An Indoor Location Method Based on Features Optimization for Different Regions With Improved Curve Smoothness Index

Traditional Wi-Fi-based indoor localization techniques are susceptible to different environments. Furthermore, the Wi-Fi fingerprint method is low efficient for localization with large database matching time-consuming. To address the above problems, this article proposes a feature fusion-based multifingerprint partition localization method. In the offline phase, the feature extraction is carried out for different regions based on the improved curve smoothness index of signal intensity descending order. In the online phase, a
${k}$
-nearest neighbor (KNN) classifier is used to locate the test points to approximate regions. Then, traditional positioning algorithms are used to achieve the precision localization in the region of interest. The experimental results show that the method based on features optimization for different regions is highly generalizable to more complex environment, improving the efficiency and accuracy.