The integration of 5G and multi-access edge computing (MEC) technologies has opened new possibilities for enhancing operational efficiency and responsiveness in smart agriculture. However, determining the optimal deployment of the 5G MEC infrastructures in open-field agriculture environments poses considerable challenges due to strict latency, coverage, and energy constraints. This study proposes a tailored deployment framework for 5G-based smart farming systems that incorporate intelligent edge devices (IEDs), MEC nodes, and 5G small cells. A hybrid metaheuristic, combining the exploration capability of Simplified Harmony Search (SHS) with the local intensification of Variable Neighborhood Search (VNS), is designed to solve the resulting placement optimization problem. The approach is evaluated through a smart farm scenario featuring spatially distributed IEDs and diverse communication demands. Results show that the proposed SHSVNS approach achieves superior performance compared to existing approaches in reducing total deployment cost while maintaining coverage and responsiveness. This demonstrates its practical potential for enhancing agricultural digitalization through efficient infrastructure deployment.

Abstract Diamond multi-wire sawing machines are essential in semiconductor manufacturing, especially for slicing hard and brittle third-generation materials such as silicon carbide (SiC) and gallium nitride (GaN). The increased difficulty in processing these materials has highlighted the urgent need for reliable machine health monitoring and anomaly detection systems. While Predictive Maintenance and Prognostics and Health Management (PHM) frameworks have been widely applied across various industries, little research has specifically addressed semiconductor cutting equipment, where operational dynamics and data confidentiality present unique challenges. This study, in collaboration with an industry partner, develops two anomaly detection models tailored for diamond multi-wire sawing machines. The first model is a rule-based approach that utilizes sliding window techniques to extract statistical features and establish dynamic thresholds for anomaly detection. The second model employs a data-driven Univariate Autoencoder (UAE) to perform unsupervised anomaly detection by learning reconstruction errors from normal operating data. Both models are trained and validated using confidential industrial sensor datasets. Experimental results demonstrate that the UAE-based model achieves high detection accuracy with no observed false positives, providing an effective solution for enhancing operational reliability and production efficiency in semiconductor wafer slicing processes.

In this work, we introduce crowdsourcing home healthcare service (CHHS) systems, where caregivers (including nurses, personal care attendants, and housekeepers) from different locations (rather than centralized institutions) offer diverse home healthcare services to caretakers at home. Powered by cloud computing, the CHHS system enables real-time, dynamic, and large-scale matching between caretakers and caregivers based on their preferences and constraints, and determines caregivers’ rostering and routing plans, involving the NP-hard nurse rostering problem (NRP) and the vehicle routing problem (VRP). This work firstly creates a mathematical programming model to jointly roster and route for the CHHS, maximizing the matching scores of caregivers and caretakers based on the preferred features through analytic hierarchy process (AHP), and minimizing caregivers’ overtime and routing costs, under constraints of caregiver skills, regulations, and vehicle routing. The proposed matching score mechanism assigns weights to caretaker preferences, enhancing pairing with preferred caregivers and reducing dissatisfaction. This work proposes a hybrid genetic algorithm with variable neighborhood search (GAVNS), respectively tailored to handle the rostering and routing aspects of CHHS. Simulation indicates that the GAVNS lowers costs by approximately 38% and 26% in rural and city cases, respectively, and outperforms standalone GA and VNS, achieving a 3% additional cost reduction and consistently yielding feasible solutions.

It has been challenging to find the optimal recipe parameters for semiconductor processes to find a balance between budgets and computing efficiency. Therefore, this study focuses on finding the optimal recipe parameters of a semiconductor process using an improves meta Bayesian optimization (MetaBO) method, which can be trained with extremely few samples and historical data so as to quickly find the optimal process parameter combinations for the product. Experimental results show that the improved MetaBO significantly improves overall quality and efficiency in both model training and new task evaluation.