To solve the time-consuming, laborious, and inefficient problems of traditional methods using classical optimization algorithms combined with electromagnetic simulation software to design antennas, an efficient design method of the multi-objective antenna is proposed based on the multi-strategy improved sparrow search algorithm (MISSA) to optimize a BP neural network. Three strategies, namely Bernoulli chaotic mapping, inertial weights, and t-distribution, are introduced into the sparrow search algorithm to improve its convergent speed and accuracy. Using the Bernoulli chaotic map to process the population of sparrows to enhance its population richness, the weight is introduced into the updated position of the sparrow to improve its search ability. The adaptive t-distribution is used to interfere and mutate some individual sparrows to make the algorithm reach the optimal solution more quickly. The initial parameters of the BP neural network were optimized using the improved sparrow search algorithm to obtain the optimized MISSA-BP antenna surrogate model. This model is combined with multi-objective particle swarm optimization (MOPSO) to solve the design problem of the multi-objective antenna and verified by a triple-frequency antenna. The simulated results show that this method can predict the performance of the antennas more accurately and can also design the multi-objective antenna that meets the requirements. The practicality of the method is further verified by producing a real antenna.

Convolutional Neural Networks (CNNs) and Transformer networks have shown impressive performance in hyperspectral image (HSI) classification. However, these models usually concentrate on examining either local or global representations of HSI data, frequently falling short in capturing multidimensional representations. Furthermore, these methods fail to fully leverage the strengths of CNNs and Transformers. This paper presents the Adaptive Coupling Transformer Network (ACTN), a parallel-hybrid network aiming to improve representation learning for HSI classification. ACTN has the ability to capture different types of representation and facilitate mutual learning. Specifically, we introduce a parallel-hybrid module called the Adaptive Coupling Module (ACM), which is designed to capture multi-faceted representations from the HSI cube. The ACM consists of two branches: a CNN branch that extracts local contextual representations and a Transformer branch that captures global dependency representations. Our proposal is an Adaptive Response Fusion Module (ARFM) that interacts with the hybrid module to merge local and global representations at different resolutions in an adaptive way. In addition, we utilize a cosine similarity function to restrict the loss function in mutual learning, guaranteeing the preservation of both local and global representations to the maximum extent. Extensive experiments conducted on three public HSI datasets demonstrate that ACTN outperforms state-of-the-art methods based on Transformers and CNNs.

In this article, the joint design for user scheduling and analog beam selection in a downlink multiuser millimeter-wave (mmWave) system is studied. Our objective is to maximize the achievable sum rate under the user scheduling constraint, the analog beam selection constraint, and the resource capacity constraint. This problem is nonconvex and NP hard. We first propose a whale optimization algorithm (WOA)-based scheme to obtain a near-global-optimal solution with fast convergence and low complexity. Since the joint optimization user scheduling and beam selection problem is a constrained integer programming problem, the binary version of WOA is applied to deal with integer variables and the penalty method is used to handle the constraints. Besides, a nonlinear convergence factor is introduced to enhance the optimal solutions. For real-time use, we also propose a low-complexity machine-learning (ML)-based scheme. In the ML-based scheme, we decompose the original optimization problem into two subproblems: 1) user classification subproblem and 2) the analog beam selection subproblem. The user classification subproblem is solved based on the $k$ -means algorithm, where the users are clustered according to channel correlation. To solve the analog beam selection subproblem, we reformulate this subproblem as a multiclass classification problem. Considering the imbalance nature of the data set of the subproblem, we train the multiclass classifiers via the biased-SVM algorithm. Finally, the simulation results of the WOA-based scheme and the proposed ML-based scheme against the state-of-the-art schemes have shown the advantages of our proposed schemes.

This article studies a network where a transmitter communicates with a receiver by hybrid communications that consist of passive information transmission (IT) via backscatter communication (BC) and active IT via wireless-powered communication (WPC). Because the circuit energy consumption in the passive IT of BC is much lower than that in the active IT of WPC, BC usually achieves a higher data transmission rate than WPC. Thus, it was suggested in the literature that the network throughput performance could not be improved by hybrid communications. However, our work in this article demonstrates that the throughput can be enhanced by a newly designed hybrid communication strategy. To demonstrate this, we develop a novel protocol that enables the transmitter to adaptively switch its operation between BC, active IT, and energy harvesting in one time block while scheduling energy consumption flexibly among multiple time blocks. Under the developed protocol, we formulate an optimization problem to jointly optimize the operation mode and resource allocation at the transmitter. The formulated problem is difficult to solve because the energy scheduling at the transmitter is coupled across multiple time blocks, and noncausal channel state information (CSI) is required. To address this problem, we first solve a simplified optimization problem via dynamic programming (DP) and a layered optimization method by assuming that the noncausal CSI is known. Then, we employ an approximate DP approach to solve the original problem with causal CSI. Finally, we verify by simulations that the proposed scheme can achieve superior throughput performance.

In this article, we study a previously overlooked work of designing the constellation for concentration shift keying (CSK) in diffusion-based molecular communication (MC) systems, considering the adverse effects of signal-dependent noise and intersymbol interference (ISI). Against this background, the conventionally used uniform constellation strategy is no longer the optimal choice for CSK. Considering the discrete nature of messenger molecules, the constellation design of CSK is essentially a nonlinear integer programming problem. However, due to the nonconvexity of the goal function, general convex optimization methods cannot be easily employed. To solve the aforementioned problem, a novel CSK strategy with a brute-force search algorithm is proposed to obtain the optimal constellation numerically. Moreover, a low-complexity search algorithm that provides new insight into reducing the complexity of MC systems is further presented. Specifically, we theoretically derive the optimal threshold and employ a maximum likelihood detector that effectively minimizes the average symbol error rate (SER) by leveraging the knowledge of ISI at the receiver. Finally, we can obtain a fitted equation for the nearly optimal constellation points under specific channel parameters, attaining the SER performance similar to the optimal one, while the search complexity is much lowered.

Deep metric learning (DML) has shown promising results in few-shot hyperspectral image (HSI) classification. The core idea of DML is to learn a generalized metric space in which pixels from unseen classes can be effectively classified with only a few labeled samples. However, existing DML methods mainly adopt traditional Euclidean distance to achieve the feature metric, which ignores the category uncertainty of spatial-spectral features in mixed and edge pixels. To address this issue, we fully exploit fuzzy logic theory and propose a deep fuzzy metric learning (DFML) method for few-shot HSI classification. First, we design a novel hybrid CNN-Transformer spatial-spectral feature extraction network to fully capture the spatial-spectral features of HSI pixels. Then, a fuzzy set representation method based on Gaussian membership function for spatial-spectral features is proposed, which describes the inherent fuzziness of the spatial-spectral features. Finally, to perform the fuzzy similarity measure between the fuzzy sets of query samples and prototypes, we construct a spatial-spectral fuzzy metric space, in which HSI pixels with category uncertainty in their features can be better classified under condition of small-scale labeled samples. Extensive experimental results on three public HSI datasets demonstrate that the proposed DFML method outperforms state-of-the-art few-shot HSI classification methods.