Survivability Evaluation Of Wireless Sensor Network Based On Continuous Time Markov Chain

With the rapid development of sensor technology,the application of sensor networks in various fields has become more and more widespread.Because sensors are often deployed in unattended,harsh conditions,the sensor network is very vulnerable to damage.In this case,it is of great significance to evaluate whether the network can continue to provide critical services or not.Survivability refers to the ability of the network to continue to provide critical services after being attacked.Therefore,it is very important to evaluate the survivability of wireless sensor networks(WSN).In order to study the survivability of wireless sensor networks,firstly,this thesis proposes a parallel combined network evaluation model to solve the dependence of the existing evaluation model on node distribution and network topology.Based on this,single node-related survivability evaluation indicators and network-related survivability evaluation indicators are designed and calculated.The evaluation model has solved the problems that the evaluation indicators are single and the network survivability cannot be fully evaluated.The main research work of this thesis is as follows:(1)Based on the combination of single-node behavior,a network survivability evaluation model is proposed.Firstly,the continuous-time Markov chain(CTMC)is used to establish the state transition model of a single node under attack and other destructive behavior.The model fully considers the state changes of the node after being attacked by the black hole and the jellyfish attack.Secondly,a parallel combination model based on CTMC is established,by introducing a connectivity function to record the connectivity state between nodes,the problem of node distribution and network topology structure in other models is effectively solved.(2)According to the evaluation of single-node survivability,three evaluation indicators are proposed,namely single-node steady-state availability,single-node instant availability and single-node expected attack times.The single-node steady-state availability evaluates the probability node is in normal working state during long-term operation,single-node instant availability evaluates the probability that node can work normally at a certain moment,and the single-node expected attack number is used to evaluate the node being attacked in a certain period of time.In addition,CTMC-based calculation methods are also designed for these three indicators.(3)Based on network survivability evaluation,two evaluation indicators are proposed,namely the network steady-state k-connection probability and the network minimum availability.The network steady-state k-connection probability evaluates the probability that nodes in normal working condition are connected to the network in the long-term operation process.The network minimum availability evaluates the probability that the number of nodes in normal working condition during the long-term evaluation network is not less than the number of nodes at the time of network failure.In addition,CTMC-based calculation methods are also designed for these two indicators.Experimental results show that the effect of transfer rate on the connectivity probability and the minimum availability of the network is consistent with the change of network connectivity and availability when the actual network is attacked or destroyed.The proposed models and indicators are effective.