Self-Configuration and Self-Healing Framework Using Extreme Gradient Boosting (XGBoost) Classifier for IoT-WSN

Abstract

In most Internet of Things (IoT) systems, Quality of service (QoS) must be confirmed with respect to the requirement of implementation domain. The dynamic nature of the IoT surroundings shapes it to complicate the fulfilment of these commitments. A wide range of unpredictable events endanger the quality of service. While execution the self-adaptive schemes handle with system’s unpredictable. In IoT-based Wireless Sensor Networks (WSNs), the significant self-management objectives are self-configuration (SC) and self-healing (SH). In this paper, Self-Configuration and Self-healing Framework using an extreme gradient boosting (XGBoost) Classifier are proposed. In this framework, the IoT traffic classes are categorized as several types under XGBoost classifier. In SC phase, the IoT devices are self-configured by allocating various transmission slots, contention access period (CAPs) on the basis of its categories with priorities. In SH phase, the source node cardinally establishes a confined route retrieval method if the residual power in-between node is truncated or the node has displaced far away. The proposed framework is executed in NS-2 and the results exhibit that the proposed framework has higher packet delivery ratio with reduced packet drops and computational cost. Therefore, the proposed approach has attained 24.7%, 28.9%, 12.75% higher PDR, and 16.8%, 19.87%, and 13.7% higher residual energy than the existing methods like Self-Healing and Seamless Connectivity using Kalman Filter among IoT Networks (SH-SC-KF-IoT), Provenance aware run-time verification mechanism for self-healing IoT (PA-RVM-SH-IoT), and Fully Anonymous Routing Protocol and Self-healing Capacity in Unbalanced Sensor Networks (FARP-SC-USN) methods, respectively.