Interference and Rate Analysis of Multinumerology NOMA

Stephen McWade, Mark F. Flanagan, Lei Zhang, Arman Farhang

TL;DR

This paper introduces and analyzes a Multinumerology Non-Orthogonal Multiple Access (MN-NOMA) scheme for 5G and beyond networks. By allowing users to transmit with different numerologies (subcarrier spacings and symbol durations) within the same NOMA framework, the scheme aims to achieve both the flexibility needed for diverse services and the spectral efficiency benefits of NOMA, without the significant spectral efficiency loss seen in prior multinumerology approaches.

What Problem Does This Paper Address?

Future wireless networks, particularly 5G and beyond, need to support a wide array of services with vastly different requirements. For instance, massive machine-type communications (mMTC) might benefit from narrow subcarrier spacing for delay tolerance, while ultra-reliable low-latency communications (URLLC) demand wide subcarrier spacing for reduced latency. This heterogeneity makes a single "one-size-fits-all" numerology, like in 4G LTE, impractical. While 5G standards allow for different numerologies to coexist, traditional Orthogonal Multiple Access (OMA) approaches struggle to manage interference between these different numerologies, often requiring guard bands that reduce spectral efficiency. Non-Orthogonal Multiple Access (NOMA) offers a way to improve spectral efficiency by allowing users to share resources, but standard NOMA assumes users have the same numerology. This paper tackles the challenge of enabling NOMA for users with different numerologies, aiming to combine the flexibility of diverse numerologies with the spectral efficiency gains of NOMA.

Methods

The core of this paper is the development and analysis of a Multinumerology NOMA (MN-NOMA) system. The authors propose a scheme where users with different numerologies share resources and are multiplexed in the power domain, a standard NOMA technique. Crucially, they employ Successive Interference Cancellation (SIC) at the receiver to decode user signals. The novelty lies in deriving analytical expressions for the Inter-Numerology Interference (INI) experienced by users when their numerologies differ. This INI is quantified using the Mean Squared Error (MSE). Based on these INI analyses, the paper derives expressions for the achievable rates of each user. The analysis considers two SIC ordering scenarios: user 1 decoded first, and user 2 decoded first, acknowledging the inherent asymmetry in interference experienced due to differing numerologies.

Datasets

Not discussed in the paper. The evaluation relies on simulated channel models.

Key Insights and Ideas

  • Asymmetric Interference: The interference experienced by users in MN-NOMA is not symmetrical. The user with the narrower subcarrier spacing (longer symbol duration) experiences interference that oscillates around the channel frequency response of the other user. This is due to the broken orthogonality between subcarriers of different numerologies.
  • Flexibility Meets Efficiency: MN-NOMA successfully combines the flexibility of accommodating diverse numerologies with the spectral efficiency benefits of NOMA. This is a significant step towards realizing the full potential of 5G and beyond networks.
  • No Significant SE Loss vs. Single Numerology NOMA: Despite the added complexity of multinumerology interference, the proposed MN-NOMA scheme achieves spectral efficiency very close to that of single-numerology NOMA (SN-NOMA), indicating that the benefits of NOMA are largely preserved.
  • Superiority over Multinumerology OMA: MN-NOMA consistently outperforms Multinumerology OMA (MN-OMA) in terms of spectral efficiency, highlighting the advantage of non-orthogonal resource sharing.
  • Ordering Matters for Interference, Less for Average SE: While the specific interference experienced by each user depends on the SIC decoding order, the average spectral efficiency across both orderings remains similar when optimal power allocation is used.

Key Results

  • Analytical expressions for Inter-Numerology Interference (INI) in terms of Mean Squared Error (MSE) were derived for both SIC ordering scenarios.
  • Achievable rate expressions for users in the MN-NOMA system were derived based on the INI analysis.
  • Numerical simulations demonstrated that MN-NOMA achieves spectral efficiency very close to SN-NOMA across various Signal-to-Noise Ratios (SNRs) and channel models (EPA and EVA).
  • MN-NOMA consistently showed superior spectral efficiency compared to MN-OMA.
  • The paper illustrates the asymmetric interference pattern, where the MSE for one user follows the channel frequency response of the other user, while the reverse is not as pronounced.

Limitations and Future Work

The authors acknowledge several limitations and suggest future research directions:

  • Constant Power Allocation: The current analysis assumes constant power allocation across all subcarriers. Future work could explore optimized power allocation strategies, such as water-filling, to further enhance performance, especially considering the oscillating MSE patterns.
  • Perfect Synchronization: The model assumes perfect time and frequency synchronization between users. Investigating the performance gains and flexibility in asynchronous scenarios would be a valuable extension.
  • Simplified Channel Model: While multipath fading channels are considered, more complex channel models could be explored.
  • Limited User Count: The analysis primarily focuses on a 2-user system. Extending the framework to support more users with varying numerologies would be beneficial.

Summary updated

Abstract

5G communication systems and beyond are envisioned to support an extremely diverse set of use cases with different performance requirements. These different requirements necessitate the use of different numerologies for increased flexibility. Non-orthogonal multiple access (NOMA) can potentially attain this flexibility by superimposing user signals while offering improved spectral efficiency (SE). However, users with different numerologies have different symbol durations. When combined with NOMA, this changes the nature of the interference the users impose on each other. This paper investigates a multinumerology NOMA (MN-NOMA) scheme using successive interference cancellation (SIC) as an enabler for coexistence of users with with different numerologies. Analytical expressions for the inter-numerology interference (INI) experienced by each user at the receiver are derived, where mean-squared error (MSE) is the metric used to quantity INI. Using the MSE expressions, we analytically derive achievable rates for each user in the MN-NOMA system. These expressions are then evaluated and used to compare the SE performance of MN-NOMA with that of its single-numerology counterpart. The proposed scheme can achieve the desired flexibility in supporting diverse use cases in future wireless networks. The scheme also gains the SE benefits of NOMA compared to both multinumerology and single numerology orthogonal multiple access (OMA) schemes.
BibTeX
@inproceedings{mcwade2020interference,
  author = {Stephen McWade and Mark F. Flanagan and Lei Zhang and Arman Farhang},
  title = {{Interference and Rate Analysis of Multinumerology NOMA}},
  booktitle = {ICC},
  year = {2020},
  doi = {10.1109/icc40277.2020.9149041},
}
Figure