step in the presence of DDMC. Numerical simulation results show the effectiveness of the proposed algorithm.

On the ground with complex electromagnetic environment, protecting the received satellite signals from interference is a key issue for the receiver. To effectively cope with the coexistence of jamming and spoofing interference, and accurately suppress both jamming and spoofing, in this paper, a hybrid interference suppression algorithm based on robust beamforming is proposed. The combination of subspace projection algorithm, despreading algorithm, multiple signal classification (MUSIC) algorithm and robust beamforming based on the linearly constrained minimum variance criterion can suppress both jamming and spoofing effectively, and ensure that the desired signal is undistorted. At the same time, it can overcome the problem of inaccurate direction estimation under low signal-to-noise ratio. Numerical examples show that the proposed algorithm can effectively suppress hybrid interference.