面向6G通信-感知-计算(通感算)融合的发展需求,亟需突破网络智能感知方法,特别是基于深度学习的业务识别与流量预测。因此,首先提出了基于卷积神经网络的业务类型估计算法,以避免人工提取数据特征的复杂过程与估计误差,并减少训练模型...面向6G通信-感知-计算(通感算)融合的发展需求,亟需突破网络智能感知方法,特别是基于深度学习的业务识别与流量预测。因此,首先提出了基于卷积神经网络的业务类型估计算法,以避免人工提取数据特征的复杂过程与估计误差,并减少训练模型参数数量;然后,将基于注意力机制的序列到序列(Sequence to Sequence,Seq2Seq)算法用于预测业务流量,以解决信息损失问题,并根据预测场景的差异性,使用不同的预测步长,在保证预测准确性的前提下减少预测时间和计算消耗;最后搭建基于微服务的智能内生融合实验平台,并在此平台上实现了流量预测与业务类型估计,该平台以微服务的形式将各种能力拆分为多个网络功能,并使用人工智能(AI)技术将各种能力进行融合,实现功能模块共享,减少网络功能冗余,赋予网络智能扩展能力。实验结果表明流量预测模型预测误差较小、业务类型估计模型准确率较高。展开更多
A heuristic theoretical optimal routing algorithm (TORA) is presented to achieve the data-gathering structure of location-aided quality of service (QoS) in wireless sensor networks (WSNs). The construction of TO...A heuristic theoretical optimal routing algorithm (TORA) is presented to achieve the data-gathering structure of location-aided quality of service (QoS) in wireless sensor networks (WSNs). The construction of TORA is based on a kind of swarm intelligence (SI) mechanism, i. e. , ant colony optimization. Firstly, the ener- gy-efficient weight is designed based on flow distribution to divide WSNs into different functional regions, so the routing selection can self-adapt asymmetric power configurations with lower latency. Then, the designs of the novel heuristic factor and the pheromone updating rule can endow ant-like agents with the ability of detecting the local networks energy status and approaching the theoretical optimal tree, thus improving the adaptability and en- ergy-efficiency in route building. Simulation results show that compared with some classic routing algorithms, TORA can further minimize the total communication energy cost and enhance the QoS performance with low-de- lay effect under the data-gathering condition.展开更多
文摘面向6G通信-感知-计算(通感算)融合的发展需求,亟需突破网络智能感知方法,特别是基于深度学习的业务识别与流量预测。因此,首先提出了基于卷积神经网络的业务类型估计算法,以避免人工提取数据特征的复杂过程与估计误差,并减少训练模型参数数量;然后,将基于注意力机制的序列到序列(Sequence to Sequence,Seq2Seq)算法用于预测业务流量,以解决信息损失问题,并根据预测场景的差异性,使用不同的预测步长,在保证预测准确性的前提下减少预测时间和计算消耗;最后搭建基于微服务的智能内生融合实验平台,并在此平台上实现了流量预测与业务类型估计,该平台以微服务的形式将各种能力拆分为多个网络功能,并使用人工智能(AI)技术将各种能力进行融合,实现功能模块共享,减少网络功能冗余,赋予网络智能扩展能力。实验结果表明流量预测模型预测误差较小、业务类型估计模型准确率较高。
基金Supported by the Foundation of National Natural Science of China(60802005,50803016)the Science Foundation for the Excellent Youth Scholars in East China University of Science and Technology(YH0157127)the Undergraduate Innovational Experimentation Program in East China University of Science andTechnology(X1033)~~
文摘A heuristic theoretical optimal routing algorithm (TORA) is presented to achieve the data-gathering structure of location-aided quality of service (QoS) in wireless sensor networks (WSNs). The construction of TORA is based on a kind of swarm intelligence (SI) mechanism, i. e. , ant colony optimization. Firstly, the ener- gy-efficient weight is designed based on flow distribution to divide WSNs into different functional regions, so the routing selection can self-adapt asymmetric power configurations with lower latency. Then, the designs of the novel heuristic factor and the pheromone updating rule can endow ant-like agents with the ability of detecting the local networks energy status and approaching the theoretical optimal tree, thus improving the adaptability and en- ergy-efficiency in route building. Simulation results show that compared with some classic routing algorithms, TORA can further minimize the total communication energy cost and enhance the QoS performance with low-de- lay effect under the data-gathering condition.