Wireless sensor networks typically consist of a significant number of low-cost sensor nodes which can collaborate among each other to execute a wide range of applications, such as environmental and health monitoring, intrusion detection, etc. Basically, sensor nodes are mainly battery-powered and have a limited amount of energy to spend, thereby they are not likely to be recharged or replenished after being deployed in an inaccessible sensed environment. As a result, in the past decade, power saving has become a critical issue for the design of a media access control (MAC) protocol in wireless sensor networks. By incorporating power-saving mechanism into MAC, we expect to maximize the lifetime of a sensor network. Intuitively, a large amount of research tends to make use of the TDMA protocol because it has inherently more favorable characteristics when compared to the CSMA protocol, that is, sensor nodes only need to wake up to transmit or receive data at its allocated time slots, and other time slots they can go into sleep mode. By doing so, it can reduce power consumption. However, in this master thesis, we found that there exists a rather acute problem, called the Multiple-Cluster Collision (MCC) problem, in the TDMA cluster-based protocol that will degrade the system performance drastically. Therefore, in this thesis, we propose the approach, called Multiple Cluster Collision Problem (MCCP), to perfectly eliminate the MCC problem by sensor nodes. Finally, we will compare our approach with SRSA, proposed by Tao WU et al., to show that the performance of our approach outperforms SRSA in terms of collision reduction, throughput and so on.