Abstract: We present a new reward design for the deep reinforcement learning (DRL)-based routing, modulation and spectrum assignment in the elastic optical networks (EONs). The performance of the EONs ...