WiFi fingerprinting positioning systems have been used in tunnels. But, the limitation and complexity of the tunnel environment, the solution to achieve a high accurate positioning system remains open. In order to solve this problem, we have done the following three tasks: 1) Different access point (AP) layout schemes are proposed for several different tunnel environments; 2) A dynamic fingerprint database based on test point signal is established; 3) An improved algorithm is designed by combining k-nearest neighbors(KNN) algorithm and fuzzy C-means clustering (FCM) algorithm. The experimental results show that the accuracy of WiFi positioning in tunnel is improved.