Edge deployments of OneM2M can suffer QoS degradation due to constrained resources and fluctuating IoT traffic, increasing latency and reducing availability under bursty workloads. We propose a lightweight framework that integrates an on-device LightGBM predictor with a proxy to classify oneM2M CSE states from CPU, RAM, and RTT and to trigger adaptive routing and cloud offloading when a critical state is predicted. Experiments on a Raspberry Pi 4 with HTTP/MQTT traffic show that LightGBM reaches 0.913 accuracy after ADASYN, and that proactive offloading restores QoS under overload by recovering the success rate to 100% and reducing RTT below 2 s.

The increasing complexity of IoT networks, particularly in critical domains such as e-health, intelligent transportation systems, and industrial environments, requires the implementation of adaptive routing strategies capable of ensuring high levels of quality of service (QoS). Although the Software-Defined Networking (SDN) paradigm offers centralized control and global network visibility, most existing controllers, such as Ryu, still rely on static routing mechanisms that are not well suited to the dynamic behavior of network traffic. In this context, we propose a dynamic QoS-oriented routing model for SD-IoT environments. This model assigns weights to network links based on several key metrics, including latency, link load, packet loss rate, and available bandwidth. It then uses the Dijkstra algorithm to compute the most suitable paths in real time. This approach allows the SDN controller to make context-aware routing decisions that account for both network topology and traffic characteristics. The model was evaluated through a series of scenarios that simulate three representative traffic types: e-health, transportation, and industrial. These scenarios involved progressively increasing load conditions. The results demonstrate significant improvements in performance, including reduced latency, lower packet loss, more balanced traffic distribution, and more efficient use of hardware resources at the switch level. These findings confirm the relevance and effectiveness of our approach and suggest promising perspectives for scalable, context-sensitive routing in SD-IoT networks.

The rapid proliferation of IoT devices and ecosystems creates significant challenges in managing increasing data traffic and service requests while maintaining system performance [1]– [3]. In oneM2M-based IoT systems, overloaded Common Service Entities (CSEs) can become bottlenecks, leading to resource saturation, higher latency, and request loss [4]. To address these challenges, this paper proposes IoTScal-2CoM-ALO, an adaptive load orchestration framework that introduces a two-level collaboration model (2CoM) enabling distributed CSEs to cooperate within and across domains. The framework incorporates an Adaptive Load Orchestration (ALO) mechanism that continuously monitors key performance indicators, including CPU utilization, memory consumption, round-trip time (RTT), and packet loss, to detect overload conditions and dynamically redirect traffic to suitable neighboring CSEs. The proposed approach is evaluated in a simulated distributed oneM2M environment under heterogeneous traffic conditions. Experimental results demonstrate significant performance improvements compared with non-collaborative and static collaboration approaches, achieving up to 73% reduction in memory consumption, RTT peak reductions of up to 4750 ms, and success rate improvements of approximately 4.8%. These results highlight the effectiveness of IoTScal-2CoM-ALO in improving resource utilization and maintaining service continuity in scalable IoT systems.

Recently, the Internet of Things (IoT) has expanded rapidly thanks to significant achievements in multiple fields. Each new device adds more data requests, demand, and network pressure [1]. Handling scale and QoS for IoT middleware platforms gets harder under high or irregular traffic loads. In most oneM2M-based middleware platforms, there are no external resource-usage mechanisms for overload conditions. Our new approach, IoTScal-CoM, presents a collaborative middleware architecture that enables QoS-based request redirection among independent oneM2M systems with the help of performance monitoring. In contrast to current techniques, the proposed solution employs only native oneM2M capabilities such as RTT, packet loss rate, CPU, and memory usage in order to guarantee the SLA conformity without changing the main standard specifications. The IoTScal-CoM middleware is deployed and tested in a simulated oneM2M environment by conducting a comparative analysis of both collaborative and non-collaborative scenarios. Experimental results demonstrate that the collaboration leads to increased stability and successful request processing as well as to improved system scalability.

IoT deployments under the OneM2M standard face a structural limitation: the platform provides robust interoperability but no built-in mechanism for autonomous overload management or QoS adaptation. This paper addresses that gap through an ML/DL decision framework that monitors QoS telemetry from Azure IoT devices, classifies system state into three classes (Keep Local, Partial Offload, Full Migration), and triggers adaptive offloading in under 700 ms. The framework combines a MAPE-K loop with seven classical classifiers, two ensemble strategies, and three deep learning architectures, evaluated on a 10,000-sample rebalanced dataset from a real Mobius CSE deployment. The Voting ensemble achieved 95

Maintaining a stable Quality of Service (QoS) in oneM2M deployments is challenging because edge-to-cloud traffic in IoT systems is bursty and resource demand changes rapidly. We propose a fuzzy-logic QoS controller, integrated into a MAPE-K autonomic loop, that adaptively decides the share of traffic offloaded from the local oneM2M platform to the cloud as a function of CPU usage, Round-Trip Time (RTT), and incoming traffic rate. The controller uses a 27-rule Mamdani inference engine, formally defined trapezoidal membership functions, and centroid defuzzification, and is integrated with the open-source Mobius platform. Compared with an unmanaged baseline under peak load, our approach reduces operating cost by 43.5%, RTT by 55.9%, and increases the request success rate by 19.4%, while keeping CPU and RAM usage in the 40–50% range. A qualitative comparison with static-threshold and recent fuzzy/learning-based offloading methods, together with a discussion of scalability to hundreds of edge nodes, positions the controller as a practical and cost-effective option for oneM2M-compliant IoT platforms.

The rapid expansion of Internet of Things (IoT) devices requires middleware capable of handling heterogeneous traffic while satisfying strict Quality of Service (QoS) targets. ITU-T Recommendation Y.1541 defines well-established performance thresholds for IP networks; however, baseline oneM2M deployments frequently fail to meet these targets under mixed workloads. This paper evaluates the open-source OM2M platform against ITU-T Y.1541 using eight QoS metrics spanning application and network layers, making the compliance gap explicit and quantifiable. Under the default configuration, the platform achieved only 20% overall compliance. To close this gap, an autonomic control architecture based on the Monitor-Analyze-Plan-Execute with Knowledge (MAPE-K) loop is integrated with a Random Forest (RF) classifier that predicts four discrete QoS operational states with 91.9% accuracy. The optimized configuration improves ITU-T compliance from 20% to 60%, achieving latency reductions of 53 to 71%, jitter mitigation of 93 to 97%, and transaction failure rate decreases of 36 to 64%, all measured during steady-state operation. The paper identifies the mechanisms responsible for the remaining non-compliant metrics and proposes a cross-layer roadmap for achieving full ITU-T compliance.

The technological world has known major advancements, leading to the creation of new paradigms like the Internet of Things (IoT). IoT is a combination between two concepts , the "Internet" and "Things" .This innovative notion connects billions of devices, enabling automation and smarter systems across industries like healthcare, smart cities, and manufacturing. By integrating concepts such as M2M, Big Data, and AI, IoT has transformed communication and innovation. However, as IoT networks expand, they face challenges like scalability, increased latency, limited bandwidth, and higher error rates, which impact Quality of Service (QoS).This paper proposes implementing Simple Automatic Repeat reQuest loss recovery mechanisms within IoT architecture to enhance QoS and address scalability challenges, ensuring reliable performance even as the network grows.

Ensuring excellent Quality of Service (QoS) while limiting financial expenses is seriously challenged by the growing scope and complexity of the Internet of Things (IoT). Though established under ETSI, the OneM2M standard provides a consistent middleware foundation but does not include a thorough QoS management technique. This work presents a dynamic optimization method based on the MAPE-K model to solve traffic- and resource-oriented QoS aspects. We balance QoS with operational costs by providing cost modeling—including cloud resource pricing and workload offloading. Supported by mathematical modeling and real-world workload situations, the results show the possibilities of cost-aware QoS techniques for scalable and efficient IoT systems based on the OneM2M standard.

The integration of IoT systems into diverse domains has increased the demand for efficient Quality of Service (QoS) optimization to handle the vast and dynamic data flows. In this work, we explore the potential of deep learning models to enhance QoS parameters such as latency, throughput, and resource utilization within the OneM2M standard. We specifically investigate AutoEncoders (AE), Long Short-Term Memory (LSTM) networks, and Convolutional Neural Networks (CNNs). Each model was designed and fine-tuned to address specific aspects of QoS optimization. Experimental results demonstrate the ability of these models to achieve high accuracy, efficient reconstruction, and improved classification performance on QoS metrics. This research presents a novel approach to leveraging deep learning models in IoT systems compliant with the OneM2M standard, facilitating intelligent, data-driven QoS management.

The Internet of Things (IoT) connects a vast number of smart devices, generating massive and diverse traffic, which poses challenges in network resource management and Quality of Service (QoS). Traditional network architectures, designed for predictable traffic, struggle to meet the increasing demands of modern IoT infrastructures. Software-Defined Networking (SDN) offers a solution by separating the control plane from the data plane, enabling centralized and flexible management of IoT traffic. This approach has led to the emergence of Software-Defined IoT (SD-IoT), which optimizes QoS based on TIPHON standard. In this study, we evaluate SD-IoT performance through a simulation in Mininet using the Ryu controller, integrating three types of IoT traffic (E-health, Transportation, Industrial) into advanced topologies such as NSFNET and ARPANET. Our tests analyze key metrics, including latency, jitter, throughput, packet loss rate and CPU/RAM consumption. The results demonstrate that applying SDN to IoT networks significantly improves QoS, aligning with TIPHON standard. This study paves the way for future research on QoS-aware routing optimization and artificial intelligence for IoT traffic management.

The exponential growth of Internet of Things (IoT) deployments has created unprecedented challenges in maintaining optimal Quality of Service (QoS) while managing dynamic resource allocation in IoT middleware platforms. Although OneM2M, standardized and adopted by ETSI (European Telecommunications Standards Institute), provides a robust framework for IoT interoperability, it does not inherently address overload scenarios and dynamic resource optimization challenges. This paper presents a comprehensive machine learning and deep learning framework for intelligent traffic offloading and QoS optimization within the OneM2M standard architecture. The primary objective is to optimize QoS metrics through a dual approach: first, traffic-oriented metrics including Round-Trip Time (RTT) and Success Rate, followed by resource-oriented metrics encompassing CPU and RAM utilization, enabling proactive system adaptation to prevent performance degradation. We developed an autonomous system based on the MAPE-K (Monitor, Analyze, Plan, Execute - Knowledge) framework that continuously monitors and collects critical QoS data from Azure IoT devices streaming real-time sensor data through Event Hubs to OneM2M Common Services Entity (CSE) servers, establishing performance thresholds for normal, acceptable, and critical operational states. The collected QoS data undergoes comprehensive preprocessing to address class imbalance using SMOTE oversampling techniques and feature engineering of composite metrics. Eight machine-learning algorithms were systematically evaluated, including Random Forest, LightGBM, XGBoost, and Support Vector Machines, followed by three deep learning approaches: Convolutional Neural Networks (CNN) for spatial pattern recognition, Long Short-Term Memory (LSTM) networks for temporal dependencies, and Variational Autoencoders (VAE) for feature compression. The optimized models are deployed through a scalable API infrastructure hosted on Hugging Face that orchestrates the entire decision-making process, serving as the central intelligence hub for receiving real-time QoS data, processing it through trained models, and providing autonomous decision recommendations for traffic management and resource allocation. The optimized Random Forest model achieved 96% accuracy with perfect precision for critical state detection, while ensemble approaches reached 98% accuracy, demonstrating RTT reduction of 30–50%, CPU/RAM usage optimization of 20–30%, and maintaining success rates above 90% across diverse traffic patterns (uniform, real-time, and burst scenarios), which proves particularly beneficial for critical IoT domains such as e-health applications where system reliability and responsiveness are paramount. This research fundamentally addresses critical limitations in current OneM2M implementations by introducing autonomous overload management capabilities, establishing a robust foundation for next-generation IoT infrastructure with promising extensions toward federated learning approaches for distributed model training across heterogeneous IoT networks, seamless integration with 5G/6G network slicing for enhanced QoS guarantees, and deployment in industrial IoT environments requiring ultra-low latency and high reliability standards.

The Internet of Things (IoT) has revolutionized multiple sectors by enabling seamless device connectivity and generating vast data streams. OneM2M, a global standard for IoT service platforms, provides interoperability across heterogeneous devices and applications. However, as IoT networks scale, ensuring quality of service (QoS) becomes a critical challenge. OneM2M, while efficient in managing device communication, struggles to maintain high performance under traffic congestion, impacting key QoS parameters such as latency (RTT), success rate, CPU, and RAM usage. This study evaluates the impact of traffic overload on OneM2M by introducing three major IoT communication protocols-HTTP, MQTT and CoAP-under real-world traffic injection scenarios. Performance analysis revealed that CoAP exhibits the best efficiency in terms of low RTT and minimal resource consumption, making it ideal for constrained IoT environments. Conversely, MQTT with QoS Level 2 ensures the highest reliability, which is crucial for mission-critical applications like e-health. HTTP, while widely supported, suffers from excessive overhead, limiting its scalability. To understand and predict system overload, we employ queueing theory modeling to simulate different traffic intensities and identify performance bottlenecks. Our model categorizes system behavior into three zones-preferable, acceptable, and critical-based on system utilization. This theoretical approach was validated through a real-world scenario, confirming the accuracy of our model in estimating QoS degradation under high traffic loads. To prevent system overload and optimize QoS, we propose an automated traffic orchestration approach based on the MAPE-K framework. This method dynamically manages IoT traffic by prioritizing data streams based on SLA requirements and leveraging cloud resources to offload excess traffic. Our findings demonstrate that cloud integration significantly enhances scalability while maintaining optimal QoS for high-priority traffic. Additionally, after integrating cloud-based traffic redirection, we conducted a cost estimation study to assess the economic feasibility of our approach. This ensures that our solution not only enhances system performance but also remains cost-effective for large-scale IoT deployments. These results provide valuable insights into protocol selection, traffic modeling, and cost-aware resource management in OneM2M-based IoT ecosystems, ensuring higher efficiency, scalability, and service reliability.

The rapid growth of the Internet of Things (IoT) generates heterogeneous traffic flows with diverse Quality of Service (QoS) requirements, necessitating efficient and adaptive routing mechanisms. The Software-Defined Networking (SDN) paradigm offers a centralized control plane conducive to optimizing these flows, but the choice of routing algorithm remains a key factor in maintaining performance. In this work, we compare the Dijkstra and Bellman-Ford algorithms in a realistic multi-flow Software-Defined IoT (SD-IoT) environment, implemented on the Ryu controller. The experimental architecture incorporates three representative traffic types: e-health, intelligent transportation, and industrial control, subjected to varying loads (50, 150, and 300 Mbps). Performance was evaluated across several metrics, including latency, packet loss rate, link load, and CPU/RAM consumption at both the controller and switch levels. The results show that both algorithms significantly improve traffic distribution and QoS stability compared to default routing. Dijkstra stands out with lower latency and faster convergence, making it suitable for time-sensitive applications, while Bellman-Ford, although more resource-intensive and These findings provide recommendations for selecting routing strategies adapted to heterogeneous SD-IoT environments and pave the way for future research into hybrid and multi-criteria approaches, including the integration of artificial intelligence for predictive traffic management and autonomous route adaptation.

The rapid growth of IoT applications has highlighted the critical importance of efficient Quality of Service (QoS) management in platforms like OneM2M. This paper presents an innovative machine learning-based approach, applied for the first time to the OneM2M standard, to optimize QoS parameters under high data traffic scenarios. This study examines the application of sophisticated data balancing techniques alongside ensemble learning methods to improve the performance of classification models for traffic management in IoT environments. Our approach begins by addressing the inherent data imbalance in IoT-generated traffic, leveraging synthetic over-sampling techniques such as SMOTE. This ensures that classification models are not biased towards dominant traffic classes. Additionally, we assess the effectiveness of various machine learning models, such as Random Forest, Support Vector Machines, XGBoost, and LightGBM, using both imbalanced and balanced datasets. To achieve superior accuracy and robustness, we implement ensemble learning strategies, including Voting and Stacking Classifiers, combining the strengths of Random Forest and LightGBM. These ensemble methods demonstrate significant improvements in prediction accuracy, achieving a weighted F1-score of 98%. Our results reveal that machine learning can effectively classify and prioritize IoT traffic, thereby optimizing critical QoS metrics such as latency, success rate, CPU, and RAM utilization. By applying these models within OneM2M, we demonstrate their potential to intelligently manage traffic congestion, filter non-IoT traffic, and autonomously migrate excessive load to the cloud. This study sets a foundation for deploying ML-driven QoS management in real-world IoT systems and highlights its transformative potential for ensuring reliable, efficient, and scalable IoT services.

The Internet of Things (IoT) is undergoing rapid evolution, promising to connect billions of devices in the near future. This technological expansion presents numerous challenges, especially in maintaining diverse Quality of Service (QoS) standards across various applications. Our research introduces an innovative approach to evaluate the performance of the OneM2M platform’s middleware, crucial for the effective management of the expanding IoT landscape. We developed scenarios that simulate heavy load conditions on the platform, utilizing both uniform HTTP traffic and realistic IoT communication patterns. These scenarios aim to rigorously test the platform’s ability to cope with high network loads and genuine IoT operational conditions. Our approach is designed to provide a comprehensive assessment of the OneM2M platform's capability in handling different traffic types and volumes, an essential factor for its deployment in real-world IoT contexts. The results from these evaluations are critical in identifying the strengths and shortcomings of the OneM2M platform under various stress scenarios. These insights are valuable for enhancing the platform’s scalability, efficiency, and adaptability, ensuring that it can maintain robust QoS in the face of complex and dynamic IoT network demands.

The increasing variety of domains within the IoT ecosystem demands precise SLAs with adaptable QoS to accommodate the rapid proliferation of connected devices. While the OneM2M standard provides a comprehensive framework, it has not sufficiently addressed QoS issues. To tackle these challenges, we propose an approach tailored to the OneM2M standard, which considers both MQTT and HTTP traffic types and targets three specific SLAs: eHealth, industrial, and transportation. Our approach involves simulating overload scenarios and deploying automated mechanisms based on the MAPE-K framework to alleviate congestion. Additionally, we integrate cloud resources to prevent request wastage and maintain optimal QoS metrics. Through these strategies, we aim to ensure that the system can dynamically adapt to varying traffic conditions and SLA requirements, thereby enhancing overall performance and reliability.

The Internet of Things (IoT) involves diverse devices with various communication needs. While HTTP is commonly used for its simplicity, it may not always be the best choice for all IoT applications. This paper provides for the first time a new process to implement MQTT and CoAP protocols within the OneM2M platform to enhance interoperability and efficiency. MQTT offers lightweight messaging for bandwidth and power efficiency, while CoAP is optimized for constrained devices. This paper details the integration process, enabling researchers and developers to utilize these protocols within OneM2M. By facilitating testing and comparative analysis, we aim to provide insights into the platform’s performance, contributing to the development of more efficient and interoperable IoT solutions.

The exponential growth of IoT devices presents significant challenges for maintaining Quality of Service (QoS) and scalability in OneM2M networks. This paper presents a comprehensive comparison of three fundamental loss recovery mechanisms—Selective Repeat, Go-Back-N, and Stop-and-Wait—implemented across both HTTP and MQTT protocols within the OneM2M architecture. Through extensive experimentation using a controlled test environment with varying packet loss rates, we evaluate the performance impact of each mechanism on transmission rates and latency. Our results demonstrate that MQTT consistently outperforms HTTP across all three recovery mechanisms, particularly in high packet loss scenarios. The performance analysis reveals significant differences in both transmission rates and latency between the protocols, with MQTT showing superior scalability and reliability. These findings provide crucial insights for selecting appropriate loss recovery mechanisms in OneM2M deployments, particularly for scenarios requiring high reliability and low latency in IoT networks.

The Internet of Things (IoT) paradigm aims to connect billions of devices, enabling wise and dynamic choice making across packages. However, the rapid expansion of IoT poses good-sized demanding situations for retaining quality of provider (QoS) requirements. This paper provides a detailed performance analysis and comparative analysis of Hypertext Transfer Protocol (HTTP) and Message Queuing Telemetry Transport (MQTT) protocol in oneM2M system; focusing on QoS aspects. HTTP, known as its simplicity and widespread use, facilitate client-server communication in many IoT applications. In contrast, MQTT, a lightweight messaging protocol, is optimized for low-bandwidth and high-latency environments, making it ideal for constrained devices and unwieldy networks reliability. We developed and implemented test scenarios incorporating homogeneous traffic and real-time data to assess the platform’s robust capabilities. Simulating realistic heavy load conditions and IoT network models, this study provides a thorough understanding of the strengths and limitations of the oneM2M platform in handling different traffic types and volumes The findings provide important insights into the scalability of the oneM2M platform, efficiency, and adaptability, which are important to ensure robust QoS in diverse IoT environments.