Contents I. Introduction
New challenges in embedded systems design include uncertainty in design parameters due to process, voltage and temperature (PVT) variations among others [1]. The uncertainty in design parameters adversely affects the accuracy of the model-based estimations of the design attributes of interest, thereby affecting the quality of the final solutions. Thus, increased uncertainties in design parameters undermine the accuracy and effectiveness of the mapping of applications to architecture platforms in embedded systems design. Therefore, researchers started to address the issue of uncertainties in design parameters during the process of designing embedded systems. For example, several recent studies presented solutions to model and handle uncertainty [2]–[4]. The work in [2] proposed uncertainty models and a mapping algorithm for robust embedded systems that used software components only. Reliability was the only objective considered in the op-timization process. The study in [3] proposed an uncertainty-aware reliability model for the design space exploration of embedded systems with consideration of correlations between components. Similarly to [2], reliability was considered as the only uncertain design attribute. The work in [4] used the uncertainty and reliability models from [2] and proposed a solution to the mapping problem similar to that from [5], [6], but formulated as a multi-objective problem considering reliability, execution time, and energy consumption together.
