Research On Routing Strategy Of Wireless Sensor Networks

Wireless sensor network (WSN) is a practical application of modern control theory in many fields likeindustrial production. For the WSN’s features of auto-organization and multi-hop transmission, the routingalgorithm has become one of the most important factors which can influence the network’s performance.While the production cost of nodes in WSN is decreasing as the new progress has been made on techniqueand manufacturing process, the scale of network is increasing consequently, which may amplify not onlythe monitoring territory, precision and flexibility, but also the effect of local communication’s instabilitycaused by network’s parameters’ dynamic wobble or the external environmental interference. Moreover,the extreme condition which sensors would always work in could make the unstable areas in networkappear more likely. To improve the important signal like fire alert’s transmitting performance under theunstable areas’ interference, this thesis proposed an improved routing algorithm AONDVjr (Ad hocOn-Demand Navigated Distance Vector Junior, AONDVjr) on the basis of hybrid routing procedure ofAODVjr (Ad hoc On-Demand Distance Vector Junior, AODVjr) and tree routing under ZigBee protocolwhich is widely used in WSN to neutralize the influence of unstable area. Furthermore, we simulate thecore mechanism of the algorithm’s theoretical capability and its actual performance, the results indicatethat AONDVjr has a better performance than AODVjr in the assumed working condition, the expecteddesign goals is well achieved.This thesis designed a set of parameters including node stability to have further research and controlon unstable area, and analyze the dynamic varying process of parameters’ discrete time series. Aftercompared and summarized the existing theory conditions, we modeled the parameters’ series and made an analogy between the model’s actual meaning and the concept of signal and noise, the model can beconsidered as the superposition of the effect to local area which produced by the whole network’s smoothenvironmental change, and the local area’s more nondeterministic change itself. Therefore, to improve themodel’s performance and accuracy on reflecting the change’s trend of the unstable area, according to theenvironment of the local area which the node belongs to, we built a real-time adjusting mechanism ofdetecting period. Such mechanism could avoid unnecessary energy consumption while tracking theenvironmental change. On the other hand, the Kalman filter recursive mechanism has been introduced in toprocess the parameters’ series, which can extract and utilize the series’ statistical properties while needn’tmany historical data, it can fit the WSN nodes well for the reason that the nodes usually have only limitedstorage space and computing capability. Simulation result shows that the proposed scheme has a goodperformance and could be applied to AONDVjr.The basic principle of AONDVjr can be established by the analysis of the parameters above. Theperiodic check on environmental parameters can detect the unstable area, then select a node callednavigator at the border to monitor and utilize the unstable area. The selecting and dismissing mechanism ofnavigator is described in detail as the core of AONDVjr’s working process. In addition, the criteria in themechanism also have dynamic relations with the node’s current environment, which make it flexible to thevarying situation. The simulation results indicate the performance advantages of AONDVjr. Furthermore,the relation between parameters’ different values and algorithm’s performance in various networksituations can be revealed by such results.