Comparing Simulative and Formal Methods for the Analysis of Response Times in Networked Automation Systems Jürgen Greifeneder, Liu Liu, and Georg Frey Electrical and Computer Engineering Department University of Kaiserslautern, Kaiserslautern, Germany e-mail: {greifeneder|liuliu|frey}@eit.uni-kl.de). Abstract: Networked Automation Systems (NAS) result from the increasing decentralization of automa- tion systems using new network structures. Those structures are less expensive and more flexible than traditional ones. However, they introduce stochastic and coupled temporal behavior. Therefore, a detailed analysis is necessary accounting for the special characteristics of NAS. In this article, two approaches for the analysis of response times in NAS are presented. While simulation using Dymola/Modelica offers a user-friendly implementation of the system models, probabilistic model checking using PRISM gives more accurate and reproducible results in less time. The strengths and weaknesses of the two approaches are discussed based on a typical NAS scenario. The results are then validated by a large number of meas- ured samples. It is demonstrated that quite accurate results are obtainable by both approaches. 1. INTRODUCTION The trend towards an increasing decentralization in automa- tion systems by means of new network structures leads to Networked Automation Systems (NAS, Fig.1). Due to those networked and decentralized architectures, a variety of delays with probabilistic duration are introduced into NAS. These aspects have direct influences on dependability, quality, safety, and reliability issues of automation processes. cyclic requests read sensors+ actuators I/O 1 answering time: 2 ms PLC 1 -I/O cycle time 17 ms Inputs Outputs PLC PLC 1 cycle time: 10 ms write I/O 3 I/O n I/O 2 sensors actuators network read PLC 2 -I/O cycle time: 11 ms write cyclic requests C C C ... PLC PLC 2 C 13 ms execution sensors+ actuators sensors+ actuators Fig. 1 Example schematic of a Networked Automation Sys- tem (NAS). The analysis of response times (i.e. delays) lies the basis for the quantitative evaluation of temporal system properties. However, only few methods are feasible for such an analysis (see section 2). This paper is arranged to cover such issues in detail, and is organized as follows: In the third section, a simulative approach using the simulation environment Dy- mola is introduced, followed by the presentation of a formal approach based on Probabilistic Model Checking (PMC) in section 4. Section 5 compares these two methods. Further, the two approaches are applied to a case study and the obtained results are compared with extensive laboratory measurements in section 6. Finally, some important points are summarized and an outlook is given. 2. REQUIREMENTS FOR ANALYSIS METHODS For the analysis of response times in a NAS, it is necessary to take account of the process shown in Fig. 2. The process to be supervised covers the signal change at a sensor, as well as the associated signal processing and the resulting reaction at the actuator. Such a procedure begins with sending the re- quest message from the PLC-I/O to the field-I/O. After being transmitted through the network, processed by the field-I/O and transmitted back, the replied message is processed by the PLC. In this course an associated actuator instruction (to- gether with the next inquiry on sensor) is sent from the PLC- I/O to the field-I/O through the network. The process ends with the activation of the actuator. PLC network PLC-I/O field-I/O sends request transport trigger event handling sensor transport proce- ssing sends request transport event handling actuator Fig. 2 Response time in NAS. If failures, errors, and queuing times shall be considered, it is necessary to know the corresponding occurrence probability functions. Furthermore, it is important to consider the times, Proceedings of the 17th World Congress The International Federation of Automatic Control Seoul, Korea, July 6-11, 2008 978-1-1234-7890-2/08/$20.00 © 2008 IFAC 5113 10.3182/20080706-5-KR-1001.3025