Determining sampling rate for computational task graphs running under the macrodataflow model of execution

Sampling Rate (SR) for a Computational Task Graph (CTG) can be determined by analysing the frequency at which the task nodes can work, the asynchrony of the data input at the nodes and the length of the feedback loops. If a CTG is fed by input data faster than its SR capacity then, some loss of data will occur. In this paper we quantify the different data loss situations that may arise in a CTG and define an approach for determining an input data SR such that no data loss occurs.