Research On Security-aware Design Optimization Algorithms For Heterogeneous Embedded Systems

Hard real-time heterogeneous embedded systems are composed of a large number of processing elements(PEs)interconnected through a shared bus to support distributed functional end-to-end computing.Aerospace,defense,industrial automation,health and medical equipment,major infrastructure,and road transportation are the main application areas.Human related heterogeneous embedded systems,such as mixedcriticality systems and automotive electronic systems,need to consider security,and unsafe factors can threaten the system itself or even human life.Security is divided into functional safety considered from embedded computing and information security considered from bus communication.The contents and main contributions of this paper are included as fellows.(1)Research on development cost optimization of multi-functional mixedcriticality heterogenous embedded systems.The determination of safety integrity levels(SIL)of the task in the early design stage is a process that meets the lower bound value.However,under the constraints of response time and functional safety,it is impossible for all tasks to be at the minimum SIL level.The elevation of SIL requires additional certification and design,which will cause the developer to pay additional development costs.Heterogeneous embedded systems are cost sensitive and high development costs are not desirable.This paper proposes a method based on key path tabu search algorithm(KPTS)combined with message movement strategy(MMS)to solve the development cost optimization problem.The proposed KPTS algorithm uses a treebased tabu list to find a solution with low development cost and meet response time constraints.The proposed MMS strategy avoids recursive task's SIL elevation and thus significantly reduces development costs.(2)Research on information security enhancement optimization of automotive electronic heterogeneous embedded systems.Encryption services can be deployed to meet the confidential information security needs of messages.However,the introduction of encrypted authentication will add additional burdens and even break the response time constraints of heterogeneous embedded systems.Secondly,choosing different encryption algorithms and encryption rounds requires different system resources and different encryption effects will be obtained.Finally,messages will move backwards due to the introduction of encryption and decryption tasks,and message conflicts need to be considered in the shared bus architecture.To solve this problem,this paper proposes a method that satisfies the message security requirements of high SIL tasks as much as possible and satisfies functional response time constraints while minimizing the total number of vulnerabilities exposed by the system in the early design stage.The proposed linear programming(LP)-based scheduling algorithm reduces the number of vulnerabilities exposed on the bus and the makespan of the overall functions while satisfying the minimum-security encryption requirements for different SIL of tasks.The proposed security enhancement algorithm(SE)choses messages on the bus to perform more complex encryption algorithms and encryption rounds to minimize the number of system vulnerabilities on the basis of LP.In summary,this paper has carried out design optimization work of heterogeneous embedded systems with security requirements,hard real-time constraints,and cost sensitivity feature.It provides a higher-performance security-aware optimization method for the security design optimization of heterogeneous embedded systems in the early design stage,which has certain theoretical significance and application value.