User behavior acquisition using Markov chain from Wi-Fi log

Abstract

In this research, we analyze the Wi-Fi log created every day in the recent widespread Wi-Fi environment, and make an effort to acquire user behavior. First, since Wi-Fi log will be huge and a calculation time to get users transitions will be very long, so we propose an algorithm for calculating the transition probability matrix and show a calculation time as a guideline for . Moreover, in order to confirm an ergodicity, that is the basic property of Markov chains, from the obtained transition probability matrix, we clarify a relationship between sites from a directed graph. By using basic properties of Markov chains of the transition probabilities obtained from Wi-Fi logs, we aim to get users behavior.