In future wireless communication systems, a substantial proportion of devices will be connected to the Internet of Things (IoT) networks for data communications, which poses three critical challenges for designing IoT networks: cost efficiency, spectrum efficiency, and energy efficiency. Ambient backscatter communication (AmBC), a novel communication scheme that offers low-cost, spectrum-efficient, and energy-efficient data communications, has found itself a competitive solution for developing and deploying future IoT networks. In AmBC systems, the main design challenge is to decode the tag signals from the composite received signals. In this thesis, the author focuses on the transceiver design and performance analysis of the AmBC system by making the following contributions. Firstly, we design a machine learning-based detector to decode the tag signals for an AmBC system. The second focus of this thesis is to study the BER performance of the AmBC systems that utilize the RF source signals with error control coding. The ongoing developments of 5G wireless networks are continuously exposing some inherent limitations. Reconfigurable intelligent surface (RIS), a promising solution to overcome the limitations in current 5G wireless networks and enable the beyond 5G and future sixth-generation (6G) networks, has attracted increasing attention recently. In the third research focus, we investigate the performance of the RIS-assisted wireless system where the signal transmitted from the transmitter is protected by error control codes. We focus on deriving the analytical upper and lower bounds on the bit error probability of the RIS-assisted wireless system with LDPC-coded source signals. We further investigate the performance of the RIS-assisted wireless system with polar codes through simulations. In addition, we show that the deployment of RIS can enhance the system BER performance significantly by increasing the number of RIS reflecting elements.