Research On Survivability For Mobile Ad-hoc Networks

The development of Mobile Ad-hoc Networks(MANETs)leads to an increase in application and promotion in various fields,and the dynamic and complexity of MANET environment propose high demand of network reliability and robustness.Survivability for MANET is a key indicator to weigh network performance of dynamic adjustment and communication security in ever-changing environment.Thus we are experiencing the increasing demand for the research of MANET survivability.Survivability for MANET requires adaptation mechanism,the system capacity and skill of quickly incorporating lessons learned from failures and adapting to emerging threats.On the foundation of summarizing relevant theories and research results,we present our work on MANET survivability into three parts considering the adaptation mechanism requirement of survivable MANET.Firstly,MANET survivability framework.In this paper,we model the topology of MANET by Time-varving Graph(TVG),in order to characterize the time-varying features of network.Based on the network model,we propose spatiotemporal survivability definition for MANET with spatial parameter N and temporal parameter T,denoted as S(N,T).Secondly,MANET survivability model and quantification.On the basis of the connection-oriented idea for survivability analysis,TVG-based model is transformed into Time-varying Line Graph(TVLG).TVLG-based model characterizes the connectedness and adjacency in MANET topology.Max-flow with temporal parameter T,denoted by F(T),is set as the quantification of survivability.And we design T-Max-flow survivable routing algorithm based on F(T).The simulation proves the performance of spatiotemporal survivability framework.Thirdly,numerical analysis of survivability model.The survivability analysis model is built by MANET spatiotemporal survivability,including assessment model based on Hidden Conditional Random Fields(HCRPs)and prediction model based on Radial Basis Function Neural Networks(RBFNNs).Besides,we adopt KL divergence as the metric of registration accuracy for anomalous event detection in MANET,and minimize the survivability quantification accuracy by solving convex optimization problem.Finally,we conclude the work in this paper and analyze the improvable research of MANET survivability.