An evolutionary cellular automaton for minimizing energy consumption in network on chip-based soft real-time embedded systems

Embedded multimedia systems with multi-application scenarios increase the number and computational power of processing elements. To satisfy this growing demand, a network on chip is used due to lower message latency, better scalability and more reliable performance predictability. Today the focus is not on the implementation of one single application on the many-core platforms, but using a single-die that support multi-application scenarios, and due to that the mapping strategies that map the application tasks on the chip should minimize the energy consumption of multiple parallel applications under timing constraints. We used a new paradigm called Evolutionary cellular automata offering diversity point of views called transition rules which are evolved in time progress in order to optimize an application's mapping.