• Proposes SAPC-DKNN, a dynamic K-nearest neighbor indoor positioning method that uses RSS path loss and RSS fluctuation area to quantify access point significance and credibility.
• Introduces ND-DKA, an automatic neighbor-density-based K-value algorithm that optimizes the number of nearest neighbors for each test point, overcoming the limitations of a fixed K value.
• Achieves 15.41%–64.74% reduction in average positioning error relative to state-of-the-art KNN methods across three indoor positioning datasets.
• Demonstrates a robust fingerprint-based Wi-Fi positioning framework with strong AP credibility weighting, suitable for patient monitoring, equipment scheduling, and laboratory safety.