Distributivity,Robustness,and Security In Complex Systems

With the popularization and development of information technology,many complex systems arise,such as automated manufacturing systems and access control systems.Such systems are usually composed of discrete states,and the evolution of states are driven by events among states.Complex systems are of large scale and extremely complex function and structure.In addition,such systems also have the characteristics of multi-objective,multi-constraint,dynamic,asynchronization,concurrency,and uncertainty.How to perform effective analysis and control on such systems so as to ensure their enforceability,robustness,and security has become a challenging scientific problem in the field of control.This is also a key technical problem to be solved urgently in the industry.However,due to the inherent characteristics of complex systems,it is difficult to model them by traditional differential or difference equations.Therefore,the modeling,analysis,and control of such systems has become a hot research issue in the field of control.This thesis focuses on the investigation of various supervisory control theories and techniques in the context of complex systems.It is supported by Natural Science Foundation of China and Major Fundamental Research Program of the Natural Science Foundation of Shaanxi Province,under grants 60474018 and 2017ZDJC-34,and entitled “Robust supervisory control of automated manufacturing systems” and “Comprehensive optimization of logic and performance of automated manufacturing systems”,respectively.Due to explicit or implicit reasons,logic control specifications usually are enforced in complex systems in the form of supervisory control so as to avoid the occurrence of undesirable behaviors,such as deadlock.Real-world systems are usually very complex,and the corresponding control specifications may be astronomical in their scales.As a result,conventional centralized resolution always suffers from computational difficulty.Therefore,it is important to take a distributed control strategy with wide range of application and high efficiency of online operation.However,in dynamic environment,each subsystem of the complex global system often has a limited perception range,and the global environment is usually unknown.Hence,it is a challenging technical problem to perform reasonable and effective supervision and control with respect to control specifications.In actual production,resource failure is a very common phenomenon.For the real-world manufacturing systems,the closed paths of different processes are crisscross.When one or more processes that use unreliable resources stagnate due to resource failures,the execution of other processes using only reliable resources may also be blocked.The stagnation obviously has a propagation effect.In the worst case,the whole system may completely stagnate due to a single resource failure.Hence,it is of great practical significance to ensure the non-blocking of automated manufacturing systems in the case of resource failures.Generally speaking,an automated manufacturing system has high flexibility and productivity if it owns more permissive behaviors.Therefore,with the complexity of the production,how to ensure the maximally permissive(optimal)robust supervision and control is of significance to improve the production efficiency and the stability of manufacturing.For complex systems,in order to protect the private data from being illegally accessed by unauthorized users,it is necessary to establish a flexible and comprehensive access control mechanism to control the permissions of administrators and ordinary users.At the same time,in order to ensure the security of the system,it is necessary to formally analyze and verify the satisfiability of various authorization security constraints in the access control strategy,so as to avoid the access failure caused by the restriction of resource access during the operation of the system.However,due to the complexity of real-world systems,it often includes a large number of users,and the roles and permissions assigned to the users are also very complex.As a result,the traditional methods that are based on state enumeration suffer from state explosion and curse of dimensionality problems.Moreover,the existing methods barely consider comprehensive and effective solutions to authorization violations.Due to their graphical nature,formal semantics,solid mathematical foundation,and abundant analysis techniques,Petri net are widely used by researchers and practitioners to model and analyze various types of discrete and parallel systems.Therefore,this thesis uses Petri nets to investigate the key problems of distributivity,robustness,and security in complex systems.This thesis makes the following pioneering and promising research contributions as briefly summarized as follows.First,this thesis realizes the distributed control of complex systems.By combining model predictive control and integer linear programming,this thesis realizes the control of the global system in a local and real-time way with the aid of shared information among different subsystems.As a consequence,the whole system can run in a cooperative autonomous way.This thesis focuses on analyzing and controlling each subsystem from the perspective of pure structure.With the help of key structural information,the control of the global system is achieved through the strategy of real-time behavior prediction.In this sense,it can realize online rather than offline,closed-loop rather than open-loop,globally optimal rather than locally optimal deadlock avoidance strategy and logic specification enforcement technology.This thesis shows a solid theoretical foundation for complex systems to realize the control of logical behavior and ensure the optimality of system performance.Second,this thesis investigates the maximally permissive robustness analysis of complex systems.The physical resource failures are formalized by removing all unreliable transitions in Petri nets.States are subdivided into strongly robust and weakly robust as well as non-robust states.By using reachability graph and integer linear programming,general methods are proposed to check the robustness of all states.Furthermore,a necessary and sufficient condition is presented to check the robustness of markings with unreliable resources and uncontrollable events,so called robustness controllability theorem.This thesis provides technical support for the maximally permissive robustness analysis of complex manufacturing systems,and a theoretical basis for subsequent research.In addition,the research in this thesis can also be successfully applied to other resource allocation systems,which fully shows its broad application prospects.Third,this thesis performs the security analysis of complex systems.Various authorization constraints in role-based access control system can be analyzed in complex,dynamic and multi-domain environment.Based on integer linear programming,generalized methods are proposed to check the satisfiability of separation of duty constraints,binding of duty constraints,and constraints of cardinality.Furthermore,the enforcement of control places and administration of access control strategy is presented to solve the authorization violations.Then,a new framework is proposed for resiliency verification and analysis of access control system.Finally,a distributed solution strategy for avoiding role inheritance violations and separation of duty violations in multi-domain environment is proposed.This thesis provides a strong support for guaranteeing the security of various access control systems.