Hybrid Deep Learning Model for Fault Diagnosis in Smart Manufacturing using Edge AI and Digital Twin Integration Lendale Venkateswarlu, Ch Ram Mohan, P. Karunakar Reddy, Vasavi Bande, M Archana, Pavan Kumar Reddy Manellore Proceedings of the 6th International Conference on Smart Electronics and Communication Icosec 2025, 2026 The rapid advancement of Industry 4.0 has accelerated the need for intelligent and autonomous fault diagnosis in smart manufacturing systems. This paper presents a hybrid deep learning approach that combines Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) networks to effectively capture both spatial and temporal features from industrial sensor data. To support real-time fault detection and diagnosis, the model is deployed via Edge AI, thereby reducing latency, conserving bandwidth, and improving operational responsiveness. In addition, a Digital Twin of the manufacturing process is developed, offering a synchronized virtual representation of the physical system. This enables continuous monitoring, simulation, and predictive analytics, improving contextual awareness and fault explainability. Experimental evaluation on publicly available benchmark datasets demonstrates that the proposed model outperforms traditional approaches in terms of accuracy, noise robustness, and response time. The integration of deep learning, edge computing, and digital twin technologies provides a scalable and intelligent framework for predictive maintenance and real-time monitoring in next-generation smart factories.
Next-Gen Cognitive IoT Architectures for Autonomous Vehicular Networks in Smart Cities Ch Ram Mohan, Mallareddy Adudhodla, Sirikonda Vamshi Krushna, Vasavi Bande, Ch G V N Prasad, Pavan Kumar Reddy Manellore International Conference on Nexgen Networks and Cybernetics Ic2nc 2025 Proceedings, 2026 Rapid urbanization and digital transformation have driven the need for integrating intelligent transportation systems with robust IoT infrastructures. Autonomous vehicular communication in smart cities demands scalable, low-latency, and secure networks for real-time decision-making. This study presents a cognitive IoT architecture optimized for autonomous vehicular ecosystems, combining edge computing, capsule attention networks for congestion control, and ambient RF based communication to improve connectivity and energy efficiency. The modular design integrates sensor fusion, digital twin orchestration, and multi-protocol gateways, tested in both simulated and real-world urban environments. Results show an average latency of 235 ms, a packet delivery ratio of 99.4%, and a 12% reduction in energy consumption compared to the conventional systems. The security framework mitigated 94% of simulated injection attacks without latency penalties, while edge processing reduced cloud traffic by 28%. The proposed approach offers a scalable, energy-efficient, and secure blueprint for urban ITS, benefiting municipal planners and automotive OEMs.
A Machine Learning Framework for Accurate and Scalable Brain Tumor Categorization in MRI Imaging D Marlene Grace Verghese, Ch Ram Mohan, Banoth Ramesh, Balvindersingh Bondili, S Vara Vinod Ungarala, David Raju Kuppala Proceedings of 5th International Conference on Soft Computing for Security Applications Icscsa 2025, 2025 For a precise diagnosis and treatment plan to be planned in a clinical setting, brain tumour categorization from magnetic resonance imaging (MRI) data is essential. Clinicians can efficiently monitor the course of the disease, select the best course of treatment, and gauge the effectiveness of that course of treatment with the help of fast and accurate tumour classification. Additionally, machine learning models can help radiologists make better diagnoses by lowering interpretation errors and enhancing the quality of MRI image interpretation. Classifying brain tumours also aids in research endeavours to discover biomarkers, comprehend the biology of tumours, and create tailored treatments for various tumour subtypes. Current techniques for classifying tumours in the brain from MRI data frequently rely on labour-intensive, inconsistent manual segmentation and feature extraction. These techniques might not be able to accurately classify tumours due to minute variations in their morphology or texture. Furthermore, the accessibility and scalability of traditional procedures in clinical settings may be limited due to the need for expertise in radiology and medical imaging. Furthermore, manual feature engineering could miss significant tumour traits or not fully utilize MRI data for categorization. To overcome the shortcomings of current approaches, the suggested system makes use of machine learning techniques to improve and automate the classification of brain tumours using MRI image data. In order to extract discriminative features directly from MRI scans, this work uses machine learning algorithms. The proposed models are capable of reliably classifying brain cancers into important categories and effectively differentiating between different types of tumours by training them on large-scale MRI datasets labelled with tumour labels.
Optimizing Cloud Performance and Energy Efficiency using Proximal Policy Optimization in Reinforcement Learning Techniques T Kavitha, Godavarthi Sesha Rudra Kalyani, T Harikrishna, Mangamma Dharavatu, Ram Mohan, K V S S R Murthy Proceedings of the 6th International Conference on Electronics and Sustainable Communication Systems Icesc 2025, 2025 The complexity and dynamism of current cloud data centers requires smart and flexible allocation mechanisms capable of achieving the performance, costefficiency and energy optimization objectives. In this paper, the concept of reinforcement learning (RL) techniques is applied to do the same in the context of creating an adaptive mechanism of managing resources in a cloud computing environment efficiently. In the research, performance of four RL algorithms (QLearning, Deep Q-Network (DQN), Policy Gradient, Proximal Policy Optimization (PPO)) is studied and compared in terms of critical factors: resource utilization, SLA violation rate, energy consumption, and response time and execution time with different workload conditions. A controlled environment simulation was carried out on Cloud Sim/iFogSim extensions to Pythonbased RL libraries. All the considered algorithms exceeded in performance the rest of them as PPO indicated better adaptability, a low number of SLA violations, greater resource utilization persistence, and showed the smallest consumption of energy. The experiment confirms the usefulness of actor, critic-based PPO in cloud that is dynamically driven and a scaleable resource management framework of RL-based practical application in cloud orchestration platforms in real-time applications.
NavigAId: A Deep Learning Framework for Real-Time Traffic Sign Interpretation with Multi-Sensor Fusion Rammohan Ch, Qubeb S.MP, Nithya Darsini P.S., Saikiran N., Kishore Dasari, Kadiyala Vijaya Kumar Ssrg International Journal of Electronics and Communication Engineering, 2024 The rapid advancement of autonomous vehicle technologies has necessitated the development of more efficient and accurate systems for real-time traffic sign interpretation. Traditional approaches predominantly rely on single-sensor data, which often suffer from limitations in accuracy and robustness under varying environmental conditions. This paper presents the NavigAId system, an advanced autonomous navigation framework leveraging a deep neural fusion model that integrates Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs) to interpret complex sensory data for enhanced navigational decision-making. Through a comprehensive empirical evaluation conducted across a variety of environmental conditions and traffic scenarios. The NavigAId system demonstrated superior performance, notably achieving accuracy rates exceeding 95% in clear weather conditions, both urban and highway, and maintaining robust performance in adverse weather and nighttime conditions with accuracy rates above 89%. The fusion model exhibited significant improvements over standalone CNN or RNN models, with accuracy enhancements ranging from 3.0% to 8.0% and precision improvements up to 8.6%, depending on the scenario. Particularly, in velocity prediction tasks, the system achieved a remarkable reduction in Mean Squared Error (MSE) by up to 33.3% compared to individual neural network models.
The BioShield Algorithm: Pioneering Real-Time Adaptive Security in IoT Networks through Nature-Inspired Machine Learning Rammohan Ch, Laxmikanth P, Doddi Srikar, Ayyappa Chakravarthi, Terrance Frederick Fernandez, Hussain Basha P Ssrg International Journal of Electrical and Electronics Engineering, 2024 This paper introduces the BioShield Algorithm, aimed at the crucial task of securing IoT networks through real-time adaptive mechanisms that draw inspiration from nature. It delves into the critical issues plaguing IoT security, such as the dynamic and heterogeneous nature of both threats and network architectures. It proposes a nature-inspired machine learning model designed for adaptive, real-time threat detection and mitigation. By employing the "UNSW-NB15" dataset, the algorithm undergoes a rigorous evaluation across various metrics, including detection accuracy, response time, and scalability. The quantitative analysis reveals the algorithm's high proficiency in dealing with diverse cyber-attack scenarios, with precision scores ranging from 95.9% for Malware to 98.4% for Tampering attacks. Recall rates also show impressive figures, peaking at 96% for DDoS attacks, alongside consistently high F1 scores that underscore the model's balanced precision and recall capabilities. Additionally, accuracy rates across different attack types further confirm the algorithm's effectiveness, with scores oscillating between 94.95% and 97.2%. These results strongly endorse the BioShield Algorithm's capacity to accurately detect and classify cyber threats within IoT environments, spotlighting its applicability in significantly enhancing the security framework of IoT networks. This algorithm stands out for its adaptive, efficient, and scalable nature, positioning it as a pivotal contribution to the field of IoT security.
An Approach for Coordinating Lane Changes between Autonomous Vehicles in Congested Areas International Journal of Intelligent Systems and Applications in Engineering, 2023
Reputation-based secure routing protocol in mobile ad-hoc network using Jaya Cuckoo optimization Ch. Ram Mohan, Venugopal Reddy Ananthula International Journal of Modeling Simulation and Scientific Computing, 2019 The advancements in Mobile Ad-hoc Network (MANET) are suitable to wide applications, which involve military applications, civilian domains, and disaster recovery systems. It is assumed that the nodes present in the routing protocols of MANETs are cooperative and trustworthy. This assumption makes MANET more prone to manipulation and interception, which generates several possibilities of different suspicious attacks, such as Denial of Service (DoS) attacks. Hence, security is considered as an important parameter as the network is fond of suspicious attacks. This paper addresses the security issues by proposing an optimization algorithm, named Jaya Cuckoo Search (JCS) algorithm, which is the combination of Jaya algorithm and Cuckoo Search (CS) algorithm for initiating secure route among the MANET nodes so that the path attained is feasible and secure. The proposed JCS algorithm uses a fitness function, which considers a multiobjective function using parameters, such as distance, link lifetime, delay, energy, trust, along with the reputation factor to select a secured path. The JCS algorithm shows maximum performance with energy, throughput, and PDR values as 27.826[Formula: see text]J, 0.554, and 0.628, respectively.
