Multi-Antenna Ambient Backscatter Communications: Performance Optimization and Analysis

Abstract

Amid the evolution of Internet-of-things (IoT), connecting the ever-increased number of devices requires sustainable solutions. Ambient backscatter communications (AmBC) emerges as a promising enabler because of its ultra-low power consumption and maximized spectrum efficiency. Wider deployment of AmBC is confronted with the need to increase the communication range and achievable data rate of a backscatter device (BD). These two system performance indicators are limited by the poor bit error rate (BER) performance of demodulating the BD signal due to its extremely low signal-to-noise ratio (SNR). In this thesis, AmBC receiver design and BD signaling matrix design are investigated to improve the AmBC system performance. Their implementation details are discussed to improve their practicality and generality. The results suggest that the proposed designs minimize BER compared to the state-of-the-art works, and thus foster a diversity of use cases. The research in the thesis exploits multiple antennas to improve the SNR of the BD signal. A machine learning (ML)-assisted coherent receiver that obtains an SNR gain is proposed. The ML classification is used to estimate and compensate for the phase offset caused by the excess path length of the backscatter path channel compared to the direct path channel. In contrast, the proposed optimum receiver considers the composite channel of the two paths. As the phase offset is included in the composite channel, the non-coherent optimum receiver achieves the same BER performance as the coherent receiver. The derived receivers generalize to different AmBC deployments regardless of the types of BD modulation and ambient signals, and thus need not be tailored to specific system setups. Their performance analyses both suggest that BD placement close to the ambient source or AmBC receiver but away from the line segment between them yields a better BER performance. Furthermore, optimizing the reflection coefficients on a multi-antenna BD, termed a BD signaling matrix, maximizes backscatter signal strength. This is a particularly attractive solution in scenarios with multiple ambient sources. The practicality of the presented designs is elaborated in depth and evaluated through experimentation. Methods for estimating the parameters of the AmBC receivers and the BD signaling matrix are proposed. Solutions are provided for mitigating the dynamic range issue of analog-to-digital converters when the receivers are implemented in the digital domain. The proposed multi-antenna AmBC receivers are verified to outperform the receivers in the literature through measurements from a prototyped AmBC system. A processing method is proposed to guarantee a successful implementation. Therefore, the thesis is composed of both theoretical derivation and practical demonstration of the designs that improve AmBC system performance.