— We consider energy efficient packet forwarding with quality-of-service (QoS) guarantee for wireless sensor networks (WSNs). In most existing wireless network protocols, route planning and transmission rate setting are considered independently, however, we observe that energy consumption can be significantly reduced by combining these two aspects together when making forwarding decision. This is because this combination potentially increases one more degree of freedom (i.e., the temporal dimension) for the original routing problem. We are motivated to exploit this additional degree of freedom and devise forwarding protocol that conserves more energy than traditional algorithms. In particular, we cast the packet forwarding problem into the framework of dynamic programming. State space partition technique and state aggregation approximating architecture are introduced to produce an approximating optimal value function. Simulation experiments are carried out to evaluate the performan...