Design Flow of Dynamically-Allocated Data Types in Embedded Applications Based on Elitist Evolutionary Computation Optimization

New multimedia embedded applications are increasingly dynamic, and rely on Dynamically-allocated Data Types (DDTs) to store their data. The optimization of DDTs for each target embedded system is a time-consuming process due to the large design space of possible DDTs implementa-tions. Thus, suitable exploration methods for embedded de-sign metrics (memory accesses, memory usage and power consumption) need to be developed. This paper presents a design flow to tackle the optimization of DDTs in multime-dia applications. By profiling of the original desktop ap-plication and using evolutionary algorithms, the proposed approach is able to find solutions 1584 × faster than other state-of-the-art heuristics in an automated way. Moreover, we study the use of elitist Multi-Objective Evolutionary Al-gorithms (MOEAs) to explore DDT implementations, which offer 75 % more optimal solutions to the system designer for the implementation of the final embedded application. To this end, we analyze the quality of the solutions by compar-ing three MOEAS and other optimization heuristics. Our results in two object-oriented multimedia embedded appli-cations show that elitist MOEAs (NSGA-II and SPEA2) of-fer better solutions than simple non-elitist schemes (VEGA) and alternative well-known optimization heuristics. 1