The combination of non-orthogonal multiple access (NOMA) using power-domain with backscatter sensor communication (BSC) is expected to connect a large-scale Internet of things (IoT) devices in future sixth-generation (6G) era. In this paper, we introduce a BSC in multi-cell IoT network, where a source in each cell transmits superimposed signal to its associated IoT devices using NOMA. The backscatter sensor tag (BST) also transmit data towards IoT devices by reflecting and modulating the superimposed signal of the source. A new optimization framework is provided that simultaneously optimizes the total power of each source, power allocation coefficient of IoT devices and reflection coefficient of BST under imperfect successive interference cancellation decoding. The objective of this work is to maximize the total energy efficiency of IoT network subject to quality of services of each IoT device. The problem is first transformed using the Dinkelbach method and then decoupled into two subproblems. The Karush-Kuhn-Tucker conditions and Lagrangian dual method are employed to obtain the efficient solutions. In addition, we also present the conventional NOMA network without BSC as a benchmark framework. Simulation results unveil the advantage of our considered NOMA BSC networks over the conventional NOMA network.