Abstract
In this paper, we analyze the performance of two-way amplify-and-forward (AF) relaying with best relay selection scheme in the presence of hardware impairment at the relays. We first obtain the tractable closed-form expression of the statistical characteristics for the end-to-end signal-to-noise ratio (SNR). To reveal significant insights into the system performance, we also computed the asymptotic symbol error probability (SEP), which is shown extremely tight in the high SNR regime and low hardware impairments level. Our analysis enables us to evaluate the impact of non-ideal hardware on the performance of two-way relaying networks and to highlight as a benchmark for system designers.
Introduction
Traditional wireless communication research often relies on the assumption of "perfect hardware." However, real-world radio-frequency (RF) transceivers are subject to physical imperfections such as amplifier non-linearities, phase noise in oscillators, and IQ imbalance. While these effects are well-documented in point-to-point links, their impact on two-way amplify-and-forward (AF) relaying is significantly more complex.
In a two-way setup, two nodes exchange information simultaneously through a relay. If that relay has "dirty" hardware, it doesn't just amplify the desired signals; it injects distortion noise that is subsequently amplified and broadcast back to the users. This paper mathematically characterizes this degradation to provide design guidelines for practical, low-cost relay hardware.
Analytical Framework
The authors employ a generalized impairment model where the received signal at the relay is distorted by two noise terms: $\eta_{3r}$ representing receiver impairments and $\eta_{3t}$ for transmitter impairments. These are modeled as Gaussian noise scaled by a hardware characterization parameter, $\kappa$, which is typically related to the Error Vector Magnitude (EVM).
The analytical approach involves:
- SNDR Derivation: Calculating the Signal-to-Noise-and-Distortion Ratio (SNDR) for both nodes, accounting for the relay's variable gain.
- Statistical Modeling: Using Rayleigh fading channel assumptions to derive exact closed-form expressions for the Outage Probability (OP) using modified Bessel functions.
- Asymptotic Analysis: Examining the "high-power regime" to determine if the system can eventually overcome hardware-induced noise.
- Numerical Validation: Verifying the mathematical formulas through Monte-Carlo simulations.
Key Insights and Results
The study reveals several critical findings regarding the performance of impaired relaying systems:
The Irreducible Error Floor
Unlike ideal systems where signal quality improves indefinitely with increased transmit power, impaired systems hit a fundamental "performance ceiling." Beyond a certain power level, the distortion noise grows proportionally with the signal, causing the SNDR to saturate. This leads to an irreducible error floor, as shown in Figure 3. Even at infinite transmit power, the Symbol Error Rate (SER) converges to a non-zero constant.
Optimal Hardware Symmetry
A significant engineering takeaway is that for a fixed "hardware budget" (total EVM), optimal performance is achieved when the relay's transmitter and receiver hardware are of equal quality ($\kappa_{3t} = \kappa_{3r}$). As seen in the figure, asymmetric configurations (e.g., $\kappa_{3t}=0, \kappa_{3r}=0.2$) result in a higher error floor compared to symmetric ones with the same aggregate impairment.
High-Rate Vulnerability
High-rate systems requiring high spectral efficiency are particularly sensitive to these impairments. If the hardware quality is sufficiently poor (e.g., $\kappa = 0.2$), the system may remain in a state of permanent outage regardless of the available transmit power.
Conclusion and Future Directions
This research provides a robust mathematical framework for evaluating two-way relaying under realistic hardware constraints. While the current model assumes perfect Channel State Information (CSI), future work could explore the compounded impact of hardware impairments on channel estimation accuracy. Additionally, extending the analysis to include specific non-linearities beyond aggregate Gaussian noise could further refine these practical design benchmarks.
Summary updated