A frequency hopping method for localization system is aimed to overcome the degradation of location accuracy due to radio interference if there are some other radio devices using the same radio frequency as a localization system. A Packet Reception Rate (PRR) thresholding or a learning-based approach for the diagnostic test is proposed. In that, a PRR thresholding or a set of parameters trained by Hidden Markov Model (HMM) is used as a criterion to decide whether or not to hop. The proposed hopping mechanism provides an accurate and stable localization with a minimum delay.