Gokaraju Rangaraju Institute of Engineering and Technology is a private engineering college in Hyderabad, Telangana, India.
Accurate air pollution prediction needs systems that monitor diverse environmental data, adapt to changing pollution patterns, and ensure secure real-time communication. Existing systems struggle to capture distributed sensor interactions and dynamic environmental variations effectively. To address this problem, the study proposes an Adaptive and Efficient Air Pollution Monitoring Network (AEPM-Net) incorporating Metal Oxide Semiconductor (MOS) nanosensors with an average particle size ranging from 20 to 40 nm for sensitive air-quality monitoring. The system monitors levels of Carbon Monoxide CO , Nitrogen Dioxide NO_2 , Sulfur Dioxide SO_2 , Nitric Oxide NO , Volatile Organic Compound VOC , and Particulate Matter PM , which includes fine particulate matter PM2.5 and coarse particulate matter PM10 , in different locations. The system uses SPECK encryption and additive Lightweight Homomorphic Encryption (LHE) to protect collected pollutant data, while noise filtering and modified Z-score normalization methods help to stabilize the data. The Spatio-Temporal Data Processing Module (ST-DPM) analyzes nanosensor data across time and space by capturing correlations among distributed sensors and environmental variations. The Hierarchical Attention Mechanism (HAM) prioritizes important spatio-temporal features to improve prediction accuracy. The Dynamic Adaptation Module (DAM) adjusts parameters in real time, while the Pollution Level Prediction Module (PLPM) forecasts pollutant levels with uncertainty handling. Over 90 days, the model achieved R² values of 0.973 CO , 0.974 NO , 0.975 NO_2 , 0.978 SO_2 , 0.971 VOC , 0.969 PM2.5 , and 0.967 PM10 . The proposed framework supports environmental sustainability by enabling early detection of hazardous pollutants, improving urban air quality monitoring, and assisting effective pollution control to protect public health.
The traditional approach to calculating anisotropy via interrupted tensile deformation does not fully capture the accurate Lankford parameters, thereby questioning the accuracy of the yield criterion. Various anisotropic yield criteria consider only the initial yield values and use them to calculate the associated anisotropic parameters. In this work, the uniaxial tensile tests were conducted at 0°, 45°, and 90° with respect to the rolling direction. The 2D DIC method is employed to capture the strain fields in the gauge region. The uniform elongation region is predicted using the Swift hardening model. The evolving Lankford parameters were predicted with a 4th-order polynomial function and subsequently implemented in Hill’48 anisotropic yield criterion. The strain-dependent Hill’s 48 criterion is incorporated due to its fewer anisotropic parameters and reduced testing requirements for fully capturing the evolving yield locus. The results indicate that the modified criteria precisely capture the anisotropic behaviour of the AZ31B alloy.
The rapid integration of large language models (LLMs) into enterprise systems has introduced new concerns around the exposure of personally identifiable information (PII) during multi-agent interactions. Existing frameworks like CAPRI attempt to mitigate such risks through contextual pseudonymization; however, these methods lack formal privacy guarantees and remain susceptible to inference attacks, particularly in structured reasoning workflows. To address this limitation, we propose CAPRI-DP, an enhanced framework that extends CAPRI with differentially private noise injection applied to sensitive fields within structured entity representations. The approach preserves semantic integrity by applying Laplace perturbations selectively, allowing external LLM agents to operate on masked data without sacrificing task accuracy. Our experiments, conducted across finance and healthcare use cases using ToolEmu-inspired tasks, reveal that CAPRI-DP improves upon fully pseudonymized setups by achieving a 52% success rate—ten percentage points higher—while maintaining fewer conversational turns. Although slightly less performant than entity-only configurations, the proposed method offers a tangible balance between privacy assurance and reasoning effectiveness. These findings suggest that CAPRI-DP can serve as a viable model for privacy-conscious, real-time LLM-agent deployments in sensitive operational domains.
This research focuses on Martensitic stainless steel (MSS) in service conditions where there is a significant amount of demandingness. Some of the possible applications include turbine blades, valves, cutting tools, and various components used in automobiles. Among the possible heat treatment processes for Martensitic stainless steel, tempering is identified as a critical process for modulating the physical properties of the material. The main aim of the research is to evaluate the impact of temperature and time on tempering process and microstructural changes in Martensitic stainless steel. The research also evaluates the impact of temperature on the degradation of Martensitic stainless steel and its improvement through an increase in toughness, ductility. The research also evaluates the impact of temperature on Martensitic stainless steel and its relationship to stiffness. The research discusses the wear mechanism of Martensitic stainless steel and its enhancement towards tempering process.
The rapid proliferation of Internet of Things (IoT) devices has intensified the need for intelligent and adaptive security mechanisms capable of addressing emerging cyber threats such as multimedia data manipulation and zero-day attacks. This paper presents a novel AI-driven intrusion detection framework that synergistically integrates Generative Adversarial Networks (GANs), Deep Q Networks (DQNs), and Federated Learning (FL) to achieve decentralized, privacy-preserving, and real-time threat detection. Unlike existing GAN-based or reinforcement learning (RL)-based models that operate under centralized architectures, the proposed system introduces a hierarchical Edge-Fog-Cloud (EFC) design to minimize latency, enhance scalability, and support distributed intelligence across IoT environments. GANs generate synthetic attack data to augment scarce training samples, while DQNs adaptively learn optimal defensive strategies against dynamic and evolving threats. Federated Learning ensures that sensitive IoT data remain local by transmitting only encrypted model updates, thereby strengthening privacy protection. Experimental results demonstrate that the proposed framework achieves a 93
