Embedded Systems Handbook

Networked Embedded Systems

Richard Zurawski, Richard Zurawski

• Éditeur: CRC Press
• Année: 2017
• ISBN:
  9781351834186
• Édition: 1
• Sujet: Technology & Engineering
• Sous-sujet: Electrical Engineering & Telecommunications

Considered a standard industry resource, the Embedded Systems Handbook provided researchers and technicians with the authoritative information needed to launch a wealth of diverse applications, including those in automotive electronics, industrial automated systems, and building automation and control. Now a new resource is required to report on current developments and provide a technical reference for those looking to move the field forward yet again. Divided into two volumes to accommodate this growth, the Embedded Systems Handbook, Second Edition presents a comprehensive view on this area of computer engineering with a currently appropriate emphasis on developments in networking and applications. Those experts directly involved in the creation and evolution of the ideas and technologies presented offer tutorials, research surveys, and technology overviews that explore cutting-edge developments and deployments and identify potential trends.

This second self-contained volume of the handbook, Network Embedded Systems, focuses on select application areas. It covers automotive field, industrial automation, building automation, and wireless sensor networks. This volume highlights implementations in fast-evolving areas which have not received proper coverage in other publications. Reflecting the unique functional requirements of different application areas, the contributors discuss inter-node communication aspects in the context of specific applications of networked embedded systems.

Those looking for guidance on preliminary design of embedded systems should consult the first volume: Embedded Systems Design and Verification.

I
Network Embedded Systems: An Introduction

1 Networked Embedded Systems: An Overview Richard Zurawski

Networking of Embedded Systems

Automotive Networked Embedded Systems

Networks Embedded Systems in Industrial Automation

Wireless Sensor Networks

Networked Embedded Systems in Building Automation

Concluding Remarks

2 Middleware Design and Implementation for Networked Embedded Venkita Subramonian and Christopher D. Gill

Introduction

Middleware Solution Space

ORB Middleware for Networked Embedded Systems: A Case Study

Design Recommendations and Trade-Offs

Related Work

Concluding Remarks

Acknowledgments

1
Networked Embedded Systems: An Overview

Richard Zurawski

ISA Group

1.1 Networking of Embedded Systems

1.2 Automotive Networked Embedded Systems

1.3 Networks Embedded Systems in Industrial Automation

1.4 Wireless Sensor Networks

1.5 Networked Embedded Systems in Building Automation

1.6 Concluding Remarks

References

1.1 Networking of Embedded Systems

The last two decades have witnessed a remarkable evolution of embedded systems from being assembled from discrete components on printed circuit boards, although, they still are, to systems being assembled from IP components “dropped” on to silicon of the system on a chip. Systems on a chip offer a potential for embedding complex functionalities, and to meet demanding performance requirements of applications such as DSP, network, and multimedia processors. Another phase in this evolution, already in progress, is the emergence of distributed embedded systems; frequently termed as networked embedded systems, where the word “networked” signifies the importance of the networking infrastructure and communication protocol. A networked embedded system is a collection of spatially and functionally distributed embedded nodes interconnected by means of wireline and/or wireless communication infrastructure and protocols, interacting with the environment (via a sen- sor/actuator elements) and each other, and, possibly, a master node performing some control and coordination functions, to coordinate computing and communication to achieve certain goal(s). The networked embedded systems appear in a variety of application domains such as automotive, train, aircraft, office building, and industrial areas—primarily for monitoring and control, environment monitoring, and, in future, control, as well.

There have been various reasons for the emergence of networked embedded systems, influenced largely by their application domains. The benefits of using distributed systems and an evolutionary need to replace point-to-point wiring connections in these systems by a single bus are some of the most important ones.

The advances in design of embedded systems, tools availability, and falling fabrication costs of semiconductor devices and systems have allowed for infusion of intelligence into the field devices such as sensors and actuators. The controllers used with these devices provide typically on-chip signal conversion, data and signal processing, and communication functions. The increased functionality, processing, and communication capabilities of controllers have been largely instrumental in the emergence of a widespread trend for networking of field devices around specialized networks, frequently referred to as field area networks.

The field area networks, or fieldbuses [1] (fieldbus is, in general, a digital, two-way, multidrop communication link) as commonly referred to, are, in general, the networks connecting field devices such as sensors and actuators with field controllers (for instance, programmable logic controllers (PLCs) in industrial automation, or electronic control units (ECUs) in automotive applications), as well as man-machine interfaces; for instance, dashboard displays in cars.

In general, the benefits of using those specialized networks are numerous, including increased flexibility attained through the combination of embedded hardware and software, improved system performance, and ease of system installation, upgrade, and maintenance. Specifically, in automotive and aircraft applications, for instance, they allow for a replacement of mechanical, hydraulic, and pneumatic systems by mechatronic systems, where mechanical or hydraulic components are typically confined to the end-effectors; just to mention this two different application areas.

Unlike local area networks (LANs), due to the nature of communication requirements imposed by applications, field area networks, by contrast, tend to have low data rates, small size of data packets, and typically require real-time capabilities which mandate determinism of data transfer. However, data rates above 10 Mbit/s, typical of LANs, have become a commonplace in field area networks.

The specialized networks tend to support various communication media like twisted pair cables, fiber-optic channels, power line communication, radio frequency channels, infrared connections, etc. Based on the physical media employed by the networks, they can be in general divided into three main groups, namely, wireline-based networks using media such as twisted pair cables, fiber-optic channels (in hazardous environments like chemical and petrochemical plants), and power lines (in building automation); wirelss networks supporting radio frequency channels and infrared connections; and hybrid networks, with wireline extended by wireless links [2].

Although the use of wireline-based field area networks is dominant, the wireless technology offers a range of incentives in a number of application areas. In industrial automation, for instance, wireless device (sensor/actuator) networks can provide a support for mobile operation required in case of mobile robots, monitoring and control of equipment in hazardous and difficult to access environments, etc. In a wireless sensor/actuator network, stations may interact with each other on a peer-to-peer basis, and with a base station. The base station may have its transceiver attached to a cable of a (wireline) field area network, giving rise to a hybrid wireless-wireline system [2]. A separate category is the wireless sensor networks, envisaged to be largely used for monitoring purposes.

The variety of application domains impose different functional and nonfunctional requirements on to the operation of networked embedded systems. Most of them are required to operate in a reactive way; for instance, systems used for control purposes. With that comes the requirement for real-time operation, in which systems are required to respond within a predefined period, mandated by the dynamics of the process under control. A response, in general, may be periodic to control a specific physical quantity by regulating dedicated end-effector(s), or aperiodic arising from unscheduled events such as out-of-bounds state of a physical parameter or any other kind of abnormal conditions. Broadly speaking, systems which can tolerate a delay in response are called soft real-time systems; in contrast, hard real-time systems require deterministic response to avoid changes in the system dynamics which potentially may have negative impact on the process under control, and as a result may lead to economic losses or cause injury to human operators. Representative examples of systems imposing hard real-time requirement on their operation are Fly-by-Wire in aircraft control, Steer-by-Wire in automotive applications, to mention some.

The need to guarantee a deterministic response mandates using appropriate scheduling schemes, which are frequently implemented in application domain-specific real-time operating systems or frequently custom designed “bare-bone” real-time executives.

The networked embedded systems used in safety-critical applications such as Fly-by-Wire and Steer-by-Wire require a high level of dependability to ensure that a system failure does not lead to a state in which human life, property, or environment are endangered. The dependability issue is critical for technology deployment; various solutions are discussed in this chapter in the context of automotive applications.

As opposed to applications mandating hard real-time operation, such as the majority of industrial automation controls or safety-critical automotive control applications, building automation control systems, for instance, seldom have a need for hard real-time communication; the timing requirements are much more relaxed. The building automation systems tend to have a hierarchical network structure and typically implement all seven layers of the ISO/OSI reference model [3]. In case of field area networks employed in industrial automation, for instance, there is little need for the routing functionality and end-to-end control. As a consequence, typically, only the layers 1 (physical layer), 2 (data link layer, including implicitly the medium access control layer), and 7 (application layer, which covers also user layer) are used in those networks.

This diversity of requirements imposed by different application domains (soft/hard real-time, safety critical, network topology, etc.) necessitated different solutions, and using different protocols based on different operation principles. This has resulted in a plethora of networks developed for different application domains.

Design methods for networked embedded systems fall into the general category of system-level design. They include three aspects, namely, node design (covered extensively in Section I of the book), network architecture design, and timing analysis of the whole system. The network architecture design involves a number of activities. One of them is selection of an appropriate communication protocol and communication medium. A safety-critical application will employ a protocol based on Time Division Multiple Access (TDMA) medium access control to ensure deterministic access to the medium. For an application in building automation and control, the choice of the communication medium may be the power line wires in the existing building or dedicated twisted pair wires in a new construction. The topology of the network heavily depends on the application area. In industrial automated systems, the prevalent topology is the bus. Building network may have a complex topology with many logical domains. Configuration of the communication protocol, among other things, involves allocation to the communication nodes priorities in the priority busses, or slots in the TDMA-based protocols, for instance. The timing analysis aims at obtaining actual times for the chosen architecture. That involves task execution time measures such as worst-case execution time (WCET), best-case execution time (BCET), and average execution time; response time of a task from invocation to completion; end-to-end delay; and jitter, or variation in execution time of a task, for instance. In the end, the whole system has to be schedulable to guarantee that deadlines of all distributed tasks communicating over the network will be met in all operational conditions the system is anticipated to be subjected to. As an example, let us consider a simple control loop comprising a sensing node with a single application task dedicated to sensing, an actuator node processing data received from the sensing node, and generating control value delivered to an actuator over a dedicated link. The composite time of data processing (WCET) and transmission (worst-case response time) has to be shorter or equal to the maximum time allowed by the process dynamics under control. In case of other nodes connected to the shared communication network and forming similar control loops, a contention for the medium access may arise to be remedied for safety-critical and hard real-time systems by adopting a fixed transmission schedule as in the case of the time-triggered TDMA-based protocols, for instance. The schedulability analysis is to determine if the worst-case response time for all those composite tasks forming control loops is less then or equal to the deadline.

1.2 Automotive Networked Embedded Systems

Trends for networking also emerged in the automotive electronic systems where the ECUs are networked by means of one of automotive-specific communication protocols for the purpose of controlling one of the vehicle functions; for instance, electronic engine control, antiloc...