Section: New Results

Data-oriented scheduling approaches

We consider the scheduling problem of tasks using an inter-task communication model based on a circular buffer, which eases the data consistency between tasks [13], [12]. The tasks are scheduled on one processor by a fixed priority preemptive scheduling algorithm and they have implicit deadlines. We provide a formal method calculating the optimal size for each of the buffers while ensuring data consistency, i.e., it is required that a buffer slot is accessed for reading the input data. This later slot will never be used by the producer task to write new data before the execution completion of the instances of all consumers that are currently reading from this slot. As a second contribution, we provide an analytical characterization of the temporal validity and reachability properties of the data flowing in between communicating tasks. These two properties are characterized by considering both tasks execution and data propagation orders. Moreover, we assume that a task instance reads all its inputs data at its activation time and writes back the output data at the completion time where this data becomes immediately available for consumption. Given that, they may be several data samples available in the buffer, we say that a data sample is fresh or temporal valid if, since the time instant it is produced, its producer has not completed another execution. Given that, we use buffers whose size may be larger than one, it is obvious that the consumer task will not implicitly know which data is temporally valid. In order to use the data that reflects the current status of the system environment (valid data), we introduce a novel parameter; the sub-sampling rate used within two scheduling algorithms. These scheduling algorithms ensure the data consistency and temporal validity, while deadlines are met.