Real-Time Scheduling of Sensor-Based Control Systems

Enhancing adaptivity via standard dynamic scheduling middleware

“…A distributable thread comprises one or more scheduling segments, which is a code sequence whose execution is scheduled according to a set of application-specified scheduling parameters, e.g., worstcase execution time, deadline, and criticality of a realtime operation is used by the Maximum Urgency First (MUF) [4] scheduling strategy. These parameters can be associated with a segment encompassing that operation on a particular endsystem, e.g., as shown for segment B in Figure 1.…”

“…Various scheduling disciplines exist that require different scheduling parameters, such as MLF [4], EDF [12], MUF [4], or RMS+MLF [13]. One or more of these scheduling disciplines (or any other discipline the system developer chooses) may be used by an open DRE system to fulfill its scheduling requirements.…”

“…The analysis below features a comparison of two canonical scheduling strategies, the hybrid static/dynamic Maximum Urgency First (MUF) [4] strategy, which assigns operations to strict priority lanes according to their criticality and then schedules them dynamically within each lane according to laxity, and the static Rate Monotonic Scheduling (RMS) [12] Operation cancellation. Figure 7 shows the effects of canceling non-critical operations in the MUF hybrid static/dynamic scheduling strategy in conditions of CPU overload.…”

Multiparadigm scheduling for distributed real-time embedded computing

“…Designing DRE systems that implement their required capabilities, are dependable, and are parsimonious in their use of limited computing resources is hard; building them on time and within budget is even harder. A particularly essential task is supporting the quality of service (QoS) demands of mission-critical DRE systems that possess a mix of hard and soft real-time requirements, such as avionics mission computing systems [1], mission-critical distributed audio/video processing [2], [3], and real-time robotic systems [4].…”

Enhancing adaptivity via standard dynamic scheduling middleware

“…For an important emerging class of open DRE systems (such as adaptive audio/video streaming [2], collaborative mission replanning [3], and robotics applications designed for close interaction with their environments [4]), however, it is often not possible to know the entire set of application tasks that will run on the system, the loads they will impose on system resources in response to a dynamically changing environment, or the order in which the tasks will execute. This dynamism can occur because the number of combinations in which application tasks can be mapped to system resources is too large to compute efficiently or because task run-time behaviors are simply too variable to predict accurately.…”