Indoor Fingerprint Location Algorithm Based On Global Locally Preserving Projection

In indoor fingerprint positioning, the received signal strength indicator (RSSI) is easily affected by interference and noise. Obstacles and signal reflections can cause temporal and peak variations in signal strength, resulting in possible dispersion of the closest reference point (RP) set and consequently affecting the accuracy of positioning. To address this issue, a novel approach called global locally preserving projection (GLPP) method has been proposed. This method effectively reduces the overall noise of RSSI fingerprints and improves the reliability and adaptability of fingerprints by using GLPP to train both offline fingerprint database and online fingerprints. During the online phase, the enhanced weighted k-nearest neighbor (EWKNN) algorithm is employed to estimate the target position. Through experimental validation, our method has demonstrated higher accuracy and stability compared to other approaches.