基于静态权值组合集成模型的WSN时钟偏差估计

Clock bias estimation for WSN based on static weights integration model

针对无线传感器网络(WSN)时钟同步精度低、复杂度高等问题,提出了一种基于静态权值组合集成模型的时钟偏差预测方法。对传感器节点的时间戳观测值进行有放回抽样,将面向回归问题的Ada Boost.RT集成学习算法的误差函数和阈值调整方法进行改进,并以改进的Ada Boost.RT算法作为集成框架,采用DPNN作为弱学习机构建集成局域模型对时间偏差进行有效预测。实验表明,对于长期预测,Ada Boost.RT模型和改进型Ada Boost.RT模型的预测效果相对于DPNN全局模型提升了20%。此外,在长期观测和短期观测两种情况下,Ada Boost.RT改进型模型的预测效果要优于Ada Boost.RT模型,能够更有效地减小时间估计偏差。

Aiming at the low accuracy and high complexity in the clock synchronization procedure for WSN wireless sensor network,this paper proposed a clock error prediction method based on static weight composite integration model. Sample back into the observed time stamp value of sensor nodes,and then improved the error function and threshold adjustment method of the algorithm for the regression problems based Ada Boost. RT integrated learning algorithm. Lastly,the improved algorithm would be used as an integration framework,and DPNN would be used as weak machine learning to build integrated local model to effectively predict the time deviation. Experiments show that for long-term prediction,prediction effect of the Ada Boost. RT model and the improved Ada Boost. RT model increased by 20% compared with DPNN global model. In addition,in both the long-term and short-term observations,the predicted effect of the improved Ada Boost. RT model is superior to Ada Boost. RT model,and it can more effectively reduce clock estimation bias of the WSNs.