In resource-limited wireless sensor network, significant energy savings and longer life time of sensors can be achieved by having an effective energy management forcing sensors to sleep as long as they can. In this paper, we propose a Route-Oriented Sleep (ROS) approach which allows sensors to sleep based on routing information and the level of network's activities. The scheme lets sensors switch their roles in the network according to the critical probability of the percolate theory without losing their sensing and communication connection coverage. This technique takes advantages of routing decisions and implements an effective sleep scheme at the network level, which allows some of sensors along the route path to sleep and wake up when they are a part of specific route. This produces balanced and low energy consumption for each node. ROS does not only achieve significant energy savings, but also reduces the communication interferences in the data transmissions, thus reduces the latency of data transmissions. The simulation results show that the proposed ROS approach significantly saves the sensor energy consumption more than 50%.