REDUCING ROUTING OVERHEAD IN THE DSR ALGORITHM FOR MANETS USING AN EFFICIENT K-MEANS CLUSTERING MECHANISM

ABSTRACT MANET is a temporary wireless network environment where in nodes that is in mobility establishes the network without aid of any fixed infrastructure. Due to increased routing overhead, network congestion and performance degradation happen. This research aimed to explores the enhancement of routing performance in Mobile Ad Hoc Networks (MANETs) by integrating K-Means clustering with the Dynamic Source Routing (DSR) protocol. Data mining and machine learning often utilize the k-means clustering technique to divide data in to categories based on similarities. Previous research has shown that the Dynamic Source Routing (DSR) protocol suffers from high routing overhead in large or highly mobile networks due to frequent route discoveries and route maintenance. Many attempts to improve DSR focus on caching or route shortening, but they do not effectively scale in dense or dynamic topologies. In this research, we proposed K-Means clustering to categorize MANET nodes according to where they are located. To mitigate this overhead, DSR uses various techniques such as caching and route discovery optimization to minimize the amount of routing information that needs to be transmitted. However, these techniques can also introduce additional complexity and overhead into the network, making it important to carefully balance the trade-offs between routing efficiency and overhead. The study addresses critical challenges, including routing overhead and packet delivery ratio (PDR), which significantly impact network efficiency. By applying K-Means clustering, we optimize cluster head selection, resulting in reduced routing overhead and improved PDR. The study demonstrate that our approach yields significant gains in network performance metrics, such as reduced latency and increased throughput, compared to traditional DSR implementations. Simulation results show that our proposed methodology increases 0.189% of packet delivery ratio than DSR and 0.1435% than CBRP. Based on the research finding we recommend that future work should be measure the proposed systems effectiveness in presence of various attacks.