Enhancing routing efficiency in Cloud MANET using KNN and Fitness Function for dynamic network environments

Abstract The rapid expansion of Mobile Ad-Hoc Networks and their integration with cloud technologies have led to the emergence of Cloud MANET, a network model that combines the flexibility of MANET with the scalability and computing abilities of cloud technologies. In such a model, several MANET can be combined into one using connection to the cloud. However, frequent node mobility in such networks results in periodic topology changes and link instability, reducing data delivery rates and increasing end-to-end latency. This negatively impacts packet delivery within Cloud MANET, especially in real-time applications. In this paper, a novel routing scheme proposed that use machine learning techniques, it presents an extension of the AODV reactive routing protocol, called AODV-KNN, which uses the K-nearest neighbor algorithm and fitness functions. The purpose of this modifying routing scheme was to improve route discovery, reduce latency, and increase network stable. The article also analyzed the impact of fitness function coefficients on routing in Cloud MANET. For different environments, coefficients were selected to show improvements in the measured metrics. The basic quality of service metrics used to evaluate the proposed routing algorithm were packet delivery ratio, packet loss ratio, number of routing packets, average throughput and end-to-end delay. A comparative analysis demonstrates the advantages of AODV-KNN routing over traditional approaches in dynamic network environments.