Modeling and simulation of adaptive Neuro-fuzzy based intelligent system for predictive stabilization in structured overlay networks
Modeling and simulation of adaptive Neuro-fuzzy based intelligent system for predictive stabilization in structured overlay networks
复制标题
基于自适应神经模糊的智能系统的建模和仿真，用于结构化覆盖网络中的预测稳定
DOI：
10.1016/j.jestch.2016.06.015
复制
发表时间：
2017
期刊：
影响因子：
--
通讯作者：
Krishan Kumar
中科院分区：
文献类型：
--
作者：
Ramanpreet Kaur;A. L. Sangal;Krishan Kumar
智能预测邻居节点（dht协议中定义的k个邻居）的动态性有助于提高结构化覆盖网络的弹性，并减少拓扑维护的开销。覆盖节点的动态行为取决于许多因素，诸如底层用户的在线行为、地理位置、一天中的时间、一周中的某一天等，如在许多应用中所报告的。我们可以利用这些特点，有效地维护结构化的覆盖网络，通过实施一个智能预测框架，适当地设置稳定参数。考虑到人类驱动的行为通常超越间歇性的可用性模式，我们使用一个混合的神经模糊预测器，以提高预测的准确性。在本文中，我们讨论了我们的预测稳定的方法，实现神经模糊的预测在MATLAB仿真和应用这种预测稳定模型的和弦为基础的覆盖网络使用OverSim作为仿真工具。MATLAB仿真结果表明，相邻节点的行为在很大程度上是可预测的，如非常小的RMSE所示。基于OverSim的仿真结果也观察到显着改善的性能，基于和弦的覆盖网络的查找成功率，查找跳数和维护开销相比，定期稳定的方法。
Intelligent prediction of neighboring node (k well defined neighbors as specified by the dht protocol) dynamism is helpful to improve the resilience and can reduce the overhead associated with topology maintenance of structured overlay networks. The dynamic behavior of overlay nodes depends on many factors such as underlying user’s online behavior, geographical position, time of the day, day of the week etc. as reported in many applications. We can exploit these characteristics for efficient maintenance of structured overlay networks by implementing an intelligent predictive framework for setting stabilization parameters appropriately. Considering the fact that human driven behavior usually goes beyond intermittent availability patterns, we use a hybrid Neuro-fuzzy based predictor to enhance the accuracy of the predictions. In this paper, we discuss our predictive stabilization approach, implement Neuro-fuzzy based prediction in MATLAB simulation and apply this predictive stabilization model in a chord based overlay network using OverSim as a simulation tool. The MATLAB simulation results present that the behavior of neighboring nodes is predictable to a large extent as indicated by the very small RMSE. The OverSim based simulation results also observe significant improvements in the performance of chord based overlay network in terms of lookup success ratio, lookup hop count and maintenance overhead as compared to periodic stabilization approach.