Content, progress and pedagogy of the module

An embedded system is defined as an electronic system which is based on a computer, but the system is not in itself a computer, e.g., like a PC. According to this definition, an average person is interacting with hundreds of embedded systems on a daily basis, typically in terms of audio/video applications, wireless/mobile communication, gaming consoles, household machines, automotive and medical devices, as well as avionic and satellite based systems. In most cases, the computer embedded in such devices is conducting some kind of signal processing, i.e., an analogue signal is registered by a sensor and sampled, and next the signal is either analyzed or modified digitally by software executing on the computer. Eventually the resulting signal is finally re-converted back to the analogue domain. An interesting feature of this overall process is that in most cases it must be conduced in hard real-time, i.e., the processing must be completed within a predefined and fixed time interval. Otherwise, the system will fail, potentially leading to hazardous situations. Taking the outset in a real-life problem/application, the purpose of this project module is to specify, design, simulate, implement, test and document (part of) an embedded real-time signal processing system. In this context, the algorithm(s) which are to perform the signal processing have to be developed, simulated/evaluated (preferably using C or Matlab) and optimized, and next compiled into an executable code which can run in real-time on a programmable digital signal processor. The overall design parameters may include, but are not limited to execution time, code size, numerical robustness, and eventually energy consumption. Primarily, the project will focus on the signal processing theories and algorithms, as well as the development of optimal source- and object codes using commercially available development boards/tools, thus excluding the design and implementation of user-specific hardware.

Learning objectives

Knowledge

• Must have knowledge about the building blocks used in a generic embedded real-time digital signal processing (DSP) system, their mutual interaction and interfaces, as well as relevant performance parameters.
• Must have knowledge about theories and methods used to design numerically robust and resource optimal DSP algorithms suitable for being executed real-time on programmable digital signal processors.

Skills

• Must be able to analyze a technical problem which naturally finds its solution in terms of real-time digital signal processing. Secondly, to formulate a set of specifications for the algorithms to be developed, and possibly also for the hardware/software platform to be used.
• Must be able to apply various methods to design, simulate, and evaluate DSP algorithms according to the specifications for functionality and numerical properties. C or Matlab are candidates for executable specifications and for simulation purposes.
• Must be able to analyze DSP algorithms from a computational complexity, structural, and data flow oriented point of view in order to specify architectural requirements for a software programmable target platform.
• Must be able to apply design tools, such as C compilers (eventually using in-line assembly language), in order to synthesize and optimize real-time executable code for DSP algorithms.
• Must be able to evaluate 1) an overall system solution, and 2) the design methods applied to derive the solution. This must be done in terms of relevant metrics such as execution time, memory usage, numerical robustness, and energy consumption. Secondly, from a micro-computer architectural point of view, the students must be able to evaluate the match between algorithms and architectures.
• Must be able to communicate the above mentioned knowledge and skills (using the terminology of the domain), both orally and in a written report.

Competences

• Must be able to identify, design, implement, and evaluate a viable solution for an embedded real-time signal processing system in a real-life context.
• Must be able to plan, structure, and conduct a project within the scientific subject of this project module.

Type of instruction

Academically supervised student-governed problem oriented project work.

Lectures together with teacher/supervisor guided self-studies and/or mini projects.

Exam

Exams