Backscatter Communication: Design and Optimisation For Emerging Use-Cases

Abstract

Backscatter communication (BackCom) holds significant potential to improve the pervasiveness and energy efficiency of future wireless networks, through its passive modulation and reuse of existing radiofrequency signals. In order to function as a key technology under the Internet of Things paradigm, issues relating to BackCom, such as its limited coverage and deployment flexibility, low data rates, and the difficulty of channel estimation, need to be addressed. To complement this, a wider range of use-cases and deployment scenarios also need to be established. This thesis focuses on addressing these issues inherent to BackCom, by exploring a series of system setups which push the boundaries in terms of coverage and flexible deployment, and then future-proofs BackCom through the study of the assistance from another emerging technology, the intelligent reflecting surface (IRS).

The first half of the thesis focuses on the coverage and deployment flexibility of BackCom devices under conventional wireless communication settings. First, we study a novel use-case in which BackCom devices replace conventional, actively transmitting relays to assist an information transmission from a source to a destination. We introduce the decode-and-forward (DF) BackCom relaying scheme and perform a detailed bit error rate (BER) characterisation of the DF BackCom scheme alongside the amplify-and-forward (AF) BackCom 'reflection' scheme. The feasibility and practical range of the BackCom relay is demonstrated through a case study, and our findings indicate that with careful selection of relay parameters, the DF scheme can improve the functionality of BackCom relays through the decoding operation, while resulting in minimal BER differences compared to the AF 'reflection' scheme. Second, we study the coverage maximisation of bistatic BackCom systems in wide-area environmental monitoring applications through judicious power beacon (PB) placement. We propose a straightforward metric to characterise coverage, the guaranteed coverage distance (GCD), to overcome the complex shape of each PB's coverage area when the performance of the BackCom link is dependent on the strength of the energy transfer link. We find that a single-tier symmetric deployment of PBs performs favourably under a practical number (24 or less) of PBs, with a GCD of more than 100m being readily achievable.

The second half of the thesis studies the incorporation of the IRS into BackCom systems, with the aim of improving BackCom performance. The IRS-assisted bistatic BackCom system is studied first, where we solve a transmit power minimisation problem at the carrier emitter involving the joint optimisation of the transmit and receive beamforming, the IRS phase shifts and the BackCom splitting coefficients. We present a unique signal model arising from this system, where a signal originating from the carrier emitter may be reflected by the IRS twice before reaching the reader, and account for this added complexity in our algorithm design. Our results indicate that transmit power savings of over 6 dB may be achieved with a moderately-sized IRS, which may be converted to nearly 50m of range increase. Then, we study the use of the IRS in an ambient BackCom system, with the goal of reducing direct-link interference and improving detection performance. We assume the absence of all ambient signal and channel knowledge, which is a practical assumption given the passively reflecting nature of both BackCom devices and IRSs. We propose a deep reinforcement learning (DRL)-based algorithm which maximises the backscatter channel difference (that is, the ratio of the energies of the direct-link interference and overall received signal) based on instantaneous signal samples, which may be converted to BER reductions. We find that the DRL approach with no channel knowledge can achieve a backscatter channel difference within 25% of that obtained using benchmarks with full channel knowledge. Show more