Modern embedded systems are complex systems as they support the
execution of concurrent tasks having multiple modes with variable
application performance metrics (e.g., accuracy), implemented on
platforms comprising of heterogeneous computing elements and various
sensors. These sensors monitor various environmental factors which
together constitute a context. As the system is expected to adapt to
variations in context, we name these systems as Context-aware Adaptive
Embedded Systems (CAES). We propose a DSE approach that helps select the
right platform components by evaluating various application performance
metrics while simultaneously considering task modes and the supported
context. Further, the flow can generate the specifications of a run-time
controller that adapts the system to changes in the context. We
demonstrate the effectiveness of the proposed design flow through a case
study on a real embedded system named MAVI. Our approach prunes upto
99.74% of the original design space and also enables generation of the
specifications of the run-time controller for the MAVI system.