Wireless RSSI-Fingerprint and KNN Algorithm for improved Accuracy Indoor Positioning system

Abstract

Indoor Positioning Systems (IPS) has gained significant attention due to the increasing demand for location-based services in indoor environments where GPS signals are unreliable or unavailable. This study investigates the application of the k-Nearest Neighbors (KNN) algorithm for indoor localization using Received Signal Strength Indicator (RSSI) values. The proposed system utilizes Wi-Fi signal strengths collected from multiple Access Points (APs) to estimate the position of a target device within a predefined indoor area. A fingerprinting technique is employed, where RSSI measurements are first collected at known reference points during an offline phase. In the online phase, the KNN algorithm identifies the k most similar signal patterns from the database and estimates the current location based on the coordinates of these nearest neighbors. The system's performance is assessed across different k values and varying environmental conditions to analyze the trade-off between positioning accuracy and computational complexity. Results show that KNN offers a robust, low-cost, and computationally efficient solution suitable for real-time indoor localization. Its simplicity and adaptability make it a promising approach for use in mobile and internet of things (IoT) applications. This paper contributes to RSSI-based indoor positioning research by evaluating the performance of the K-Nearest Neighbors (KNN) algorithm in a 10.6 m × 5.8 m room using data from four Wi-Fi access points and 60 reference points. The results demonstrate KNN's feasibility with an average positioning error of approximately 0.5 meters, while also addressing its limitations in handling signal variability and scalability in dynamic or larger environments.