This thesis focuses on addressing four research problems in designing embedded streaming systems. Embedded streaming systems are those systems thatprocess a stream of input data coming from the... Show moreThis thesis focuses on addressing four research problems in designing embedded streaming systems. Embedded streaming systems are those systems thatprocess a stream of input data coming from the environment and generate a stream of output data going into the environment. For many embeddedstreaming systems, the timing is a critical design requirement, in which the correct behavior depends on both the correctness of output data and on the time at which the data is produced. An embedded streaming system subjected to such a timing requirement is called a real-time system. Some examples of real-time embedded streaming systems can be found in various autonomous mobile systems, such as planes, self-driving cars, and drones. To handle the tight timing requirements of such real-time embedded streaming systems, modern embedded systems have been equipped with hardware platforms, the so-called Multi-Processor Systems-on-Chip (MPSoC), that contain multiple processors, memories, interconnections, and other hardware peripherals on a single chip, to benefit from parallel execution. To efficiently exploit the computational capacity of an MPSoC platform, a streaming application which is going to be executed on the MPSoC platform must be expressed primarily in a parallel fashion, i.e., the application is represented as a set of parallel executing and communicating tasks. Then, the main challenge is how to schedule the tasks spatially, i.e., task mapping, and temporally, i.e., task scheduling, on the MPSoC platform such that all timing requirements are satisfied while making efficient utilization of available resources (e.g, processors, memory, energy, etc.) on the platform. Another challenge is how to implement and run the mapped and scheduled application tasks on the MPSoC platform. This thesis proposes several techniques to address the aforementioned two challenges. Show less
In real-time systems, the application's behavior has to be predictable at compile-time to guarantee timing constraints. However, modern streaming applications which exhibit adaptive behavior due to... Show moreIn real-time systems, the application's behavior has to be predictable at compile-time to guarantee timing constraints. However, modern streaming applications which exhibit adaptive behavior due to mode switching at run-time, may degrade system predictability due to unknown behavior of the application during mode transitions. Therefore, proper temporal analysis during mode transitions is imperative to preserve system predictability. To this end, in this paper, we initially introduce mode-aware data flow (MADF) which is our new predictable model of computation to efficiently capture the behavior of adaptive streaming applications. Then, as an important part of the operational semantics of MADF, we propose the maximum-overlap offset which is our novel protocol for mode transitions. The main advantage of this transition protocol is that, in contrast to self-timed transition protocols, it avoids timing interference between modes upon mode transitions. As a result, any mode transition can be analyzed independently from the mode transitions that occurred in the past. Based on this transition protocol, we propose a hard real-time analysis as well to guarantee timing constraints by avoiding processor overloading during mode transitions. Therefore, using this protocol, we can derive a lower bound and an upper bound on the earliest starting time of the tasks in the new mode during mode transitions in such a way that hard real-time constraints are respected. Show less
