Self-adaptation of Cyber Physical Systems: Flexible HW/SW computing

Eduardo de la Torre

Embedded systems with stringent computing and low power consumption needs within a restricted amount of computing resources may have, on top of this, the need of adapting to different conditions. Many systems may benefit from self-adaptation at run-time. This autonomous adaptation may be originated from different needs: changes in the mission of the embedded control system, changes in the environment, or changes in the control system itself (including faults and how to self-recover from them). In this talk, a review of different techniques for self-adaptation is one. Topics such as self-awareness, scalable performance, dynamic HW and SW adaptation for scalabaility, graceful degradation and fault mitigation techniques for self-healing, will be addressed. The aim is to offer different techniques that may be used in order to increase the autonomy and adaptation capabilities of embedded systems, in general, and of cyber-physical systems, in particular. HW and SW adaptation are a very interesting feature for the new reconfigurable MPSoCs solutions,like the Xilinx Zynq or Zynq UltraScale+ devices, which include a variety of computing resources such as FPGAs, multi-core systems or embedded GPUs.

Lecture Slides