【正文】 外文原文和譯文 15th IFAC World Congress on Automatic Control, Barcelona, Spain, July 2022 . TrueTime: Simulation of Control Loops Unde Shared Computer Resources Dan Henriksson, Anton Cervin, KarlErik 197。rz233。n Department of Automatic Control Lund Institute of Technology Box 118, SE221 00 Lund, Sweden {dan,anton, Abstract The paper presents TRUETIME, a MATLAB/Simulinkbased simulator for realtime control systems. TRUETIME makes it possible to simulate the temporal behavior of multitasking realtime kernels containing controller tasks and to study the effects of CPU and work scheduling on control performance. The simulated Realtime kernel is eventdriven and can handle external interrupts as well as finegrained Details such as context switches. Arbitrary scheduling policies may be defined, and the control tasks may be implemented using C functions, M functions, or Simulink block diagrams. A number of examples that illustrate the use of TRUETIME are presented. 1. Introduction Most puter control systems are embedded systems where the puter is a ponent within a larger engineering system. The controllers are often implemented as one or several tasks on a microprocessor using a realtime kernel or a realtime operating system DRTOSE. In most cases the microprocessor also contains other tasks for other functions, ., munication and user interfaces. The kernel or OS typically uses multiprogramming to multiplex the execution of the different tasks on a single CPU. The CPU time and the munication bandwidth, hence, can be viewed as shared resources which the tasks pete for. Computerbased control theory normally assumes equidistant sampling intervals and negligible or constant control delays, ., the latency between the sampling of the inputs to the controller and the generation of the outputs. However, this can seldom be achieved in practice. Tasks interfere with each other through preemption and blocking due to munication. Execution times may be datadependent or vary due to, ., the uses of caches. The result of this is jitter in sampling periods and latencies. An additional cause of this temporal nondeterminism is the increasing use of mercial offtheshelf (COTS) ponents in control systems, ., general purpose operating systems such as Windows and Linux and general purpose work protocols such as Ether. These are designed to optimize average case performance rather than worstcase performance, and therefore increase the nondeterminism. The effects of this type of temporal nondeterminism on control performance are often very hard, if not impossible, to investigate analytically. A natural approach is then to instead use simulation. However, todays simulation tools make it difficult to simulate the true temporal behavior of control loops. What is normally done is to introduce time delays in the control loop representing average case or worstcase delays. In this paper the new simulation toolbox TRUETIME is presented. TRUETIME, which is based on MATLAB/ Simulink, makes it possible to simulate the temporal behavior of a multitasking Realtime kernel containing controller tasks. The controller tasks control processes modeled as ordinary Simulink blocks. Different scheduling policies may be used, .,prioritydriven or deadlinedriven scheduling. The execution times of the controller tasks can be modeled as being constant or timevarying, using some suitable probability distribution. The effects of context switching and interrupt handling are taken into account, as well as task synchronization using events and monitors. With TRUETIME it is also possible to simulate the timing behavior of munication works used in, ., worked control loops. TRUETIME can be used for several purposes: to investigate the true effects of timing nondeterminism on control performance, to develop pensation schemes that adjust the controller dynamically based on measurements of actual timing variations, to experiment with new, more flexible approaches to dynamic scheduling, ., feedback scheduling [Ekeret al., 2022] and QualityofService (QoS) based scheduling approaches, and to simulate eventbased control systems, ., bustion engine control systems and distributed controllers. Figure 1 The interfaces to the Simulink blocks. The Schedule and Monitors ports provide plots of the allocation of mon resources (CPU, monitors, work) during the simulation. Related work While numerous tools exist that support either simulationof control systems (. Simulink) or simulation of realtime scheduling (. STRESS [Audsleyet al., 1994G and DRTSS [Storch and Liu, 1996]) very few tools support cosimulation of control systems and realtime scheduling. An early, tickbased prototype of TRUETIME was presented in [Eker and Cervin, 1999]. Since it was not eventbased this early version had very little support for interrupt handling and could not handle finegrained simulation details. Also, there was no support for simulation of works. The RTSIM realtime scheduling simulator (a standalone C++ program) has recently been extended with a numerical module (based on the Octave library) that supports simulation of continuous dynamics, see [Palopoli et al., 2022]. However, it lacks a graphical plant modeling environment, and so far its work capabilities are limited. Outline of the paper The simulation environment is described in some detail in Section 2. Three examples are then given to illustrate the use of the simulator. The first example treats scheduling during overload conditions. The subject of the second example is worked control system, whereas the last example evaluates an improved scheduling technique for controller tasks. 2. The Simulator The TRUETIME simulation environment offers two Simulink blocks: a puter block and a work block, t