In wake-up radio (WuR) enabled wireless sensor networks (WSNs), a node triggers a data communication at any time instant by sending a wake-up call (WuC) in an ondemand manner. Such wake-up operations eliminate idle listening and overhearing burden for energy consumption in duty-cycled WSNs. Although WuR exhibits its superiority for light traffic, it is inefficient to handle high traffic load in a network. This paper makes an effort towards improving the performance of WuR under diverse load conditions with a twofold contribution. We first propose three protocols that support variable traffic loads by enabling respectively clear channel assessment (CCA), backoff plus CCA, and adaptive WuC transmissions. These protocols provide various options for achieving reliable data transmission, low latency, and energy efficiency for ultra-low power consumption applications. Then, we develop an analytical framework based on an M/G/1/2 queue to evaluate the performance of these WuR protocols. Discrete-event simulations validate the accuracy of the analytical models.