Abstract
This paper investigates a novel framework for integrating Non-Orthogonal Multiple Access (NOMA) with millimeter-wave (mmWave) communications through the use of random beamforming. By leveraging the inherent directionality of mmWave signals and employing stochastic geometry, the authors demonstrate that NOMA can significantly enhance spectral efficiency while drastically reducing the feedback overhead typically associated with full channel state information (CSI). The study introduces a wedge-based scheduling mechanism and evaluates various limited feedback schemes, showing that even a simple one-bit feedback approach can achieve substantial performance gains.
Introduction
As wireless networks transition toward 5G and beyond, the combination of mmWave communications and NOMA has emerged as a promising solution to meet the demands for massive bandwidth and high spectral efficiency. mmWave offers vast frequency resources, while NOMA allows multiple users to share the same time-frequency resource block through power-domain multiplexing.
However, a significant challenge in implementing NOMA in mmWave networks is the requirement for the Base Station (BS) to possess accurate CSI for all users to perform effective power allocation and Successive Interference Cancellation (SIC). In dense networks with highly directional beams, the signaling overhead for full CSI acquisition can be prohibitive. This paper addresses this bottleneck by proposing a random beamforming approach that eliminates the need for precise beam alignment and extensive feedback.
Methods and System Model
The proposed framework utilizes random beamforming to simplify the communication process. Instead of calculating optimal beams for each user, the BS generates random beams and schedules users based on their spatial alignment with these beams.
Wedge-Based Scheduling
To ensure that scheduled users have sufficient channel gains, the BS employs a wedge-based scheduling strategy. Only users located within a specific angular "wedge" or sector ($\Delta$) around a randomly generated beam are considered for scheduling. This approach exploits the high directionality of mmWave signals to maintain high effective channel gains for the paired NOMA users.
Stochastic Geometry and Blockage Modeling
The researchers model the user distribution using a Homogeneous Poisson Point Process (HPPP). To account for the realistic propagation environment of mmWave, a "thinning" process is incorporated to model blockages (e.g., buildings, foliage). This ensures that the analysis reflects the probability of Line-of-Sight (LOS) connections, which are critical for mmWave performance.
Feedback Schemes
The paper analyzes three distinct feedback levels to evaluate the trade-off between overhead and performance:
- Full CSI: Users feedback their exact effective channel gains, serving as a performance benchmark.
- Distance-Only Feedback: The BS orders users based on their distance (path loss), reducing feedback complexity at the cost of ignoring instantaneous fading.
- One-Bit Feedback: Users send a single bit indicating whether their channel gain exceeds a pre-defined broadcast threshold.
Key Insights
The analytical framework reveals several critical insights into the behavior of mmWave NOMA systems:
- Channel Correlation as an Advantage: Unlike traditional MIMO where high correlation can be detrimental, the directional nature of mmWave leads to high channel correlation between users in the same beam. This simplifies the interference environment, making it ideal for NOMA pairing.
- The Efficacy of One-Bit Feedback: A key finding is that a properly designed one-bit feedback scheme can achieve the full diversity gain for the "strong" NOMA user, nearly matching the performance of systems with full CSI.
- Diversity Order Trade-offs: The authors prove that while full CSI allows the $i$-th ordered user to achieve a diversity gain of $i$, distance-based ordering limits the diversity gain to one, highlighting the importance of capturing some level of fading information.
Results
The theoretical derivations were validated through extensive Monte Carlo simulations. Key results include:
- Sum Rate Gains: NOMA consistently outperforms conventional Orthogonal Multiple Access (OMA). For instance, at a transmission power of 30 dBm, NOMA achieved a sum-rate gain of approximately 5 bits per channel use (BPCU) over OMA.
- Outage Probability: The proposed random beamforming NOMA scheme effectively reduces outage probability, particularly for users with favorable channel conditions.
- Inter-Beam Interference: In multi-beam scenarios, the high directionality of mmWave allows for effective suppression of inter-beam interference, provided there is sufficient angular separation between the beams.
Conclusion and Future Work
This research demonstrates that random beamforming is a viable and efficient strategy for implementing NOMA in mmWave networks. By reducing feedback requirements, the proposed framework enables high spectral efficiency in dense environments.
Future research directions include investigating the impact of user mobility and the resulting beam-tracking overhead, exploring more complex multi-path scattering models beyond LOS-dominant paths, and extending the analysis to multi-user clusters per beam to further push the limits of spectral efficiency.
Summary updated