DATA CONGESTION HANDLING IN SENSORS ENABLED INTERNET OF THINGS

Abstract

The Internet of Things (IoT) is a network of physical devices connected to the Internet, enabling them to interact with their internal states or the external environment. This transformative technology has diverse applications in various domains, such as healthcare monitoring, transportation, and smart cities. The growth of IoT networks led to a significant increase in data traffic. However, this surge in data, coupled with the limitations of constrained IoT devices, has created a bottleneck in the network, resulting in congestion. Congestion poses several issues, particularly in terms of packet delivery. The overwhelming data traffic strains the network infrastructure, causing delays and hindering the timely delivery of packets. Additionally, the limited storage capacity of IoT devices exacerbates the problem, leading toa substantial number of packet losses. This congestion problem hampers the efficiency and throughput of IoT networks. It disrupts the smooth flow of data and jeopardizes the integrity of the entire network. Addressing congestion in IoT networks is crucial to ensure seamless communication and optimal performance. Therefore, it becomes imperative to develop effective strategies and solutions to mitigate congestion, improve data traffic management, and enhance the overall performance of IoT networks. By doing so, we enabled the successful delivery of packets, reduce packet losses, and ensure the smooth operation of IoT applications and services. This thesis aimed to investigate the existing literature on congestion control in IoT networks and identify the research gap. Specifically, we examined the focus of most authors, which primarily revolved around congestion control without adequately determining its occurrence in the IoT network. While packet loss and delay were commonly used indicators of congestion, congestion problems could also be influenced by factors such as link failure and channel noise, making them less reliable. Therefore, we proposed more accurate schemes to detect and control congestion in IoT networks by considering a broader set of parameters in the prediction process. Additionally, we addressed the limitations of applying traditional IP- . vi based congestion control approaches to resource-constrained IoT environments. We emphasized the importance of incorporating congestion prediction approaches to effectively manage congestion before it impacted network performance. Furthermore, we recognized the significance of accounting for the limited resources of IoT devices, the heterogeneous nature of IoT networks, and the dynamic changes in network conditions when designing congestion control techniques. By filling this research gap, we aimed to contribute to the development of robust and efficient congestion control mechanisms for IoT networks that were focused on resource control schemes where we offloaded data packets or routed the packets in a congestion-aware manner. To address the above research gaps, we defined three primary objectives: The first objective was to design an approach for predicting congestion in IoT networks by considering multiple parameters. We utilized a Deep Neural Network-Restricted Boltzmann Machine (DNN-RBM) model to detect data congestion. The input to the Deep Neural Network (DNN) included performance factors such as congestion window, throughput, propagation delay, Round Trip Time (RTT), number of packets sent, and packet loss. The Restricted Boltzmann Machine (RBM) was employed to optimize the weights of the proposed DNN RBM model. The second objective focused on designing an approach for congestion control by implementing data offloading techniques. Data offloading techniques involved transferring data from one network or device to another to relieve congestion or improve performance. Rather than requiring the child node to select a new parent node, our approach identified a suitable neighbour node capable of sharing the load and assisting the congested node. The approach involved two steps: identifying the congested node and selecting the appropriate neighbour node to carry the data packets. The third objective entailed designing an approach for congestion-aware data transmission in IoT networks. We employed an improved Analytic Hierarchy Process (AHP) method to select the most suitable node based on multiple parameters such as the distance, hop count, residual energy, link quality, and . vii buffer occupancy. This approach enabled efficient hop-to-hop data communication/ transmission while considering congestion levels and network efficiency. By addressing these objectives, our research aimed to enhance congestion prediction accuracy, optimize congestion control through intelligent data offloading, and improve overall data transmission efficiency in IoT networks while accounting for congestion awareness.