Simulation-Based Optimization Methods For Network-Wide Signal Control Under Cyber-Attacks

With the continuous development of urban traffic signal control technology,an increasing number of traffic control devices and systems rely on network technology,highlighting the issue of network security.Meanwhile,due to the numerous vulnerabilities and potential weaknesses in the network security of signal control systems,the current network security defense measures are unable to guarantee satisfactory operational performance of the traffic signal control system.It is urgent to seek practical and feasible countermeasures to mitigate the negative impact of cyber-attacks on traffic signal control systems.Therefore,based on the traffic simulation optimization technologies,the urban traffic signal control theories and simulation-based optimization algorithms,this dissertation conducts relevant research on the simulation-based optimization method of network-wide traffic signal control under cyber-attacks.The main research achievements include the following aspects:First,a microscopic simulation platform called PYTHON-SUMOSSAM was constructed based on the Tra CI interface technology.This platform enables traffic network simulation control(including traffic signal lights,vehicles and so on),the simulation of traffic control and cyberattacks,and evaluation of traffic efficiency and safety performance.Using real traffic data,including traffic sensors data and traffic control data,a SUMO traffic signal control simulation model for the Changsha road network was modeled.The key parameters of the traffic signal control simulation model were calibrated using the simulation-based optimization calibration method.This ensures that the simulation model accurately reflects the characteristics of the real-world traffic system.Second,from the perspective of attackers,four network attack models were established,including: risk-averse direct attack model,worst-case direct attack model,worst-case indirect attack model,and signal shutdown attack model.These models aim to deteriorate the performance of traffic system with the consideration of attack risk and attack resource.Surrogateassisted simulation-based generic optimization algorithms were designed to solve these models.Based on the constructed traffic simulation optimization platform and the calibrated traffic network signal control simulation model,simulation optimization experiments were conducted to analyze the vulnerability of the road network under the four attack modes.Third,for direct cyber-attacks on fixed-time traffic signal control systems,this dissertation proposes a passive countermeasure by designing robust signal control plans to mitigate the worst-case attack and risk-averse cyber-attack.To achieve this objective,the dissertation develops two bilevel simulation-based optimization(SO)models and solves them by an improved biobjective robust simulation-based optimization(IBORSO)algorithm.The attack-defense process is formulated as a Stackelberg game,where the lower level attacker aims to find optimal attack plans to degrade traffic efficiency and safety performance,while the upper level defender adjusts the signal control plan to compensate for the effects of the attack on traffic efficiency and safety performance.Fourth,for indirect cyber-attacks,this dissertation designs a robust road network control method to mitigate the negative effects of indirect attacks on the adaptive traffic signal control systems.This robust road network control method utilizes nonlinear mapping as the control logic.It initially fits the nonlinear mapping control logic with the model predictive control method.Then,a gradient-enhanced Kriging assisted expected improvement simulation optimization algorithm is developed to obtain the robust key parameters of nonlinear mapping signal control logic,which can generate robust signal timing plans.Simulation experiments were conducted with the developed traffic network simulation platform,and it was found that the proposed method achieves better control effectiveness compared to other traffic signal control methods.Fifth,to defend against the signal shutdown attacks,this dissertation designs a weighted maximum pressure-based traffic signal control method.It develops a robust network signal control simulation optimization technology with consideration the signal shutdown attacks,which can adjust the key parameters of turning importance in the weighted maximum pressure-based control to optimize traffic network signal control logic.The optimized traffic network signal control logic allows real-time adjustment of the traffic signal control plan based on the traffic network state,which helps regulate the traffic flow entering traffic jam areas,alleviating traffic congestion caused by signal shutdown attacks and increasing the traffic network efficiency.Simulation experiments show that the method can efficiently mitigate the traffic congestion caused by signal shutdown network attacks.