As an interesting and challenging problem, generating image caption automatically has attracted increasingly attention in natural language processing and computer vision communities. In this paper, we propose an end-to-end deep learning approach for image caption generation. We leverage image feature information at specific location every moment and generate the corresponding caption description through a semantic attention model. The end-to-end framework allows us to introduce an independent recurrent structure as an attention module, derived by calculating the similarity between image feature sequence and semantic word sequence. Additionally, our model is designed to transfer the knowledge representation obtained from the English portion into the Chinese portion to achieve the cross-lingual image captioning. We evaluate the proposed model on the most popular benchmark datasets. We report an improvement of 3.9% over existing state-of-the-art approaches for cross-lingual image captioning on the Flickr8k CN dataset on CIDEr metric. The experimental results demonstrate the effectiveness of our attention model.

Ensuring physical layer security is crucial for Dual-Function Radar-Communication (DFRC) systems. The synergism of Artificial Noise (AN) and Reconfigurable Intelligent Surface (RIS) is proposed to improve transmission security. Aiming to maximize the secrecy rate, beamforming for transmit signals and AN or jamming signals at the dual-function radar and base station (radar-BS), along with passive reflection phase-shift matrix (PSM) of the RIS are jointly optimized. This optimization is subject to satisfying the transmit power budget at the radar-BS, the unit modulus constraint at the RIS, and the SINR thresholds for both communication and radar sensing at the radar-BS. To render this nonconvex problem tractable, we decompose it into three convex sub-problems, which are solved by reformulating into fractional programming(FP) problems and employing the techniques of successive convex approximation (SCA) and semidefinite relaxation (SDR). Numerical simulations demonstrate that the proposed algorithm can help improve the transmission security of DFRC systems while ensuring target sensing performance.

The acquisition of accurate channel state information is critical for enhancing the transmission quality and energy efficiency of reconfigurable intelligent surface-aided millimeter wave systems with multiple-input multiple-output antenna arrays at the transceiver ends. Due to the cascaded doubly sparse channel property in angular and delay domains, it is difficult to acquire nonzero angular gains sampled by overcomplete dictionaries with low pilot overhead. In this work, a channel-estimation method based on the sparse Kalman filter (SKF) framework is proposed, in which a state-space model is established for a sparse complex angular gain vector in both static and dynamic scenarios, and a zero-padding block-diagonal pilot pattern is designed. A linear l ₁ -norm subdifferential pseudo-measurement equation is deduced. The on-grid angular gain vector, along with the corresponding support, is recovered by the sequential processing of multiple pilot tone observations. To enhance the l ₁ -norm convergence and thus reduce the pilot overhead, an accelerated algorithm is developed, which incorporates an exponential factor to accelerate the gradient descent procedure. The simulation results demonstrate that the method can capture the angular support using a few pilot blocks. The normalized mean square error performance for the angular gain vector was better than that for the orthogonal matching pursuit. The accelerated SKF algorithm achieves faster convergence and higher accuracy compared with the ordinary SKF algorithm, and it is easy to implement online.

Sparse arrays are widely employed in array signal processing due to their obvious advantages in array element distribution and uniform degrees of freedom (uDOFs). In this paper, a generalized augmented multi-subarray nested array (GAMSNA-I) and its variant, GAMSNA-II are proposed, with the objective of increasing uDOFs and reducing mutual coupling. Based on two subarrays of the prototype nested array (NA), GAMSNA-I is constructed by reconfiguring the dense uniform linear array (ULA) and forward-shifting the sparse ULA. GAMSNA-II is obtained by sparsifying the dense part of GAMSNA-I, ensuring constant uDOFs while further reducing mutual coupling. Subsequently, the closed-form expression for the uDOFs of GAMSNA-I with an arbitrary number of sensors is derived, and the proof is provided that the uDOFs of GAMSNA-II remain unchanged relative to that of GAMSNA-I. Compared to some existing array configurations, both GAMSNA-I and GAMSNA-II exhibit improved uDOFs, with GAMSNA-II achieving lower mutual coupling. Simulation results show the superior performance of the proposed GAMSNA-I and GAMSNA-II.

Single-snapshot direction-of-arrival (DOA) estimation plays an important role in dynamic target detection and tracking applications. Because a single-snapshot signal provides few information for statistics calculation, recently compressed sensing (CS) theory is applied to solve single-snapshot DOA estimation, instead of the traditional DOA methods based on statistics. However, when the unknown sources are closely located, the spatial signals are highly correlated, and its overcomplete dictionary is made up of dense grids, which leads to a serious decrease in the estimation accuracy of the CS-based algorithm. In order to solve this problem, this paper proposed a two-step compressed sensing-based algorithm for the single-snapshot DOA estimation of closely spaced signals. The overcomplete dictionaries with coarse and refined grids are used in the two steps, respectively. The measurement matrix is constructed by using a very sparse projection scheme based on chaotic sequences because chaotic sequences have determinism and pseudo-randomness property. Such measurement matrix is mainly proposed for compressing the overcomplete dictionary in preestimation step, while it is well designed by choosing the steering vectors of true DOA in the accurate estimation step, in which the neighborhood information around the true DOAs partly solved in the previous step will be used. Monte Carlo simulation results demonstrate that the proposed algorithm can perform better than other existing single-snapshot DOA estimation methods. Especially, it can work well to solve the issues caused by closely spaced signals and single snapshot.