Symbol Detection in Multi-channel Multi-tag Ambient Backscatter Communication Under IQ Imbalance
Yuxin Li, Guangyue Lu, Yinghui Ye, Liqin Shi, Daniel Benevides da Costa
TL;DR
This work analyzes symbol detection in a direct-conversion wideband AmBC system with 2M backscatter tags and FDMA channel allocation under IQ imbalance that induces image-channel crosstalk. A differential-encoding and energy-difference detector framework is developed, yielding a Gaussian-approximation BER expression and a near-optimal detection threshold $\gamma_{m,k}^{iq}$, with $\Delta_+^2 = \Delta_0^2 + \Delta_1^2$. Since the needed parameters are not available, a blind threshold-estimation scheme uses sample statistics: $\mathbb{E}(|T_{m,k}^{iq}|)$ and $\mathbb{D}(T_{m,k}^{iq})$ to estimate $|\vartheta_{m,k}^{iq}|$ and $\Delta_+^2$ (Lemma 4), solving $f(\Delta_+)=0$ via a one-dimensional bisection (Algorithm 1). Numerical results corroborate the analysis and show substantial BER gains over using ideal-threshold detectors, especially under moderate-to-strong IQ imbalance.
Abstract
Ambient backscatter communication (AmBC) offers low-cost and low-power connectivity for Internet of Things (IoT), where a backscatter tag (BT) modulates incident signals transmitted by an ambient radio frequency (RF) source and reflects them to its associated AmBC receiver. In multi-channel multi-tag AmBC, one of major challenges from the aspect of symbol detection is the image channel crosstalk, which is induced by the inevitable in-phase/quadrature (IQ) imbalance. To address this issue, in this paper, we study symbol detection in multi-channel multi-tag AmBC under IQ imbalance. Considering the differential encoding scheme at the BTs, we propose a novel symbol detection model that incorporates IQ imbalance parameters, the presence or absence of both the incident signal and the backscattered signal of the image channel. On this basis, considering an energy difference detector at the AmBC receiver, we derive the closed-form expressions for the bit error rate (BER) as well as the near-optimal detection threshold to minimize BER. However, calculating the near-optimal detection threshold requires prior information, such as the IQ imbalance parameters, the presence probability of the incident signal of the image channel and the backscattered signal of the image channel, the signal power of the ambient RF source, and the noise power, which are typically unknown to the AmBC receiver in practice. To eliminate the need for the prior information, we propose a threshold estimation method using the received samples. Numerical results indicate that under IQ imbalance, directly using the existing method leads to a significant degradation in BER performance. However, this degradation can be effectively mitigated by our derived detection threshold.