In this study, we consider four data transmission protocols for the wireless power transfer-based Internet-of-Things (IoT) data collection system and then design the resulting resource allocation mechanisms to enhance user fairness. First, we adopt the time division multiple access protocol, in which the joint slot partition and power allocation can be equivalently transformed into a convex problem. However, this protocol poses stringent requirements for synchronization precision in IoT devices. Thus, we introduce the nonorthogonal multiple access (NOMA) protocol to avoid the partition of the transmission slot. We then propose a block coordinate descent algorithm-based mechanism to achieve a stationary point of the original non-convex problem by alternately solving two convex subproblems. However, this approach leads to severe co-channel interference, thus, impairing the transmission performance. Therefore, we design two fixed-length sliced partial NOMA protocols to balance transmission performance and implementation complexity. We also leverage the matching theory and successive convex approximation algorithm to propose two-joint slice assignment and power allocation mechanisms, respectively. Our simulation results verify the trade-off between transmission performance and implementation complexity and show the convergence of the proposed mechanisms as well as the impact of system parameters on optimized results.