RFID: A Communication System Perspective

This dissertation focuses on three key areas relating to backscatter modulation andRadio Frequency IDentification (RFID) systems: 1) analyze the co-design of backscatter modulation and error correction coding, 2) characterize space-time coding and MIMO performance limits of the dyadic backscatter channel, and 3) hybrid-ARQ (HARQ) and random access improvements to the Gen2 protocol.;Passive and semi-passive RFID tags depend on scavenged energy to power their IC. While backscatter modulation itself consumes a negligible amount of energy, the modulator creates an impedance mismatch between the tag's antenna and power harvester, thereby decreasing the antenna to the tag power transfer efficiency. This required impedance mismatch couples the link performance to the power harvester's performance, so to quantify this tradeoff, we introduce a new metric: backscatter power efficiency loss per bit. Higher order constellations improve the link's spectral efficiency, but have lower power efficiency as compared to binary modulation schemes. We propose new coded modulation schemes based on unequal error protection, which improves both the spectral efficiency and the backscatter power efficiency loss metric.;MIMO processing is a canonical technique to improve wireless link capacity and reliability, which will require future tags and readers to have multiple antennas. The dyadic backscatter channel (DBC) models the behavior of small-scale fading in RFID MIMO systems, however, its statistics differ from those of the classic Rayleigh fading MIMO channel. We analytical characterize the performance of space-time trellis codes and orthogonal space-time block codes, derive an upper bound to the pairwise error probability (PEP), and derive the maximum diversity order of the DBC. Unlike Rayleigh fading, the diversity order only depends on the number of tag antennas but not the number of reader receive antennas.;In fading channels, MIMO techniques offer two opposing performance gains: diversity (reliability against outage events) or multiplexing gain (spectral efficiency). The diversity multiplexing tradeoff (DMT) is an asymptotic measure that quantifies the achievable diversity for a given multiplexing gain. Starting from the definition of the DBC and the DMT of the double scattering channel, the corresponding DMT of the DBC is derived. The statistics of the DBC limit the amount diversity when compared to the Rayleigh MIMO channel, although the available multiplexing gain is unchanged. Increasing the number of receive antennas improves both diversity and multiplexing gain until the receive antenna count equals the number of tag antennas, otherwise additional receive antennas offer no gains with respect to the DMT.;The current EPC Gen2 standard does not use any form of error correction and does not allow for fast link adaption between reading separate tags. We consider a protocol that uses HARQ algorithms without requiring major changes to the Gen2 protocol. In addition, we develop theoretical models that capture EPC Gen2's baseline performance and capacity in terms SNR and tag read rate. Existing HARQ algorithms, such as Chase combining (CC) and incremental redundancy (IR), are studied via simulations and the performance quantified in terms of tag read rate. The simulation results show that CC allows for graceful system degradation and IR achieves read rates close to EPC Gen2's capacity limit.;Random access plays a critical role in RFID tag singulation. EPC Gen2 uses frame slotted ALOHA (FSA) to arbitrate channel resources between tags, but FSA has low efficiency due to empty slots and collisions. To aid in tag collision resolution, we consider multiuser detection (MUD) and incremental redundancy enhancements to the FSA protocol. The theoretical performance of the MUD receiver is analyzed from a compressive sensing viewpoint. As an example of a practical code construction, we evaluate the performance of punctured second order Reed Muller codes. These enhancements improve FSA's throughput and its response to high system loads. (Abstract shortened by UMI.).