Wireless sensor networks (WSN) is the key resource of perception and is widely used in the systems based on Internet of Things (IoT). The smart sensor nodes are used in applications like infrastructure monitoring, medical health care systems, etc. But these nodes are energy constraint devices. Efficient clustering and proper cluster head (CH) selection schemes are required, in order to improve energy saving of sensor nodes. In this paper, dynamic CH selection method (DCHSM) is used where CHs are selected in two phases. This algorithm improves energy saving on large scale thus can be used for IoT applications. Initially, Voronoi diagram is used to divide the monitoring area in polygonal shaped clusters. Then, CH election is performed in two phases. First class of CH is elected based on perceived probability and the second class is elected on the basis of survival time estimation. Simulation analysis show that DCHSM outperforms the conventional methods in terms of network lifetime. © 2017 IEEE.