KF‐OLSR: A Novel Routing Protocol for Mobile Ad Hoc Networks Utilizing Kalman Filter and Fuzzy Logic Based on OLSR Routing Protocol

Mobile ad hoc networks (MANETs), and especially Flying ad hoc networks (FANETs), operate in highly dynamic 3D environments that demand routing protocols capable of adapting to rapid topology changes. This paper presents KF‐OLSR, a novel OLSR extension that combines an Extended Kalman Filter (EKF) for predictive mobility estimation with Mamdani‐style fuzzy inference systems to compute fuzzy costs for both Multipoint Relay (MPR) selection and routing‐table construction. The EKF processes historical GPS positions and velocities to produce accurate current and short‐term predicted positions, from which we derive new mobility‐aware metrics: Predicted Relative Displacement (PRD), Predicted Link Lifetime (PLL), and Mobility Variance (MV). These are fused with traditional topology and link indicators—node degree, centrality, and a Link Quality Indicator (LQI, e.g., ETX/Hello reception)—to produce (i) an MPR suitability cost that selects stable, well‐positioned relays and (ii) composite link costs used by a modified Dijkstra algorithm to build routing tables favoring long‐lived, high‐quality paths. Hello and TC messages are extended to carry compact EKF predictions and metric summaries so nodes can compute fuzzy costs locally without additional message types. We also present an analytical modeling and formal analysis framework that derives theoretical performance bounds on packet delivery ratio, end‐to‐end delay, and route stability as functions of prediction accuracy, node density, and mobility dynamics, and quantify the protocol's computational and communication overhead. These analyzes show that KF‐OLSR's gains persist under bounded prediction errors and identify operational regions where the protocol provides provable improvements over baselines. NS‐2 simulations using a Gauss–Markov mobility model validate the analytical results and show that KF‐OLSR significantly outperforms E‐OLSR, ETX‐OLSR, ML‐OLSR, and MD‐OLSR—reducing end‐to‐end delay by up to 28.57%, increasing packet delivery ratio by up to 79.13%, and improving throughput by up to 120.41%—demonstrating the effectiveness of combining predictive analytics with fuzzy decision‐making for airborne ad hoc networks.