Marilyn Wolf

Industrial Internet of Things

Industry is a leading application domain for the Internet of Things. The promise of the technology is expected to revolutionize manufacturing as we know it and influence application domains like transport, health, and energy with the models that are emerging. In this chapter, we present the evolution of the Industrial Internet of Things (IIoT) through efforts in Europe and the USA, we review the related reference architectures, and we describe applications and challenges.

Processes and Operating Systems

Multiple processes may be executed on a single computer processing unit (CPU) to save hardware and, to some extent, to save energy; however, it is important to be careful to share the CPU so as to meet all the real-time deadlines. A large body of literature has been written on real-time scheduling, taking into account various conditions on execution and assumptions. Real-time operating systems must be designed to efficiently implement basic operations such as scheduling, interrupt handling, and interprocess communication. Several scheduling protocols can be used to guarantee basic real-time behavior. However, it may not be possible to use 100% of the CPU if it has to be guaranteed that all the deadlines will be met. Scheduling for dynamic voltage scaling tries to stretch the execution time of processes just to meet their deadline to take advantage of the lower operating voltages allowed by slower execution. Context switching and interrupt overhead may be important in heavily utilized systems. File systems implemented in flash memory must use specialized techniques to avoid wearing out the flash memory. Concurrent system verification builds and searches state machine models to determine whether desired properties are satisfied.

Hourglass-shaped architecture for model-based development of safe and secure cyber-physical systems: poster

Inspired by the hourglass-shaped architecture of Internet, we propose an approach to model-based development of networked cyber-physical systems (CPS) that is centered on the notion of a standardized CPS design specification language. The proposed design specification language can be used to build a CPS design specification model that can serve as a narrow interface between a set of platform-aware feedback controller design techniques and a set of runtime CPS computing platforms. As a result, this model-based development approach can support the goals of an integrated, cross-layer CPS design and development methodology, while still acknowledging the differences between the domain-specific skillset that control system engineers and embedded system engineers typically possess. The poster will outline a number of CPS-related safety and security concerns that the proposed hourglass-shaped architecture for networked CPS development must address. The poster will also document a number of requirements that any standardized CPS design specification language must meet in order to address the CPS-related safety and security concerns. The proposed approach is inspired by the hourglass-shaped architecture of Internet. The narrow waist of hourglass-shaped architecture suggests that there is less diversity of protocols at this layer of Internet [4]. Any application that can operate based on the services of IP layer can be deployed on the Internet, and any underlying technology that can transport bytes from one point to another according to IP services can be used in the Internet. Similarly, according to the proposed approach to the model-based development of networked CPS, a wide range of DSMLs (and associated analysis tools) can be utilized to develop a platform-aware feedback controller design, which is then specified using a standardized design specification language. The proposed feedback controller design can then be analyzed for mapping on to wide range of runtime CPS computing platforms by utilizing their corresponding DSMLs (and associated analysis tools). This approach can support the goals of an integrated CPS theory and development methodology while still taking into account the differences between the domain-specific skillset that control system engineers and embedded system engineers typically possess. However, this poster will outline a number of CPS-related safety and security concerns that the proposed hourglass-shaped architecture for networked CPS development must address. The poster will also document a number of requirements that any standardized CPS design specification language must meet in order to address the CPS-related safety and security concerns.

The IoT Landscape

This chapter introduces basic concepts in Internet of Things (IoT) systems. A survey of sample applications for IoT systems provides a basis for identifying system architectures. The characteristics of wireless networks and VLSI IoT devices are surveyed. Security and privacy are key design goals for IoT systems. An event-driven model provides a useful basis for analyzing IoT systems. The chapter closes with a survey of the remainder of the book.

Image Capture Systems and Algorithms

This chapter considers the design of cameras and all the processes that are required to capture perform the initial processing an image. We will concentrate in this chapter on algorithms that provide traditional photos, such as sharpening and compression. Imaging chain algorithms must be designed for efficiency. We measure efficiency along several axes: n n nExecution time. Cameras—both still and video—are real-time systems. We care about the rate at which we can capture, process, and store images. Algorithms must be designed to run fast. We are also concerned about variations in their execution time, which can require additional buffer memory that imposes other costs and limitations. n n nEnergy and power consumption. Energy and power are related but distinct concerns. Energy is important because most cameras are battery-powered; lower energy per consumption per image results in more images per battery charge. Energy-efficient algorithms and systems must avoid unnecessary or duplicative work. Power consumption—energy per unit time—is important in large part because of thermal requirements. Power consumption results in heat. Thermal power dissipation is the primary limitation on performance in high-performance computer systems [Wol17]. Heat generated in a camera can also affect sensor performance—most device and circuit noise increases with temperature, typically exponentially. n n nMemory bandwidth and capacity. Multimedia algorithms are memory-intensive. Memory and mass storage devices can absorb and produce data at limited rates. High memory access rates can limit system performance; it can also drive up energy and power consumption. We are also concerned with the total memory usage of an algorithm. Certain parts of the imaging pipeline, particularly those near the image sensor, provide only constrained amounts of memory. Sloppy use of buffer memory can, for example, limit the number of images in a burst.

Image and Video Enhancement

This chapter considers algorithms to enhance photos and still images. High-dynamic range algorithms, for example, generate a composite image from several images at different exposures. These algorithms are increasingly available on cameras, but they are not part of the traditional imaging chain. Many of the algorithms here require substantially more computation than was the case for the methods of Chap. 3. In the next chapter, we will take this development a step further to look at algorithms that do not produce images at all—they analyze images to produce succinct descriptions.

Image and Video Analysis

The human visual system is much more than a camera—most of the visual system is dedicated to analyzing the imagery captured by our eyes. We perceive the world both as scenic imagery and as our understanding of those scenes—people, objects, and places. Digital cameras have allowed us to move photography beyond imaging to image understanding. A camera does not need to take a picture—it can report on what it sees.

Security Testing IoT Systems

System implementations need to be tested for security, because implementation bugs provide an attack surface that can be exploited to penetrate the systems. In this chapter, we introduce testing for security for IoT systems and especially fuzz testing, which is a successful technique to identify vulnerabilities in systems and network protocols. We describe an example fuzzer for the industrial protocol Modbus.

Security and Safety

Safety is a critical requirement for IoT systems and services in numerous application domains, such as health, transportation, energy, and manufacturing. Security is a prerequisite of safety, because its violation leads to unsafe systems. In this chapter, we review security technologies and challenges for IoT systems, from the device level to the application and process level.

Event-Driven System Analysis

Event-driven models provide a rich basis for the analysis of IoT systems. This chapter introduces event-driven analysis methods that allow us to characterize key design parameters for IoT systems. After surveying related work, the chapter describes an example that motivates our work. A model of IoT networks includes communication links and hubs and a timewheel to model the temporal relationships between events. Event analysis allows us to derive characteristics of the network’s event population over time.