摘要
在传感器网络中,多个传感器对于同一目标的识别结果经常会发生冲突.本文采用基于D em pster-Shafer证据推理理论的数据融合方法来解决这一问题.然而,采用D-S证据组合公式计算融合结果,计算量过于巨大,对处理能力有限的感知结点来说负担过重,此外,计算所造成的延时也将严重影响系统的实时性和同步性.本文提出了一个基于矩阵分析的快速融合算法,该算法采用了D-S证据理论的思想,计算得到的融合结果与D-S证据组合公式计算得到的融合结果相同.本文用数学归纳法证明了这一结论.经过模拟实验验证,和直接采用D-S证据组合公式相比,该算法的计算量和所需的计算时间明显减少.
In this paper,a data fusion method based on the Dempster-Shafer evidence reasoning theory is introduced to solve the problem of recognition confliction of multi-sensor in the data processing layer of sensor network. However, the huge computational complication of the D-S evidence combination formula which is beyond the processing ability of the sensor nodes impacts the real-time and synchronization requirements of sensor network. A fast fusion algorithm based on the matrix analysis is presented. This algorithm implements the idea of D-S evidence theory. The fusion results of the algorithm and the D-S combination formula are coincident exactly. This paper proves this conclusion by the induction method. Compared with the use of the D-S evidence combination formula,this algorithm surprisingly decreases the time complexity.
出处
《小型微型计算机系统》
CSCD
北大核心
2006年第7期1321-1325,共5页
Journal of Chinese Computer Systems
基金
国家自然科学基金项目(60475012)资助
国防创新基金项目(KMAQ22217004)资助
总装备部预研基金项目(514160400IHT0157)资助.
