Target assignment for robotic networks: asymptotic performance under limited communication