Federated Genetic Optimization for Secure and Privacy-Preserving Sensor Localization in Consumer IoT Applications

Traditional localization methods in the Internet of Things often rely on centralized processing of distance measurements, which makes them vulnerable to adversarial data injection, privacy leakage, and scalability limitations. Methods such as the received signal strength indicator, the time of arrival, and range-free protocols like DV-Hop are typically designed for static, noise-free, and resource-rich environments. In this work, a novel Federated Genetic Algorithm (FedGA) is proposed for robust and privacy-aware sensor localization in consumer Internet of Things environments. FedGA logically divides the network into several federated clusters, where nodes use genetic optimization to compute location estimations. Only elite candidate solutions are shared with a central aggregator, ensuring data confidentiality and minimal communication overhead. Through rigorous simulation under varying node densities and measurement noise, FedGA demonstrates high localization accuracy, resilience to noise and partial data tampering. It has been observed that the FedGA improved localization accuracy by 19% as compared to the state of the art federated localization algorithms.