Application area: Healthcare, Automotive

Researcher in charge: Sajid Mohamed

Supervisor: Prof.dr.ir. Twan Basten, Dr. Dip Goswami

Host: Eindhoven University of Technology, The Netherlands

Secondments: Inchron GmbH, Germany and Ericsson, Sweden

 The main objective of this research is to develop a state-of-the-art design method for applications running on integrated architecture. The researcher will consider a setting where streaming applications share the platform with the control applications. The focus will be on tradeoff analysis considering given resource constraints (computation and communication). Such a state-of-the-art design method will require: 1) a scientific method to integrate the architecture-level timing information into the high-level modelling, 2) a method for accurate system-timing analysis and 3) scalable formal verification and synthesis tools or methods.

A preliminary design space exploration reveals several advantages in using max-plus algebra as a semantic modelling and analysis framework for our objective. Max-plus algebra has been used as a foundation for solving industrial design problems. Max-plus semantics can inherently integrate timing constraints and can help to scale down our model for computation. Max-plus algebra is also known to be used in solving problems in model predictive control and adaptive control. This research will focus on design space exploration and leveraging the advantages of max-plus algebra for an efficient design methodology.

Expected contributions:

This research aims to develop new analysis and design methods that are capable of taking into account various tradeoff possibilities between QoC (control loops) and QoS (streaming applications). This will require novel platform-aware control algorithms that are able to adapt their sampling rate when necessary. In particular, the research will focus on combining the theoretical foundation of multi-rate controllers with synchronous dataflow/synchronous-aware dataflow and max-plus formalism to address the co-design problem. Initially, the researcher will implement and verify the developed formalism using the xCPS machine (figure 1). Then the researcher will develop the libraries for max-plus functions required for reusability of this formalism. These libraries and the co-design methodology will then be integrated into industry standard tool chain developed under oCPS. The applicability will be illustrated considering applications from healthcare (e.g., the interventional x-ray system) and automotive (e.g., advanced driver assistance system).

machine
Figure 1: xCPS machine (xcps.info)

The research topic of this ESR is tightly connected to the research lines RL1, RL2, and RL3. We aim for a synthesizable co-design method for control and streaming applications (RL1), considering tradeoffs between QoC and QoS under given constraints (RL2) and platform (RL3).