A Novel Indoor Intelligent Localization Strategy Based on KPCA and CPO-SVR Technology

To improve the accuracy and robustness of the indoor localization strategy based on SVR, a localization strategy based on KPCA (Kernel Principal Component Analysis) fingerprint feature enhancement and CPO-SVR (Support Vector Regression based Crested Porcupine Optimizer, CPO-SVR) localization model is proposed. The KPCA algorithm is used to reduce the dimensionality and enhance the features of the fingerprint data, effectively separating the complex relationship between signal strength and spatial position. During the training SVR model process, the CPO optimizer is introduced to optimize the hyperparameter combinations of SVR, which simulates the defense behaviors of the crested porcupine, thereby improving localization accuracy. The experimental results show that the average localization error is 2.35 m. Compared with the genetic algorithm optimizer, particle swarm optimizer, and snake optimizer, the localization accuracy has been improved by 10.0%, 24.4%, and 25.7%, respectively.