Embedded Systems Design
Scott DeCota
Modern Operating Systems

Embedded systems are all around us. They run the engine, brakes, seatbelt, airbag, and audio system in your car. They digitally encode your voice and construct a radio signal to send it from your cell phone to a base station. They control your microwave, dishwasher, and DVD player. They command robots on a factory floor, power generation in a plant, processes in a chemical plant, and the traffic lights in the city. Embedded operating systems are growing increasingly common in everyday appliances and other devices. These operating systems provide application interfaces for a wide range of different applications. The challenge for developers is to determine which operating system or programming language to use in a particular device or project, and how to utilize the parallelism and concurrency capabilities of the hardware and programming language chosen for a project to design a system of adequate timing precision and efficiency. Developers working with embedded systems need to be extremely diligent in regards to timing and optimization. Embedded systems are frequently interacting with the physical environment in which multiple sources of data may be interacting with the system simultaneously. Developers should have a solid understanding of the parallelism and concurrency capabilities of the hardware for a selected project. These features can occur in different forms that can greatly affect the timing of the program execution. Developers have a few choices when selecting a language to design their system logic. Embedded systems can be written in Assembly, Java, and C, with the latter two being more prevalent in current devices. However let’s not forget that Assembly is still there underneath a majority of these higher level languages. Embedded systems written in C can be singular processes with multiple logic threads within, or a more general purpose embedded OS, such as embedded Linux and Windows CE. The choice of language often will be dictated by the requirements of the project and the hardware selected to fulfill the physical design. Routers and POS systems are some of the more common embedded systems that run on more general purpose operating systems like Linux and Windows CE. DD-WRT is an example of an embedded version of Linux, specifically tasked with network traffic management and other routing capabilities. Physical processes are made up of multiple events happening all at once, as opposed to traditional software processes which are deeply sequential. Measuring and controlling the dynamics of these events by taking actions that influence the process is the central purpose of embedded systems. Many of the hurdles in designing and analyzing embedded systems are inherent in the need to utilize programming languages that are based on sequential semantics to function in a more parallel environment. There are a variety of controllers and processing units that can be used in design and construction of embedded systems and therefore may have vastly different instruction sets. In order to build efficient systems developers need to evaluate our applications needs and select an appropriate set of hardware. Additionally developers need to design the program logic to address the appropriate instruction set for the selected hardware. Modeling of physical systems can give developers valuable insight into how the system handles input and output as well as calculations. Modeling allows developers to test and debug logic through imitation and virtualization. Interaction with the physical world requires that developers understand how to capture and process different input streams and generate output streams. Each type of processor or controller board has its own set of peripheral interfaces of which designers need to be aware. There is typically a great deal of data being captured from the sensors connected to the system. Some of this data is unwanted input of the same signal type, otherwise called signal noise, but not the specific variance being collected. Signal conditioning is used to minimize the amount of noise being captured with the wanted data. Specific algorithms are utilized to strip most, or all of the unwanted signal from the input stream. This type of optimization is extremely important to designing stable efficient systems. Concurrency within the logic of the system is essential since often embedded systems need to process multiple input streams and simultaneously control multiple output devices. Since these systems are often interacting with and processing input from the real world, program efficiency is not just about completing tasks as quickly as possible, but completing tasks at the correct time for the system to function properly. For example you wouldn’t want the controller in your engine to fire the spark plugs sooner or later than it should. In order for the engine to run properly, the spark plugs need to be fired at the exact correct time. Therefore developers need to understand the timing capabilities of the processors and control boards selected in order to avoid selecting equipment that would prevent the system from achieving the required level of timing precision. Furthermore developers need to consider the memory architecture of the controller and the memory model of the chosen programming language. Developers need to be aware of the address spaces of volatile and non-volatile memory, and they need to implement logic for garbage collection and memory fragmentation. Total memory availability and caching are important factors as well when creating embedded systems. To create functional and efficient embedded systems developers need to have an in-depth understanding of memory management, input/output streams, signal processing and conditioning, and process control. All of these, along with selection of appropriate hardware and programming languages are integral parts to designing a functional embedded system.

Works Cited 1. https://chess.eecs.berkeley.edu/eecs149/ Introduction to Embedded Systems (Fall 2013) 2. Wikipedia “Embedded Systems” 2014 3. Design Automation for Embedded Systems – Springer Link 2012 4. Embedded.com 2009 5. IBM DeveloperWorks 2014 6. E. A. Lee and S. A. Seshia, Introduction to Embedded Systems - A Cyber-Physical Systems Approach, LeeSeshia.org, 2011. 7. Industrial Controls and Manufacturing, Edward W. Kamen, 1999