Applications running in heterogeneous MPSoCs, as multimedia and networking, normally contain a dynamic workload of tasks. This implies a varying number of tasks simultaneously running, with their number possibly exceeding the available resources. This may require the execution of task mapping at run time, to meet real-time constraints. Most works in the literature propose the use of static mapping [1][2], where the best placement of tasks is defined at design time. Consequently, such methods are not appropriate for dynamical workloads. Task migration [3][4] has also been used in heterogeneous MPSoCs to optimize the performance at run-time. It consists either in relocating tasks when a performance bottleneck is detected, or to distribute the workload more homogeneously among the MPSoC processors. Differently from task migration, dynamic mapping can insert new tasks into the system at run time.