Motivated by the benefits of cognitive radio (CR), energy harvesting (EH), and backscatter communication (BC) technologies to support Internet of Things (IoT) systems, we investigate the backscatter-aided EH CR networks (EH-CRNs) in a multichannel scenario. To achieve high throughput on various channels, we propose a novel hybrid communication scheme that the secondary transmitter (ST) selects one channel for spectrum sensing, and performs multiple actions based on the sensing result. To be specific, if the selected channel is detected as busy, the ST potentially performs underlay mode transmission, ambient BC (AmBC), or radio frequency (RF) EH. Otherwise, the ST performs interweave mode transmission. Based on the ST’s knowledge of the channel availability and the amount of the available energy, the decisions of channel and specific action selections are made. Furthermore, the sequential decision problem is formulated as a mixed observability Markov decision process (MOMDP), and addressed by the classic value iteration algorithm. The proposed scheme could be flexibly adapted to the changes in energy and channel availabilities. Simulations demonstrate the superiority of this scheme in terms of throughput, and show that even without channel selection, the proposed scheme conducted on the channels with different idle probabilities always achieves high throughput.