Cyber-Physical System Requirements - A Model Driven Approach

Background

Cyber-physical systems (CPS) – networked

software based systems that extensively

interact with the physical world. Eg: Aircraft

control devices

FDA’s initiative explores various methods to

improve the safety of medical CPS systems

such as infusion pumps.

Complete and consistent requirements are

central to analysis, verification, and assurance

for safety.

Collecting a good set of requirements is not a

straightforward task.

A generic patient controlled analgesia (GPCA)

infusion pump system was used as a case

example to provide an archetype of system

development artifacts that demonstrates safety.

Publications

1. A.Murugesan, S.Rayadurgam, M.Heimdahl. Using Models to

Address Challenges in Specifying Requirements for Medical

Cyber-Physical Systems. Accepted at Fourth workshop on

Medical Cyber-Physical Systems, Philadelphia, PA, Apr

2013.

2. Mats.Heimdahl, L.Duan, A.Murugesan, S.Rayadurgam.

Modeling and Requirements on the Physical Side of Cyber-

Physical Systems. Accepted at Second International

Workshop on the Twin Peaks of Requirements and

Architecture, San Francisco, CA, May 2013.

3. A.Murugesan, S.Rayadurgam, M.Heimdahl. Modes, Features,

and State-Based Modeling for Clarity and Flexibility.

Accepted at Fifth International Workshop on Modeling in

Software Engineering , San Francisco, CA, May 2013.

Research Challenges

Requirements Analyses and Modeling Techniques

Abstract

Gathering and analyzing Cyber-Physical System

requirements pose unique challenges to the

requirements engineering community - a

perspective that is sensitive to the scoping and

interplay between the cyber, physical and

behavioral aspects of the system. A model driven

requirements analysis approach using various

categories of models served as a useful aid in

discovering, understanding, and analyzing certain

classes of requirements about crucial aspects of

the cyber-physical systems which were otherwise

latent.

©2008 Regents of the University of Minnesota. All rights reserved.

Anitha Murugesan, Lian Duan, Sanjai Rayadurgam, Mats Heimdahl Department of Computer Science and Engineering, University of Minnesota-Twin Cities

Next Steps

Create assurance cases and perform reasoning to demonstrate GPCA safety

Create a catalog of requirements patterns to capture system dynamics in the

physical domain. (Control System Modeling)

Develop guidelines for scoping a system and flow-down of requirements to

decomposed system components. (Architectural Modeling)

Build a catalog of solutions for various behavioral modeling problems and

build a repertoire of modeling patterns. (Stateflow Modeling)

Develop a standard set of reference artifacts for demonstrating the safety of

CPS systems.

Requirements to address the complexity in the continuous domain.

Requirements categories related to continual behavior of the system

Precise scoping of the system.

Understanding requirements and system decomposition.

Understanding modes and features interactions

Analyzing system behavior during mode transitions.

Modeling for change/expansion/contraction.

Control System Modeling Architectural Modeling

• Examine different architectures.

• Visualize hierarchical system decomposition

• Precisely fixed system boundary to express

requirements.

Stateflow Modeling

• Evaluate various control strategies

• Explore system responses in its

intended environment.

• Identified categories of requirements:

• Accuracy

• Rise / Drop Time

• Rate of Change

• Overshoot (Max deviation)

• Settling Time

• Cumulative Error

• Complement requirements to

understand the dynamic behavior of

complex systems.

• Analyze mode-logic requirements

and behavior.

• Identified mode-logic modeling

patterns that are flexible for

changes.

• Sequential vs Parallel structure of

mode logic

• Complex feature behaviors

Figure: Generic Patient Controlled Analgesia Infusion Pump System.

Figure: Infusion Pump Plant Model.

Figure: Infusion System – GPCA Device Architecture Model

Figure: GPCA Software Model Overview

Figure: GPCA Software Model – Infusion Manager Subsystem

Funded by CNS-0931931

and CNS-1035715