WiFi based indoor localization with multiple kernel learning

To solve the problem of low accuracy in real-time localization in indoor environment, we propose a novel localization algorithm with Multiple Kernel Learning (MKL). In our work, the indoor localization is viewed as multiple classification. First we select some reference nodes in the indoor area, and measure the WiFi signal strength of reference nodes for multiple times to construct the classifiers based on multiple kernel learning. When the object enters into the location area, we measure its WiFi signal strength and input it into classifiers to discriminate its label. According to the classification result, the location of the object can be estimated. Experimental results demonstrate that our algorithm is able to effectively locate the object in indoor environment.