Multi-Cell Sparse Activity Detection for Massive Random Access: Massive MIMO versus Cooperative MIMO

Zhilin Chen, Foad Sohrabi, Wei Yu

Abstract

In the evolving landscape of 5G and 6G, massive machine-type communications (mMTC) present a significant challenge: detecting sporadic activity among millions of devices. This paper investigates two primary network architectures for sparse device activity detection—Massive MIMO and Cooperative MIMO—using the Approximate Message Passing (AMP) algorithm. While Massive MIMO relies on a high number of antennas at individual base stations to suppress inter-cell interference, Cooperative MIMO utilizes a Cloud-Radio Access Network (C-RAN) to jointly detect users across cells. Our analysis and simulations demonstrate that while Massive MIMO can theoretically eliminate detection errors as the antenna count grows, Cooperative MIMO provides a more efficient solution for cell-edge users, achieving comparable reliability with significantly fewer antennas through base station cooperation and quantized log-likelihood ratio (LLR) forwarding.

Introduction

The proliferation of Internet of Things (IoT) devices necessitates support for massive machine-type communications, where millions of devices are connected to a single network. A defining characteristic of mMTC is sporadic traffic; at any given moment, only a small fraction of the total device population is active. Consequently, the base station (BS) must identify active users before data transmission can be decoded.

Due to the sheer volume of potential users, assigning unique orthogonal signature sequences to every device is impractical. Instead, non-orthogonal sequences are employed, which inherently leads to inter-cell interference. This paper explores the trade-offs between two architectural strategies to mitigate this interference: increasing the spatial resolution at individual base stations (Massive MIMO) or fostering collaboration between base stations (Cooperative MIMO).

System Architectures and Methodology

The AMP Framework

The problem of user activity detection is framed as a sparse recovery challenge. Given that only a small subset of users is active, the system seeks to recover a sparse activity vector from the received signal. The authors utilize the Approximate Message Passing (AMP) algorithm, integrated with a Bayesian denoiser, to solve this problem efficiently. To characterize the performance of these architectures, the paper employs state evolution, a mathematical framework that tracks the MSE and error probabilities of AMP in the asymptotic regime.

Massive MIMO (Non-Cooperative)

In the Massive MIMO architecture, each base station operates independently. It attempts to detect the active users within its own cell while treating signals from neighboring cells as additive white Gaussian noise. This approach simplifies the network backhaul but requires a substantial number of antennas ($M$) to overcome the interference floor through spatial multiplexing gains.

Cooperative MIMO (C-RAN)

In the Cooperative MIMO architecture, multiple base stations are connected to a Central Unit (CU). Each BS runs the AMP algorithm to detect not only its local users but also users in neighboring cells. Instead of making hard decisions, the BSs compute soft information in the form of Log-Likelihood Ratios (LLRs). These LLRs are forwarded to the CU, which aggregates the information to make a final decision on user activity. This method treats inter-cell interference as a signal to be recovered rather than noise to be suppressed.

Key Insights and Performance Trade-offs

The Antenna vs. Cooperation Trade-off

A central finding of this research is the quantitative trade-off between hardware scale and network cooperation. Analytical results show that in Massive MIMO, the probabilities of false alarm ($P_{fa}$) and missed detection ($P_{md}$) can be driven toward zero as $M \to \infty$. However, Cooperative MIMO is significantly more effective at protecting cell-edge users who suffer most from inter-cell interference.

Numerical simulations of a 19-cell network reveal that a Cooperative MIMO system involving just three cooperating base stations (each with 16 antennas) achieves the same cell-edge detection reliability as a non-cooperative Massive MIMO system with 48 antennas per BS. This suggests that cooperation can reduce the required antenna count by a factor of four in certain practical scenarios.

Fronthaul Efficiency and Quantization

To address the practical limitations of C-RAN, specifically the finite capacity of the fronthaul links connecting BSs to the CU, the paper investigates LLR quantization. A standard uniform quantization fails because the dynamic range of LLRs varies significantly based on a user's large-scale fading coefficients.

The authors propose a user-specific uniform quantization scheme. By setting a quantization coverage parameter (e.g., $\zeta = 0.97$), the system can represent LLRs with high precision using only 3 to 4 bits. Simulations show that this low-resolution quantization allows the Cooperative MIMO system to nearly match the performance of a system with infinite-capacity fronthaul.

Conclusion and Future Directions

This study highlights that while Massive MIMO is a robust "brute-force" method for reducing detection errors, Cooperative MIMO offers a sophisticated and resource-efficient alternative, particularly for improving the experience of cell-edge users. The transition from treating interference as noise to actively recovering it through C-RAN architectures provides a clear path forward for dense mMTC deployments.

Future work remains to address the assumptions of uncorrelated Rayleigh fading and the requirement for base stations to know the signature sequences of out-of-cell users. Incorporating spatial correlation into the Bayesian denoiser and reducing the signaling overhead for inter-cell coordination are essential steps toward implementing these findings in real-world 5G-Advanced and 6G networks.

Summary updated

Abstract

This paper considers sparse device activity detection for cellular machine-type communications with non-orthogonal signatures using the approximate message passing algorithm. This paper compares two network architectures, massive multiple-input multiple-output (MIMO) and cooperative MIMO, in terms of their effectiveness in overcoming inter-cell interference. In the massive MIMO architecture, each base station (BS) detects only the users from its own cell while treating inter-cell interference as noise. In the cooperative MIMO architecture, each BS detects the users from neighboring cells as well; the detection results are then forwarded in the form of log-likelihood ratio (LLR) to a central unit where final decisions are made. This paper analytically characterizes the probabilities of false alarm and missed detection for both architectures. Numerical results validate the analytic characterization and show that as the number of antennas increases, a massive MIMO system effectively drives the detection error to zero, while as the cooperation size increases, the cooperative MIMO architecture mainly improves the cell-edge user performance. Moreover, this paper studies the effect of LLR quantization to account for the finite-capacity fronthaul. Numerical simulations of a practical scenario suggest that in that specific case cooperating three BSs in a cooperative MIMO system achieves about the same cell-edge detection reliability as a non-cooperative massive MIMO system with four times the number of antennas per BS.
BibTeX
@article{chen2019multi,
  author = {Zhilin Chen and Foad Sohrabi and Wei Yu},
  title = {{Multi-Cell Sparse Activity Detection for Massive Random Access: Massive MIMO versus Cooperative MIMO}},
  journal = {IEEE Trans. Wireless Communications},
  year = {2019},
  url = {https://openreview.net/forum?id=qr3vj1PlsI},
}
Figure