Abstract
Constructing a Large Scale Wireless Sensors NetworkA real challenge of wireless sensors network applications is how to expand it to a larger scale. We proposed the idea of “One Zone Multi-PAN (Personal Area Network)” to build a Large-Scale Wireless Sensors Network (LS-WSN). Its key points include: the elementary component of LS-WSN is PAN; different PANs work on different channels; different PANs use customized policies and execute different tasks; every PAN covers and surveillances the same area, but senses different circumstance information.
Based on the idea, we proposed a two-level distributed LS-WSN model. The first level is the PAN-level, which works in a distributed mode. On this level, the up to 16 PANs bridged by a specially designed gateway coexist like a multi-layer sandwich. This gateway is a common children node of different PANs; therefore it can join in different PANs in a time-sharing mode. The second level is the inner-PAN-level, whose work mode is determined by the specific manufacturer.
The size of this model can be expanded up to 16 times of the size of a single PAN. Furthermore, QoS of LS-WSN can be guaranteed, data fusion and data management can be simplified and improved. Using Zigbee stack, we set up a prototype system of LS-WSN, whose performances are validated by our experiments. Resources Scheduling among PANsFor executing resource reconfiguration in LS-WSN, we proposed an application framework of resources scheduling (RSAF). Its key principles include: radio channels, superframe time-slots, and energy resources are named as “Common Resources;” “the whole network is more important than a single PAN” in LS-WSN; keep the key PANs alive; the lifespan of a LS-WSN is determined by that of key PANs.
RSAF can be comprehended as a collaboration mechanism at PAN level. It includes three components: superframe time-slots scheduling, radio channel scheduling, and energy scheduling. Their algorithms are different. Especially, a typical working cycle of energy scheduling contains four steps. First, the coordinators of different PANs gather different PANs' state information in a piggyback mode. Then the gateway periodically evaluates different PANs' states and makes the decisions of resources scheduling. After that, the specified coordinators send the control messages to certain nodes. Finally, the specified nodes will join and reinforce the key PAN in danger due to certain reasons (e.g., out of energy). Our experimental results validate the feasibility of RSAF in our prototype system.