Communication And Control Co-Design For Real-Time Cyber-Physical Systems

In real-time cyber-physical systems,wireless communication design needs to pro-vide ultra-reliable and low-latency wireless communication(URLLC)to support real-time wireless control.Most conventional methods only focus on wireless system design from communication perspective,where diversity transmission and short packet transmission techniques are adopted to guarantee stringent quality-of-service(QoS)in URLLC.This consumes significant amount of wireless resources and impedes the development of cyber-physical applications.This thesis intends to conduct research from the perspective of com-munication and control co-design,and study the interdependence relationship between wireless communication system and wireless control system.In particular,this thesis jointly considers wireless transmission method design and optimal control parameter de-sign,which is expected to be the key to overcome the bottleneck of URLLC for future wireless control in real-time cyber-physical systems.Specifically,this thesis first considers a typical real-time control system,exploits the dynamic features of the control system,and designs the corresponding dynamic resource matching methods.Then,this thesis considers the the timeliness of control information,exploits the control performance based on the timeliness of control information,and pro-vides guideline for control information update scheme in real-time wireless control sys-tems.The detailed works of this thesis can be summarized as follows.1)This thesis proposes a communication and control co-design scheme based on communication dynamic QoS allocation,where the usage of extremely high QoS in URLLC can be significantly reduced.This makes it possible to use dynamic QoS allocation dur-ing control process to reduce wireless resource consumption.First,this thesis analyzes the relationship between communication QoS and control process,and obtains the influ-ence of different communication QoS on control performance.Based on that,the control process can be divided into two phases:in the first phase,high QoS has better control performance than low QoS;in the second phase,low QoS outperforms high QoS in con-trol performance.Then,this thesis provides the proof for the above phenomenon.Finally,a dynamic communication QoS allocation algorithm is proposed,and the threshold for communication QoS allocation is obtained.Compared with only using URLLC through-out the control process,the proposed dynamic communication QoS allocation method can significantly reduce communication energy consumption while maintaining good control performance.2)Based on the above work proving that communication and control co-design can significantly improve the overall system performance.This thesis studies the optimal wireless communication resource allocation by communication and control co-design for real-time wireless control systems,where this work focuses on centralized resource allo-cation at the base station(BS).While maintaining the control requirement,the optimal communication bandwidth,energy consumption allocation,and control convergence rate can be obtained by maximizing the spectral efficiency of the communication.The formu-lated optimization problem considers both the QoS constraint in URLLC and the control convergence rate constraint.The key to solving this problem is to find a way to convert the constraint on the control sub-system to the constraint of the communication sub-system.Then,this thesis analyzes the relationship between the control convergence rate constraint and the URLLC requirement in communication.The lower bound of communication reli-ability is found to be constrained by a function respect to control convergence rate where the original hybrid co-design problem can be converted into a regular communication op-timization problem.Then,it is proved that the optimization problem based on the above conversion is a concave problem.Finally,an iterative algorithm is proposed to find the optimal bandwidth and power allocation.3)Considering distributed wireless control scenario,a new autonomous device-to-device(D2D)transmission scheme is proposed to deal with the extremely high QoS re-quirement in URLLC for real-time wireless control systems.The goal is to minimize the energy consumption of D2D transmission by jointly considering URLLC and con-trol constraints.This thesis focuses on both how to activate D2D transmission and how to maintain control requirements.First,a probability-based D2D activation and transmission power allocation scheme is proposed to guarantee control requirements,which allows each transmission device to autonomously decide whether to activate to participate in the con-trol process without exchanging information with each other.This guarantees extremely high QoS constraints in URLLC.Then,by converting the convergence rate of communi-cation into communication reliability constraint,the effect of control constraints on D2D transmission can be obtained.Finally,an optimal method is proposed by balancing the failed reduction efficiency.The proposed method can optimize the activation probability and minimize the transmission power under the constraints of wireless communication reliability,while maintaining control requirement.4)A generalized age of information(AoI)is proposed for actuation update to evaluate timeliness of control information in real-time wireless feedback control systems.This thesis adopts a first-come-first-served M/M/1/1 queuing model to describe the actuation update for a typical predictive wireless feedback control system.Then,it can be found that the initial time for each actuation update is the prediction time of the most recently received actuation update,which is significantly different from the conventional AoI calculation.Considering different prediction lengths,the calculation of the average AoI in this thesis can be divided into two cases,where the conventional AoI calculation for status update is a special case of this thesis.Then,the closed-form expressions for the average AoI calculation in both cases are obtained.In addition,the mean-square-error(MSE)of the control state update is proposed as a criterion to evaluate the control performance affected by actuation update.The relationship between the average AoI and MSE is analyzed.Based on the above analysis,some properties are proposed to summarize the relationship between them,which provides a guideline for the future control system design.The results of this thesis are expected to minimize the wireless resource consumption while maintaining control performance requirement,and provide a guideline for control system design when considering timeliness of system update,which provides a new di-rection for URLLC research in real-time wireless control systems to enable future cyber-physical applications.