Artificial Intelligence, Computer Vision and Pattern Recognition
63
Scopus Publications
Scopus Publications
Digital Twin-Based Framework for Predictive Modeling of Infrastructure in IoT-Enabled Smart Cities R Udayakumar, P Balamurugan, Ashu Nayak, A Haja Alaudeen, Bahromjon Urmanov Icfnds 2025 2025 the 9th International Conference on Future Networks and Distributed Systems, 2026 Innovative, customizable data collection and intelligent decision-making have driven rapid advances in IoT technology, and, in turn, robust, complex smart urban infrastructure is envisioned to achieve the goals of smart cities. Nevertheless, the ability to assess and maintain 'real' urban infrastructure systems remains a challenge. This paper introduces a Digital Twin (DT) model as a framework for targeted, 'predictive' infrastructure systems in IoT smart cities. This approach combines the IoT smart city infrastructure predictive urban systems Digital Twin framework for predictive urban systems managed for analytics, and data-driven machine learning to bridge the technology gap in 'virtual' smart cities infrastructure systems managed to provide a 'digital' urban infrastructure smart systems to provide predictive analytics for ongoing real-time maintenance and optimization of fault prediction and maintenance scheduling of infrastructure smart systems. Operational decision systems, predictive decision systems, and targeted and informed decision systems in urban systems. A constructed urban transportation network for managed, predictive urban infrastructure systems for smart cities, as a developed, targeted system to assess and determine system fault, predictive maintenance, and optimization opportunities, with closed-loop technology for maintenance. This promotes the sustainability of smart cities.
Dynamic Resource Allocation for Software-Defined Networks using Predictive Analytics in Multi-Cloud Deployments Roohee Khan, Ashu Nayak Proceedings of 6th International Conference on Expert Clouds and Applications Icoeca 2026, 2026 Multi-cloud scenarios are on the rise and becoming more complex, which is driving the need to find new ways of managing and allocating resources on these multi-cloud solutions. SDN (software-defined networking) has emerged as a valuable technology for improving the efficiency and flexibility of network management in a multi-cloud environment. This paper provides a predictive analytics-based dynamic resource allocation model for SDN within a multi-cloud deployment. The main aim of this model is to make the most efficient use of resources and improve services by anticipating how network traffic will behave (In terms of percentage increase) and the content needed by applying machine learning techniques and time series forecasts to accurately forecast resource requirements and make real-time adjustments for allocating and reallocating resources across multiple clouds. The model will dynamically allocate resources per cloud, therefore improving network performance, eliminating resource wastage, and reducing Latency. Further, this system assesses bandwidth usage, response times, and resource use to ensure optimal allocation. Results from experiments conducted in a simulated multi-cloud environment show that the proposed model is significantly more efficient, scalable, and adaptable than traditional static resource allocation techniques. This study presents a comprehensive solution to the Obsolete Way of Operating Such Environments.
AI-Powered Microservices Orchestration for Efficient Cloud Resource Management and Service Scaling Priya Vij, Ashu Nayak Proceedings of 6th International Conference on Expert Clouds and Applications Icoeca 2026, 2026 Microservice architecture's wide implementation has helped to enhance modularity and scalability of cloud applications greatly. Still, efficient resource management and service scaling are challenging due to highly dynamic, unpredictable workloads. Traditional threshold-based autoscaling systems tend to respond slowly to workload changes, leading to overprovisioning, high latency, and failure to meet service-level objectives (SLOs). This paper proposes an AI-based microservices orchestration system for the smart management of cloud resources and service scalability. The framework combines nonstop monitoring, machine-learning-based workload forecasting, and AI-based scaling guidelines into a closed-loop, cloud-native orchestration process. The proposed system can dynamically adjust service replicas and resource allocations in real time by actively predicting demand and optimizing scaling decisions. The experimental analysis of a containerized cloud testbed shows that the proposed solution is far superior to conventional autoscaling methods. The outcomes indicate that average resource utilization has improved by about 30-35%, response latency has been reduced by 29%, and the rate of SLO violations has been reduced by 69%. Moreover, the AI-based orchestrator reduces scaling behavior by approximately 45% and scaling decision time by 46%, leading to a stable, more efficient system under variable workloads. The overall system's productivity also increases over time, reaching up to 40% efficiency during peak demand. These results support the claim that AI-based microservices orchestration offers a scalable, responsive, and cost-effective solution for next-generation cloud resource management.
Integrative Architectures for Cognitive and Agentbased Intelligence Through Neural Reasoning and Autonomous Coordination Priya Vij, Ashu Nayak Proceedings of 2nd International Conference on Multi Agent Systems for Collaborative Intelligence Icmsci 2026, 2026 The combination of cognitive architectures with agent-based systems has made possible the creation of autonomous intelligent systems that are able to perform better decision-making and problem-solving. In this paper, a hybrid architecture that uses neural reasoning to enhance the performance of cognitive agents in dynamic and uncertain environments is introduced. This research aims to study how neural networks can be integrated into multi-agent systems, including the decision-making process, scalability of agent coordination, and reliability of real-time agent systems. A new paradigm is presented to provide self-managed flexibility and enhance resolutions in robotics, autonomous cars, and humanmachine interactions. The results of the experiment indicate that the framework is more effective than traditional cognitive systems in the aspects of agility, extensiveness, and operational effectiveness. In particular, the proposed model decreased the time to complete the tasks by 29 %, the efficiency of the coordination by 18 %, and the flexibility to adapt to the changes in the environment by 22 %, as opposed to the legacy systems. These findings indicate that the autonomous systems have been greatly improved in terms of performance and scalability. The next line of research will utilize the ways in which neural reasoning can be further combined with agent-based coordination to handle more complex problems in real-life situations.
Hybrid Neural and Agent Ecosystems for Multimodal Learning and Adaptive Intelligence in Complex Environments Roohee Khan, Ashu Nayak Proceedings of 2nd International Conference on Multi Agent Systems for Collaborative Intelligence Icmsci 2026, 2026 The paper will offer a hybrid neural-agent ecosystem that is expected to improve adaptive intelligence and multimodal learning in dynamic and complicated environments. The main aim of this study is to combine deep neural networks with agent-based systems, which incorporate emotional intelligence and swarm intelligence to provide real-time decision-making, task execution, and cooperation between agents on decentralized systems. Applications to the system include resource allocation, predictive maintenance, and dynamic optimization, which are compared to its performance when compared to the traditional agent-based and neural network models. The most important results indicate that the hybrid system is superior in various parameters as compared to the traditional models. The time spent on completing the task is decreased by <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{2 8 \%}$</tex> (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{1 8. 2}$</tex> seconds for the hybrid system and <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{2 5. 4}$</tex> seconds for traditional agent-based systems), and the efficiency of collaboration has also grown by 15 % (9.5 and 7, respectively). Also, the flexibility of the system is observed as the difference in the time of executing the tasks is minimal (standard deviation of 1.2 seconds through datasets), which indicates that the system is stable and robust in practice. These improvements are statistically significant (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{p}<0.05$</tex>) as checked with statistical analysis using ANOVA. This study has shown that hybrid neural-agent ecosystems have the potential to be applied in realworld robotics, health, and autonomous systems. The work in the future will be aimed at streamlining the system to operate on resource-constrained systems, further exploiting the emotional intelligence capacity in large-scale multi-agent systems, and studying the efficiency of the system in other complex real-world settings. Scalability and deployment issues of real-time applications are also major fields that can be significantly developed further.
Data-Driven Decision Support in Smart Ubiquitous Agriculture Zaed Balasm, Dilnavoz Shavkidinova, Dr. Deepa Rajesh, Dr.N. Prabakaran, Islom Kadirov, Ashu Nayak Journal of Wireless Mobile Networks Ubiquitous Computing and Dependable Applications, 2025
Energy-Aware Buildings Reconfigure Internal Systems to Reduce Peak Demand Ramy Riad Hussein, Ashu Nayak, R. Venkatasubramanian, Mahmudov Kahramon Shuhratjon Ugli, D. Aarthi, Thella Preethi Priyanka, K. U. Khamraev, Arnav Jain Proceedings of 2025 International Conference on Intelligent Systems and Pioneering Innovations in Robotics and Electric Mobility Transforming Mobility and Automation Through Intelligent Engineering Inspire 2025, 2025
