To tackle the data link selection problem for upper layer protocols in WSN, the paper proposes an EPRR algorithm that incorporates ARIMA model. First of all, it determines parameters according to historical data, next comprehensively considers the asymmetry in the link to predict PRR. Experiments prove that, compared with prediction results from moving window mean model upon PRR, the model s prediction accuracy is significantly superior to SPRR algorithm; additionally it reflects quite well the changing trends of link quality.%针对无线传感器网络里上层协议选择数据链路的问题,提出一种采用ARIMA(差分自回归滑动平均模型)模型的EPRR算法,首先根据历史数据确定参数,然后综合考虑链路中的不对称性对PRR( Packet Reception Ratio)进行预测.实验证明,对比使用移动窗口平均模型对PRR的预测结果,该模型的预测准确度比SPRR( Smooth Packet Reception Ratio)算法有很大提高,并且能够较好地反映链路质量的变化趋势.
展开▼