Energy minimization is one of the major challenges in the design of sensor nodes and networks, since most of the sensors operate on limited power supply from batteries. Reducing energy is not trivial for a sensor node, which is a complex system with hybrid hardware components and software applications. The problem becomes further complicated at the sensor network level. In this presentation, we will introduce two of the most effective power management techniques at the system-level. They are designed to significantly reduce the system energy dissipation by dynamically slow-down or turn off the system components that are not used or under-utilized. At the same time, the performance constraints in sensor networks, such as data processing deadline, communication latency and quality-of-service, are still met by the optimization algorithms.