基于数字孪生的自组网多模态数据快速融合

Multimodal data fast fusion based on digital twins in ad hoc networks

随着自组网性能的提升与大规模应用,其内部数据呈现出多模态特征,数据量更是达到了海量级别,为数据融合处理工作带来了极大的挑战。为此,提出基于数字孪生的自组网多模态数据快速融合方法。实时采集自组网数据后,利用NLM算法与卡尔曼滤波算法去除数据中的噪声与冗余信息。然后,构建数字孪生自组网(包括自组网、孪生网络与服务系统),从服务系统加载的数据中提取多模态数据特征,搭建双线性融合模型,从而实现对多模态数据的快速融合处理。实验表明:应用该方法后,多模态数据融合过程的时延始终保持在3 s以下,融合后多模态数据质量系数可达到0.9,证明该方法具有更优的数据融合性能。

With the improvement of the performance and large⁃scale application of ad hoc networks,their internal data presents multimodal features,and the amount of data has reached a massive level,bringing great challenges to data fusion processing.For this reason,a fast multimodal data fusion method based on digital twinning is proposed.After collecting the data of the ad hoc network in real time,the NLM algorithm and Kalman filter algorithm are used to remove the noise and redundant information in the data.Then,construct the results of digital twin ad hoc network(including ad hoc network,twin network and service system),extract the features of multimodal data from the data loaded by the service system,and build a bilinear fusion model,so as to realize the rapid fusion processing of multimodal data.The experiment shows that the time delay of multimodal data fusion process is always less than 3 s after applying this method,and the quality coefficient of multimodal data after fusion can reach 0.9,which proves that this method has better data fusion performance.