An Intelligent Adaptive Neuro-Fuzzy for Solving the Multipath Congestion in Internet of Things

Abstract

The Internet of Things (IoT) has recently become a significant focus in research circles. IoT facilitates the integration of numerous physical entities with the Internet. Adhering to a standardized structure is imperative to manage the vast amount of information effectively. Although many researchers in the field of IoT have proposed various layered architectural designs, none have yet fulfilled all the requisite architectural criteria. Network congestion occurs when the volume of data packet traffic surpasses the network's handling capacity. Apart from addressing congestion issues, it is crucial to harmonize network resources like energy, bandwidth, and latency. The Quality of Service (QoS) in IoT applications chiefly depends on proficient congestion management, which is the central subject of this research. The research employs the Adaptive Neuro-Fuzzy Inference System (ANFIS) to regulate congestion, while the Membership Function (MF) undergoes adjustments through the application of the Modified Squirrel Search Algorithm (MSSA). This ANFIS amalgamates the advantages of Fuzzy Logic (FL) and Artificial Neural Networks (ANN) to form a unique framework. Utilizing ANFIS, adaptive analysis services are available to interpret complex patterns and nonlinear interactions, featuring quick learning capabilities. The MSSA aids in tweaking the Membership Function within the ANFIS model, achieving a successful global convergence rate. An adaptive method considering predator presence probability is employed to harmonize the algorithm's exploration and exploitation functionalities, further bolstered by a dimensional search approach. The simulation results demonstrate that the proposed Swarm Intelligence Adaptive Neuro-Fuzzy Inference System (SI-ANFIS) method significantly reduced traffic overhead and attained an impressive accuracy rate of 93.58%.