The invention introduces a method for indoor positioning, executed by a processing unit of a first reference node, which contains at least the following steps. Broadcast signals of second reference nodes are listened. Signal measurements of the broadcast signals are obtained. Identification information is obtained from broadcast messages sent by the second reference nodes. A distance associated with each identification information is obtained. A MLM (Machine Learning Model) is updated according to the signal measurements and the distances, where the MLM describes a linear function between signal measurements and distances.