Download now Free registration required
Aiming at the unavoidable factors of random noise interference, sensor node perception vulnerability, network transmission instability, uncertainty actual application environment in WSN, the problem of large amount of data transmission and the transmission energy consumption in centralized data fusion, distributed data fusion algorithm based on Kernel Density Estimation and Nonparametric Belief Propagation(KDE-NBP) is proposed. First, the burst interference was overcome by collaborative monitoring of constructing multi-sensor union; then node sampling data was accurately characterized by KDE for it can approximate any form of density distribution function only with the sampling data; at last, redundant information was removed and data transmission was reduced by NBP distributed information processing method with multi-source data collection, Gaussian mixture and Gibbs sampling.
- Format: PDF
- Size: 817 KB