In multi-chip IGBT power modules, package-level failures typically manifest first as localized distortions in the copper baseplate temperature field before evolving into catastrophic faults. However, existing Power Internet of Things (Power IoT) monitoring approaches mainly rely on single-point indicators, such as on-state voltage drop or case temperature, making them inadequate for capturing spatially nonuniform thermal anomalies under sparse sensing. This paper proposes a terminal–edge–cloud collaborative method for parallel multi-module temperature-field reconstruction in Power IoT, which maps sparse thermocouple measurements to a high-resolution temperature field while meeting real-time constraints. At the terminal layer, the thermocouple array layout is optimized using a condition-number minimization criterion to maximize system observability. At the edge layer, a physics-constrained conditional generative adversarial network (PC-cGAN) is deployed on an MPSoC platform, and a multi-objective particle swarm optimization algorithm is used to automatically search network architectures and hyperparameters under resource constraints, thereby balancing reconstruction accuracy and inference latency. Experimental results demonstrate that a single edge node can monitor 4–6 power modules in parallel, with an inference latency of 11.8 ms for four-module parallel processing, achieving a 3.1× efficiency improvement over serial architectures. Using 12 sparse measurements, the proposed method achieves a root mean square error of 1.89°C, a mean absolute error of 1.52°C, and a hotspot deviation of 2.34°C; compared with CNN baselines, these metrics improve by 29.5%, 29.3%, and 32.2%, respectively, while reducing communication bandwidth by 98% relative to cloud-centric approaches. The proposed method provides an engineering-ready solution for scalable, real-time thermal-state monitoring of power-module arrays in converter systems.

A novel curvature-compensated bandgap reference (BGR) using a four-input current feedback amplifier is presented in this paper. High-order curvature compensation terms (T•ln(T) dependence) are produced by the VBE’s difference of bipolar transistors (BJT) whose currents are biased at proportional to absolute temperature (PTAT) and complementary to absolute temperature (CTAT) currents, which can diminish the temperature-dependent nonlinearity of traditional BGR. A four-input current feedback summation amplifier innovatively incorporates an additional input pair to complete the summation of the first-order and high-order curvature compensation terms. Combined with an auto-zero technique built on a new timing constraint, the proposed BGR greatly enhances the BGR’s accuracy and suppresses the low-frequency noise. Implemented in a standard 180nm CMOS process, the proposed BGR achieves the average temperature coefficient (TC) of 3.5 ppm/°C and the worst case is 5.08 ppm/°C from -40 °C to 125 °C with merely one-point 4-bit trimming. Measurement shows a 3σ initial inaccuracy of ±0.14% without trimming, and ±0.05% is achieved with trimming. The low-frequency noise (0.01Hz to 10Hz) is 6.8 µVRMS.

In the context of massive machine-type communications (mMTC) within the Internet of Things (IoT), joint activity detection and channel estimation (JADCE) is a key challenge in enabling massive access due to sporadic device access patterns. In this work, we consider both single-antenna and multiple-antenna base station scenarios and formulate the JADCE problem using the least absolute shrinkage and selection operator (LASSO) framework. To address this problem, we propose a model-driven network that utilizes compressive sensing (CS) and deep learning techniques. Specifically, we first design a prediction-correction alternating direction method of multipliers (PC-ADMM) as the underlying algorithm of the model-driven network. Then, the network is developed based on the PC-ADMM and is designed to be complex-valued. Furthermore, we also model the proximal operator, typically used to generate sparse solutions in LASSO, as a channel attention module within the model-driven network to enhance robustness. Numerical results show that the proposed PC-ADMM framework outperforms existing LASSO-based methods in terms of channel estimation and device activity detection.

Abstract Within the federated learning (FL) framework, the client collaboratively trains the model in coordination with a central server, while the training data can be kept locally on the client. Thus, the FL framework mitigates the privacy disclosure and costs related to conventional centralized machine learning. Nevertheless, current surveys indicate that FL still has problems in terms of communication efficiency and privacy risks. In this paper, to solve these problems, we develop an FL framework with communication-efficient and privacy-preserving (FLCP). To realize the FLCP, we design a novel compression algorithm with efficient communication, namely, adaptive weight compression FedAvg (AWC-FedAvg). On the basis of the non-independent and identically distributed (non-IID) and unbalanced data distribution in FL, a specific compression rate is provided for each client, and homomorphic encryption (HE) and differential privacy (DP) are integrated to provide demonstrable privacy protection and maintain the desirability of the model. Therefore, our proposed FLCP smoothly balances communication efficiency and privacy risks, and we prove its security against “honest-but-curious” servers and extreme collusion under the defined threat model. We evaluate the scheme by comparing it with state-of-the-art results on the MNIST and CIFAR-10 datasets. The results show that the FLCP performs better in terms of training efficiency and model accuracy than the baseline method.

The accelerated development of the Industrial Internet of Things (IIoT) is catalyzing the digitalization of industrial production to achieve Industry 4.0. In this article, we propose a novel digital twin (DT) empowered IIoT (DTEI) architecture, in which DTs capture the properties of industrial devices for real-time processing and intelligent decision making. To alleviate data transmission burden and privacy leakage, we aim to optimize federated learning (FL) to construct the DTEI model. Specifically, to cope with the heterogeneity of IIoT devices, we develop the DTEI-assisted deep reinforcement learning method for the selection process of IIoT devices in FL, especially for selecting IIoT devices with high utility values. Furthermore, we propose an asynchronous FL scheme to address the discrete effects caused by heterogeneous IIoT devices. Experimental results show that our proposed scheme features faster convergence and higher training accuracy compared to the benchmark.

Grant-free random access is an effective solution to enable massive access for future Internet of Vehicles (IoV) scenarios based on massive machine-type communication (mMTC). Considering the uplink transmission of grant-free based vehicular networks, vehicular devices sporadically access the base station, the joint active device detection (ADD) and channel estimation (CE) problem can be addressed by compressive sensing (CS) recovery algorithms due to the sparsity of transmitted signals. However, traditional CS-based algorithms present high complexity and low recovery accuracy. In this manuscript, we propose a novel alternating direction method of multipliers (ADMM) algorithm with low complexity to solve this problem by minimizing the$\ell _{2,1}$norm. Furthermore, we design a deep unfolded network with learnable parameters based on the proposed ADMM, which can simultaneously improve convergence rate and recovery accuracy. The experimental results demonstrate that the proposed unfolded network performs better performance than other traditional algorithms in terms of ADD and CE.

This brief proposes a wideband interdigital bandpass filter (IBPF), exploiting the through-silicon via (TSV)-based 3-D integrated circuit (3-D IC) technology. IBPF is a structure composed of multiple groups of parallel-coupled line resonators, which can increase the bandwidth and out-band suppression (OBS) characteristics of a bandpass filter. Through the transformation of the Chebyshev low-pass circuit model, a seventh-order IBPF centered at 85 GHz is obtained. By combining TSV technology with the coupling coefficient method, the design process from the Chebyshev low-pass circuit model to the final optimization of IBPF is given. The fractional bandwidth (FBW) of IBPF based on TSV is 63%, the out-of-band suppression is greater than 80 dB, the insertion loss is 1.3 dB, the return loss is 20 dB, and the size of the compact IBPF is only$0.50\times0.34$mm2($0.48\times 0.33\,\,\lambda \text{g}^{2}$).