The inclusion of various intelligent electronic devices such as smart meters for AMI is expected to result in intermittent or frequent communications network congestion if additional network infrastructure investments are not made. One approach to deal with such a data volume challenge in smart grids without additional investments to increase network capacity is to aggregate data streams within the network whenever network congestion happens, but this needs to be done carefully so as not to significantly impact the applications that need the data. This paper proposes a novel approach to manage AMI data traffic volume in multi-level data collection trees through data aggregation that estimates the expected network delays messages would suffer and dynamically determines an aggregation policy to be applied at forwarding nodes of the tree to reduce the network delays with which the electric utility can get the necessary information. The proposed algorithm is evaluated for different network congestion scenarios using the NS-3 simulator. The results illustrate that the algorithm is immensely effective in controlling the increase in network latencies as congestion levels increase in AMI networks. In addition, it does well in satisfying quality-of-service (QoS) requirements in terms of the data granularity required by smart grid applications.

Human Activity Recognition (HAR) using Inertial Measurement Unit (IMU) sensors is critical for applications in healthcare, safety, and industrial production. However, variations in activity patterns, device types, and sensor placements create distribution gaps across datasets, reducing the performance of HAR models. To address this, we propose LanHAR, a novel system that leverages Large Language Models (LLMs) to generate semantic interpretations of sensor readings and activity labels for cross-dataset HAR. This approach not only mitigates cross-dataset heterogeneity but also enhances the recognition of new activities. LanHAR employs an iterative re-generation method to produce high-quality semantic interpretations with LLMs and a two-stage training framework that bridges the semantic interpretations of sensor readings and activity labels. This ultimately leads to a lightweight sensor encoder suitable for mobile deployment, enabling any sensor reading to be mapped into the semantic interpretation space. Experiments on five public datasets demonstrate that our approach significantly outperforms state-of-the-art methods in both cross-dataset HAR and new activity recognition. The source code is publicly available at https://github.com/DASHLab/LanHAR.