Abstract
In-band full-duplex (FD) wireless communication offers the potential to double spectral efficiency by enabling simultaneous transmission and reception on the same frequency. The primary challenge is self-interference (SI), where a transmitter's own signal masks incoming weak signals. This paper presents a theoretical framework using the Poon-Broderson-Tse (PBT) antenna model to analyze the spatial degrees of freedom (DoF) available for SI isolation. We demonstrate that FD performance is governed by the physical footprint of the base station and the angular geometry of environmental scattering. Our analysis shows that FD can achieve up to a 2x capacity gain over traditional half-duplex systems when the base station possesses sufficient spatial dimensions or when scattering paths are sufficiently distinct.
Introduction
The "holy grail" of wireless communication is in-band full-duplex (FD) operation. By transmitting and receiving at the same time on the exact same frequency, network capacity could theoretically double. However, the practical implementation is hindered by self-interference (SI): the base station’s high-power transmission can drown out the weak incoming signal from a distant user.
Engineers typically combat SI through two mechanisms: digital/analog cancellation and spatial isolation. Spatial isolation uses antenna beamforming to create "nulls" in the direction of interference. However, utilizing spatial degrees of freedom for isolation may reduce the resources available for data transmission. This paper investigates the fundamental trade-off between SI suppression and data rate, determining the environmental conditions under which FD provides a net benefit.
Methodology
The study departs from standard discrete antenna matrix models, instead employing the PBT model. This approach considers the physical size of antenna arrays and the continuous angular distribution of environmental scatterers. The model is applied to a three-node topology: a full-duplex base station (BS), one uplink (UL) user, and one downlink (DL) user.
By treating signals as functions in a Hilbert space and performing singular value decomposition (SVD) on scattering operators, the authors derive the Degrees-of-Freedom (DoF) region. This region defines the achievable data rate pairs for UL and DL in the high-signal-to-noise-ratio (SNR) limit.
Key Insights and Analysis
The analysis reveals several critical factors influencing FD feasibility:
- Geometry and Angular Spread: The success of FD depends heavily on the environment's "angular spread." If the objects causing backscattered SI are located in different directions than the intended users, isolation is "cheap," and FD gains are easily realized.
- The "Extra Dimension" Advantage: If the base station's antenna array is physically larger than those of the users, it possesses surplus spatial dimensions. These can be used to zero-force SI without sacrificing data-carrying capacity.
- Active Spatial Isolation: While passive shielding only addresses direct-path SI, active beamforming can effectively suppress backscattered SI—interference reflecting off nearby environmental features like walls or trees.
Key Results
The paper mathematically characterizes the FD DoF region as a pentagon defined by single-user bounds and a combined sum-DoF bound. Key findings include:
- Overlapped Scattering: In the worst-case scenario where interference and signal directions perfectly overlap, FD only outperforms half-duplex (HD) if the base station array is physically longer than the user arrays.
- Symmetric Spread: In general cases, FD is strictly superior to HD unless scattering intervals are perfectly overlapped, as any exclusive spatial direction can be leveraged for improvement.
- Capacity Doubling: If the backscatter interval is sufficiently distinct from the forward scattering interval, the base station can effectively "hide" its SI, achieving a full 2x gain over HD.
Limitations and Future Work
While providing a rigorous theoretical foundation, the study has limitations. The use of continuous linear arrays is an approximation; real-world discrete elements may introduce leakage. Furthermore, the high-SNR focus ignores noise floors and transceiver impairments like phase noise. Future research should address dense network scenarios where inter-user interference becomes a factor and extend the analysis to multi-user massive MIMO topologies.
Summary updated