Reliability Analysis and Modeling of ZigBee Networks

Abstract

The architecture of ZigBee networks focuses on developing low-cost, low-speed ubiquitous communication between devices. The ZigBee technique is based on IEEE 802.15.4, which specifies the physical layer and medium access control (MAC) for a low rate wireless personal area network (LR-WPAN). Currently, numerous wireless sensor networks have adapted the ZigBee open standard to develop various services to promote improved communication quality in our daily lives. The problem of system and network reliability in providing stable services has become more important because these services will be stopped if the system and network reliability is unstable. The ZigBee standard has three kinds of networks; star, tree and mesh. The paper models the ZigBee protocol stack from the physical layer to the application layer and analyzes these layer reliability and mean time to failure (MTTF). Channel resource usage, device role, network topology and application objects are used to evaluate reliability in the physical, medium access control, network, and application layers, respectively. In the star or tree networks, a series system and the reliability block diagram (RBD) technique can be used to solve their reliability problem. However, a division technology is applied here to overcome the problem because the network complexity is higher than that of the others. A mesh network using division technology is classified into several non-reducible series systems and edge parallel systems. Hence, the reliability of mesh networks is easily solved using series-parallel systems through our proposed scheme. The numerical results demonstrate that the reliability will increase for mesh networks when the number of edges in parallel systems increases while the reliability quickly drops when the number of edges and the number of nodes increase for all three networks. More use of resources is another factor impact on reliability decreasing. However, lower network reliability will occur due to network complexity, more resource usage and complex object relationship.