Design of an iterative congestion control model for sensor networks via ensemble classification with bioinspired optimizations

Abstract

This paper presents a novel and all-encompassing strategy for addressing the challenges of communication congestion and time synchronization in dense wireless sensor networks. Existing machine learning and deep learning models that provide bio-inspired and pre-emptive packet-analysis solutions for these problems frequently suffer from high complexity and implementation costs, as well as a lack of scalability for various node types and traffic scenarios. We propose a congestion-aware routing model with time synchronization capabilities to overcome these limitations. Our model uses a combination of Naive Bayes (NB), k Nearest Neighbour (kNN), Multilayer Perceptron (MLP), and Support Vector Machine (SVM) classifiers. These classifiers are trained using transformed components of temporal clock states and packet delivery performance data, resulting in an improved determination of optimal clock deviations and routing path ways. In addition, we incorporate a Bacterial Foraging Optimization (BFO) Model, a meta-heuristic optimization algorithm inspired by the foraging behaviour of bacteria. Utilizing a temporal fitness function that considers throughput, communication latency (or Round-Trip Time), energy levels, and packet delivery performance, the BFO Model facilitates the identification of congestion-aware routing paths. This novel approach optimizes routing paths and outperforms existing routing algorithms, particularly in environments with a high density of nodes and heterogeneous network topologies. Our model's scalability, which allows it to adapt to different node types and traffic conditions, is a significant advantage. This feature increases its applicability for real-time wireless sensor network deployments in a variety of applications. Our proposed congestion-aware routing model with time synchronization capabilities represents a substantial advancement in the field of wireless sensor networks. It effectively addresses the critical problems of communication congestion and time synchronization, resulting in more precise routing decisions and enhanced network performance. Scalability ensures the model's applicability in real-time settings. Experimental results demonstrate its superiority, with observed benefits including a 10.5% reduction in delay, an 8.3% reduction in energy consumption, a 12.4% increase in throughput, and a 1.5% increase in packet delivery ratio versus existing methods. These results highlight the practical implications and potential impact of our proposed model for enhancing the performance and effectiveness of dense wireless sensor networks.