The rapid expansion of autonomous and AI-driven networked systems across smart cities, transportation, healthcare, and cyber–physical infrastructures has intensified concerns related to trust, safety, transparency, and regulatory compliance. Existing approaches often address governance, ethics, and technical assurance in isolation, leaving a gap between high-level principles and practical system implementation. To address this challenge, this study proposes a comprehensive trustworthy AI governance framework that integrates system-theoretic modeling with governance constrained decision-making. The framework models autonomous systems as distributed cyber–physical–social systems and embeds ethical, safety, and legal constraints directly into the learning and optimization process using constrained Markov decision processes and Lagrangian optimization. A composite trustworthiness metric is formulated by aggregating Fairness, Robustness, Privacy, Explainability, Security, and Accountability dimensions. The methodology combines analytical modeling with simulated datasets representing multi-agent autonomous networks. Experimental results demonstrate a composite trust score of approximately 0.813, explainability stability of 0.905, and near-zero governance violations, confirming improved compliance, reduced risk, and enhanced trust compared to unguided AI systems.

Social media platforms now sit at the center of everyday communication for individuals and organizations. That visibility and reach also make them attractive to attackers. As Artificial Intelligence (AI) becomes embedded in social media ecosystems through content creation tools, recommendation systems, automated interaction, and moderation, well-known threats such as phishing and impersonation are being reshaped rather than replaced. This article reviews double-masked peer-reviewed research on cybersecurity risks tied to social media use, with emphasis on how AI-enabled techniques amplify deception, accelerate malicious content spread, and complicate detection. A thematic analysis synthesizes empirical findings across social engineering, account compromise, malicious link distribution, data leakage, and human factors. The literature indicates that social media security problems are socio-technical: attacks succeed through a mix of platform features, organizational workflows, and user judgment under pressure. The article closes with stakeholder-specific implications for platform developers, organizational security teams, and policymakers, and identifies research gaps that remain unresolved.

Enterprise deployments of Large Language Models need database connectivity solutions that work under absolute security and performance pressures. This paper describes an architecture we built for Model Context Protocol servers in Kubernetes, refined through actual production use. Three problems dominated our design work: authentication, network security, and scaling capacity. We went with OAuth2 authentication paired with connection pooling. Once deployed, this setup achieved a 96% cache hit rate while reducing latency by 71% relative to our baseline measurements. Horizontal scaling keeps the system responsive during traffic surges. Our production environment ran this architecture continuously for six months, processing roughly 45,000 queries per day. We saw 99.9% uptime throughout, and 95th-percentile latency came in at 180 Ms. The numbers show this architecture can manage what security-focused enterprises need, even in heavily regulated sectors. For engineers and architects working on AI-powered database services, this paper provides concrete lessons particularly relevant when security, reliability, and performance constraints are non-negotiable.

This paper talks about how Data Ops concepts can be utilized in creating survey questions like HEDIS. The Data Ops principles, like Structured Data Pipeline, Automation, and Governance, are applied with the clinical rigor of HEDIS measures. Since DataOps can be applied to any business process or analytic solution that involves extracting, a time-consuming process like creating and collecting HEDIS data will have much scope for automation.

Autonomous mobile robots have transformed industries by enabling operations without human intervention. Fuzzy logic controllers (FLCs) are widely used in robots for path planning due to their capability to handle uncertainties. However, large and complex fuzzy rule sets often result in computational inefficiencies, leading to increased navigational time. This study proposes a similarity-based rule reduction method to optimize fuzzy inference systems. The approach leverages a similarity threshold algorithm to identify and merge redundant rules using a similarity index. The hypothesis suggests that simplifying the rule set will either sustain or improve the robot’s task completion time. The study is validated through simulations of an FLC-based path planning nonholonomic wheeled mobile robot navigating in a static, unknown environment. The robot detects obstacles, localizes itself, and maps its surroundings in real-time to achieve effective navigation. The primary inputs are obstacle distances, and the output determines wheel velocity. MATLAB and CoppeliaSim are used to optimize and evaluate the robot’s performance, with traversal time as the primary metric. The results show that cutting down the number of fuzzy logic rules shortened the robot’s traversal time and boosted its processing speed and overall performance. By simplifying the rule base by nearly half (48.15%), the robot was able to complete its navigation tasks more quickly and efficiently. This demonstrates a practical way to tackle the long-standing issue of fuzzy rule complexity, making decision making systems more adaptive and responsive. These results are particularly significant for real-time applications where speed and dependability are crucial, like robotics, driverless cars, and predictive maintenance.

This study explores how IP Multimedia Systems (IMS) applications can be optimized and scaled effectively within telecom networks that rely on cloud infrastructure. IMS supports a wide range of services, including Voice over IP (VoIP), Video over IP, and Rich Communication Services (RCS), and utilizes standardized components such as P-CSCF, S-CSCF, I-CSCF, TAS, RCS, and MRF in accordance with 3GPP specifications. A unified method for gathering diverse operational data has been proposed, covering performance indicators from the network, infrastructure usage, application-level behavior, and surrounding environmental factors. From this pool of data, key indicators—like call success and failure rates, scam call detection, and voicemail activity—are extracted and analyzed. To make sense of these trends, several machine learning models, including Random Cut Forest (RCF), XGBoost, and a basic K-Nearest Neighbors (KNN) approach, are used. Their predictive strength is measured using common statistical tools such as R-squared (R²), Mean Squared Error (MSE), and Root Mean Squared Error (RMSE). This evaluation helps determine which model is most suitable for anticipating specific challenges such as SIP errors, call drops, latency issues, and network congestion. Based on these predictions, the system can automatically fine-tune its settings to adapt to changing network conditions. By integrating with CI/CD pipelines, these adjustments can happen in near real-time. The end result is a responsive and cost-effective framework for managing IMS resources in cloud-based telecom environments.

Generative AI, Model Context Protocol (MCP), and Applied Machine Learning (ML) combine to provide a compelling means to power advanced agentic AI systems that are able to make decisions autonomously and to learn how to do better over time. This paper explores how these technologies can be integrated to improve the performance, flexibility, and intelligence of agentic AI systems. With generative AI, businesses can generate new data and, therefore, new insights, but MCP is what ensures these AI models are contextually aware and can function in different environments or conditions. Applied ML adds a new level of realism to model training and real-world deployment, proving that AI can learn, adapt, and make intelligent decisions on the fly. This paper presents the synergy of these building blocks and proposes a unified framework that combines their strength to develop intelligent context-aware agents that work independently in dynamic, complex environments. This integrated framework is widely applicable in domains like healthcare, finance, and robotics, where adapting decision-making abilities is important.

Autonomous vehicle perception systems require robust pedestrian detection, particularly on geometrically complex roadways like Type-S curved surfaces, where standard RGB camera-based methods face limitations. This paper introduces YOLO-APD, a novel deep learning architecture enhancing the YOLOv8 framework specifically for this challenge. YOLO-APD integrates several key architectural modifications: a parameter-free SimAM attention mechanism, computationally efficient C3Ghost modules, a novel SimSPPF module for enhanced multi-scale feature pooling, the Mish activation function for improved optimization, and an Intelligent Gather & Distribute (IGD) module for superior feature fusion in the network's neck. The concept of leveraging vehicle steering dynamics for adaptive region-of-interest processing is also presented. Comprehensive evaluations on a custom CARLA dataset simulating complex scenarios demonstrate that YOLO-APD achieves state-of-the-art detection accuracy, reaching 77.7% mAP@0.5:0.95 and exceptional pedestrian recall exceeding 96%, significantly outperforming baseline models, including YOLOv8. Furthermore, it maintains real-time processing capabilities at 100 FPS, showcasing a superior balance between accuracy and efficiency. Ablation studies validate the synergistic contribution of each integrated component. Evaluation on the KITTI dataset confirms the architecture's potential while highlighting the need for domain adaptation. This research advances the development of highly accurate, efficient, and adaptable perception systems based on cost-effective sensors, contributing to enhanced safety and reliability for autonomous navigation in challenging, less-structured driving environments.

Mobile devices, particularly smartphones, have become integral to daily life. The personal and sensitive information stored on these devices has become hot targets for cybercriminals. Crimes such as mobile banking fraud, online scams and payment fraud are taking new forms daily. Mobile phones are also being used for money laundering operations to transfer or exchange illicit funds through mobile payment platforms. Either way, the involvement of mobile phones in committing crimes or mobile devices being targets of attacks has made them a critical source of digital evidence in forensic investigations. To extract evidence from these devices, mobile forensic tools have become essential for law enforcement agencies to preserve, extract and acceptably analyze data from mobile devices in law courts. An in-depth examination of the current state-of-the-art mobile forensic tools is essential to help the investigators make an informed choice while selecting the tools. This paper explores four mobile forensic tools: Cellebrite UFED, XRY, Oxygen Forensics Detective, and Magnet AXIOM. A comprehensive overview of the tools is presented, focusing on their capabilities, technicalities and shortcomings in the ever evolving mobile ecosystem. Key functionalities such as accessing locked devices, deleted data recovery, app data parsing, cloud storage access, decryption of encrypted data, and timeline generation are discussed to assess the effectiveness of the tools. Even though these tools have become sophisticated over time, they face several technical challenges. No single tool provides a complete solution for mobile forensic investigation. It is recommended that a combination of tools, based on the case-specific requirements, device types and the nature of data extraction, be used. This paper underscores the importance of constant updates, a wide range of device support, and legal compliance to perform mobile forensic tools effectively.