This study presents a compact multi-band fork-shaped multiple input multiple output (MIMO) antenna design that integrates Carrel's log-periodic concept with the planar log-periodic array (PLPA) mechanism to improve C, Ku, K, Ka, and Q band performance. The suggested MIMO antenna full size is 54 × 44 × 1.58 mm³. The study empirically shows improved isolation and high gain of wideband MIMO antenna, addressing the requirements for modern wireless communication. The suggested self-multi-band coupling counteraction, characterized by a straightforward antenna design and extensive self-decoupling capabilities, is expected to offer considerable application value for compact and wideband MIMO systems. The observed data supports the assertion that the proposed MIMO antenna is designed to operate across a variety of frequency bands, including the C-band (4.4-5.21 GHz), Ku-band (14.30-17.64 GHz), K-band (20.10-28.10 GHz), Ka-band (28.95-30.76 GHz) and (32.38-36.42 GHz), and a segment of the Q-band (38.34-39.55 GHz). The proposed MIMO antenna's wideband performance is achieved by applying the Carrel log-periodic theory, which prescribes the geometric scaling of radiating elements. Moreover, it was found that a microstrip feed line with a diameter of 50 Ω was noteworthy in conjunction with the carrel concept. The microstrip feed line is connected to two orthogonal branches. Decoupling structures eliminate the necessity for the MIMO antenna to achieve its maximum gain of 7.21 to 9.53 dBi and exceptional element isolation of 21.54 dB. MIMO performance measurements, encompassing envelope correlation coefficient (ECC), diversity gain (DG), and channel capacity loss (CCL), demonstrate favorable characteristics and field correlation performance over the operational spectrum band. The design suggested suits Ku/K-band and multi-band MIMO applications across diverse wireless systems in communication.

The integration of artificial intelligence (AI) and the Internet of Things (IoT) has given rise to AI of Things (AIoT) ecosystems, which, while revolutionizing various domains, face challenges in energy efficiency and environmental sustainability. This article introduces LSCEA-AIoT, a low-carbon sustainable computing framework for energy-efficient acquisition and allocation in AIoT. The framework addresses two critical aspects: 1) energy-efficient heterogeneous data acquisition and 2) low-carbon task offloading optimization. We develop a multisource model for data acquisition that considers acquisition time, load balancing, and energy consumption, coupled with an adaptive sensor node deployment strategy for optimized channel allocation. The task offloading component formulates a joint optimization problem, balancing local and edge computing models to minimize ecosystem costs and carbon emissions. We propose a carbon-aware multichannel exploration offloading decision algorithm based on a Monte Carlo tree search to obtain near-optimal solutions. Extensive experiments compare LSCEA-AIoT with state-of-the-art methods across various metrics. Results demonstrate that LSCEA-AIoT significantly outperforms existing approaches, achieving lower data acquisition errors, reduced energy consumption, extended network lifetimes, and increased data acquisition volumes. LSCEA-AIoT exhibits superior performance in task offloading scenarios in normalized rewards, system costs, and adaptability to diverse network configurations. These findings validate LSCEA-AIoT’s effectiveness in achieving low-carbon, sustainable computing for AIoT ecosystems.

In the competitive market of wind power forecasting, ensuring data security while enhancing forecasting accuracy is crucial. Federated learning emerges as a key solution to these challenges, offering a way to preserve data privacy and security across multiple entities. This paper presents a groundbreaking approach that leverages Federated Transfer Learning (FTL) in conjunction with Additive Attention Temporal Neural Network architectures, focusing on the competitive wind power forecasting. Our method integrates the sophisticated capabilities of Gated Residual Networks with a strategic process of feature selection, tailored for the unique demands of secure and efficient power forecasting. Through experiments involving data from various geographical regions, this research uncovers the specific challenges of applying federated learning in a competitive landscape. The resulting FTL-Net model demonstrates exceptional performance, surpassing existing benchmarks with metrics such as a Mean Absolute Error (MAE) of 6.011, Mean Squared Error (MSE) of 13,855.45, Root Mean Squared Error (RMSE) of 11.7, a coefficient of determination (\({R}^{2}\)) of 0.9899, and a correlation coefficient of 0.998. These results not only confirm the FTL-Net model's efficacy but also emphasize the significance of meticulous feature selection in federated learning contexts. Compared to other models, FTL-Net's superior performance, bolstered by feature selection, establishes it as a leading approach in the secure and competitive realm of wind power forecasting. This research underscores the potential of combining advanced neural network architectures with federated learning for future advancements in the field, highlighting the importance of data security and collaborative learning in achieving high accuracy in power forecasting.

With the explosive growth of consumer electronic devices, edge computing has emerged as a promising paradigm to process large-scale data in real-time and enhance data privacy. However, consumer electronic devices’ limited computing power and energy pose significant challenges to efficiently executing computation-intensive tasks. To tackle this problem, we present a cloud-edge collaborative computing offloading method that relies on deep reinforcement learning. More precisely, we construct an optimization problem with the goal of minimizing the combined delay in task execution and energy consumption, taking into account the computing resources, bandwidth, and offloading policies. We then develop an asynchronous cloud-edge collaborative deep reinforcement learning (CEC-DRL) algorithm to solve the optimization problem. The CEC-DRL algorithm leverages the computing capabilities of both cloud and consumer electronic devices to satisfy the demand for efficient data processing in large-scale consumer electronics scenarios. Moreover, it can adaptively adjust the offloading policy to minimize the system cost under various and dynamic environments. Simulation results demonstrate that the CEC-DRL algorithm provides fast convergence, high resilience, and almost ideal offloading policies while incurring the lowest computation cost.

The integration of wearable healthcare devices and fog computing has emerged as a promising paradigm for providing personalized and efficient healthcare solutions. However, the security and privacy concerns associated with the sensitive health data transmitted in such systems pose significant challenges. This paper proposes a secure and efficient fog-assisted wearable healthcare Internet of Things (IoT) system, which leverages physical layer security techniques and quantum-inspired resource allocation to enhance the confidentiality and performance of the system. First, we present a system model for a secure fog-assisted wearable healthcare IoT system, incorporating wearable healthcare devices, fog computing servers, and potential eavesdroppers. Next, we formulate an optimization problem with the goal of maximizing the secrecy rate of the system by allocating both spectral and computing resources jointly. Ultimately, we develop a bean optimization method influenced by quantum principles to address the issue of allocating resources collaboratively. The experimental findings show the efficacy of the suggested system and resource allocation strategy regarding secrecy rate performance under various scenarios.

Tool condition monitoring technology is an indispensable part of intelligent manufacturing. Most current research focuses on complex signal processing techniques or advanced deep learning algorithms to improve prediction performance without fully leveraging the end-to-end advantages of deep learning. The challenge lies in transforming multi-sensor raw data into input data suitable for direct model feeding, all while minimizing data scale and preserving sufficient temporal interpretation of tool wear. However, there is no clear reference standard for this so far. In light of this, this paper innovatively explores the processing methods that transform raw data into input data for deep learning models, a process known as an input paradigm. This paper introduces three new input paradigms: the downsampling paradigm, the periodic paradigm, and the subsequence paradigm. Then an improved hybrid model that combines a convolutional neural network (CNN) and bidirectional long short-term memory (BiLSTM) was employed to validate the model’s performance. The subsequence paradigm demonstrated considerable superiority in prediction results based on the PHM2010 dataset, as the newly generated time series maintained the integrity of the raw data. Further investigation revealed that, with 120 subsequences and the temporal indicator being the maximum value, the model’s mean absolute error (MAE) and root mean square error (RMSE) were the lowest after threefold cross-validation, outperforming several classical and contemporary methods. The methods explored in this paper provide references for designing input data for deep learning models, helping to enhance the end-to-end potential of deep learning models, and promoting the industrial deployment and practical application of tool condition monitoring systems.