A communication device includes a smart antenna, a storage device, and a processor. The smart antenna is capable of switching between a plurality of antenna modes. In each of the antenna modes during a training stage, the smart antenna receives a first feedback datum, and the processor calculates a reward indicator according to the first feedback datum. The processor compares all of the reward indicators with each other and controls the smart antenna to select a specific mode of the antenna modes according to a comparison between all of the reward indicators. In the specific mode, the smart antenna receives a second feedback datum, the processor determines the weight function of the first feedback datum and the second feedback datum of the specific mode, and the processor updates the reward indicator of the specific mode according to the weight function.