基于改进神经网络的光通信组网抗毁技术研究

针对光通信组网对网络传输实时性不足的影响,以提高光通信组网的性能为目的,提出了基于改进神经网络的光通信组网抗毁技术研究。在改进神经网络的基础上,模拟网络节点在光通信组网中传输,运用节点的变化计算出组网节点的度参数,结合平均路径长度参数和介数参数,设计了光通信组网特征参数,利用光通信组网的路由策略,设计了光通信组网抗毁算法,最后通过重构光通信组网,实现了光通信组网的抗毁。实验结果表明,基于改进神经网络的光通信组网抗毁技术在网络鲁棒性和效率方面,都具有较强的网络优势,提高的光通信组网的性能。

基于EPON技术的配网光通信网络建设

随着我国电力市场的逐渐开放,我国电力企业面对的竞争压力也越来越大,为了更好的适应电力市场的发展与竞争,我国电力企业应进行深入改革,建设高技术含量的智能电网即数字化配电网,而数字化配电网的运行质量取决于配电网的通信系统。EPON是一种成本造价不高、维护简单的新型通信技术,是配网光通信系统的应用基础,也是数字化配网建设的最佳选择。尽管EPON技术仍有着一些不足,但随着该项技术的研发和创新,以EPON技术为支持的网络运行模式必将在未来电力技术改革和发展中占有相当重要的地位。

Self-similar traffic analysis in optical burst assembly

This paper investigates the traffic properties before and after assembly at edge node of Ethernet over optical burst switching （OBS） network for the first time. Burst and inter-arrival time distributions are simulated under time-based and length-based assembly schemes. Self-similar traffic Hurst parameter is compared through R/S and V/T plot. Finally three self-similar traffic generating methods are given. Simulation resuhs demonstrate that, muhi-source traffic increases self-similar degree, however after assembly, time-based scheme can decrease self similar degree, and aggregated burst size is close to Gaussian distribution. Length-based method has no effects on the self-similarity of input traffic. RMD is fit for study of burst network with large self-similarity.