The document explores event-triggered real-time scheduling of control tasks to ensure semi-global asymptotic stability. It investigates scheduling algorithms that preempt running tasks to execute control tasks when an error exceeds a state norm threshold. This simple execution rule results in control task inter-execution times always being lower bounded, provided sensor state reading time is sufficiently small. An example on a linear control system illustrates this lower boundedness and that the estimated values do not overestimate simulated values by more than a factor of 3.
The document explores event-triggered real-time scheduling of control tasks to ensure semi-global asymptotic stability. It investigates scheduling algorithms that preempt running tasks to execute control tasks when an error exceeds a state norm threshold. This simple execution rule results in control task inter-execution times always being lower bounded, provided sensor state reading time is sufficiently small. An example on a linear control system illustrates this lower boundedness and that the estimated values do not overestimate simulated values by more than a factor of 3.
The document explores event-triggered real-time scheduling of control tasks to ensure semi-global asymptotic stability. It investigates scheduling algorithms that preempt running tasks to execute control tasks when an error exceeds a state norm threshold. This simple execution rule results in control task inter-execution times always being lower bounded, provided sensor state reading time is sufficiently small. An example on a linear control system illustrates this lower boundedness and that the estimated values do not overestimate simulated values by more than a factor of 3.
Event-triggered Real -Time scheduling of Stabilizing Control Tasks
(REVIEW)
Summary: The author is trying to explore real time scheduling of
tasks based on event rather than periodic scheduling, ensuring semi- global asymptotical stability for control tasks. Now a days embedded microprocessors have to work under multitasking environment. Scheduler decides which task to be executed, when there are more tasks are waiting for processor. The tasks can be control specific tasks such as sampling of states and inputs, computing control law and sending actuator signal or non- control specific tasks such relay of information packets over the network. Normal approach to digital control is to use periodic sampling in time also called Riemann sampling or Lebesgue sampling also called as event based sampling. The author is trying to investigate the real time scheduling algorithm based on event triggering that pre-empts running tasks, to execute the control task whenever a certain error becomes large when compared with state norm. Previously, real time scheduling of control tasks to achieve better control performance was done by providing more processor time to control tasks or scheduling control tasks more frequently than other tasks. These scheduling strategies are obtained through optimization algorithms in order to maximize performance criterion. The main contribution of the paper is that the simple execution rule results in a sequence of inter execution times for the control task which is always lower bounded provided the time required to read the state from the sensors is sufficiently small. An example is provided to illustrate the results on linear control system where lower boundedness for inter execution time is provided, further it is shown that estimated value does not overestimate the values obtained by simulation by a factor of 3.