Success Probability and Area Spectral Efficiency in Multiuser MIMO HetNets

Chang Li, Jun Zhang, Jeffrey G. Andrews, Khaled Ben Letaief

Abstract

This paper investigates the performance of Multi-User MIMO (MU-MIMO) in heterogeneous networks (HetNets). While single-antenna networks exhibit "SIR invariance," where increasing base station density does not degrade signal quality, this study proves that this property fails in MU-MIMO environments. By employing stochastic geometry and a novel Toeplitz matrix representation, the authors characterize the fundamental tradeoff between Area Spectral Efficiency (ASE) and link reliability, providing optimization frameworks for 5G network deployment.

Introduction

The "more is better" philosophy in cellular network densification was historically supported by the Signal-to-Interference Ratio (SIR) invariance property in single-input single-output (SISO) systems. In such networks, the increased interference from a denser deployment of base stations (BSs) is perfectly offset by the reduced distance between users and their serving BSs. However, modern 5G architectures rely heavily on Multi-User MIMO (MU-MIMO) to serve multiple users simultaneously. This paper re-evaluates whether densification remains universally beneficial in these multi-antenna regimes.

System Model and Methodology

The authors model a K-tier HetNet where base stations of different types (e.g., macro, micro, pico) are distributed according to independent Poisson Point Processes (PPPs). Each tier is characterized by its own transmit power, antenna count, and number of multiplexed users.

A 3-tier HetNet model showing Macro, Micro, and Pico base stations serving mobile users via transmit links.

To analyze this complex environment, the researchers utilize stochastic geometry. A key technical innovation is the use of a lower triangular Toeplitz matrix to derive closed-form expressions for success probability. This approach simplifies the handling of gamma-distributed channel gains inherent in MIMO systems. Furthermore, an asymptotic analysis for high SIR regimes provides a computationally efficient tool for network planning.

The Departure from SIR Invariance

A central finding is that MIMO HetNets are not SIR-invariant. Unlike SISO systems, the success probability in a MIMO HetNet depends monotonically on the density of the base stations. Specifically, increasing the density of a particular tier can either enhance or degrade link reliability depending on the ratio of antennas to users in that tier. This implies that simply adding more hardware without considering the spatial and antenna configurations can be counterproductive.

The ASE-Reliability Tradeoff

The paper highlights a critical conflict between two key performance metrics:

  1. Area Spectral Efficiency (ASE): The total system capacity measured in bits per second per unit area.
  2. Link Reliability (Success Probability): The probability that an individual user's SIR exceeds a required threshold.

Maximizing ASE typically requires activating as many base stations as possible to maximize spatial reuse. However, this creates a significant interference floor that reduces the success probability for individual links. The authors demonstrate that to achieve high reliability, it may be necessary to deactivate certain tiers—a phenomenon they term the "One-Tier Paradox," where the highest reliability is often achieved by using only the single most efficient tier.

Optimization and Practical Implications

To navigate these tradeoffs, the authors propose two optimization algorithms. Algorithm 1 addresses simpler uniform SDMA networks, while Algorithm 2 utilizes Dinkelbach’s method and Sequential Convex Programming (SCP) to optimize BS densities in general MIMO HetNets.

Simulations of a 3-tier network confirm that the mathematical models are highly accurate for SIR thresholds above 10 dB. The results suggest that network operators must carefully tune BS densities and tier activation to balance the needs of high-capacity hotspots with the requirement for reliable wide-area coverage.

Conclusion

This research provides a rigorous mathematical foundation for understanding the scaling laws of MU-MIMO HetNets. By debunking the SIR invariance myth for multi-antenna systems, it offers a more nuanced view of network densification, emphasizing that the path to 5G and beyond requires a sophisticated balancing of capacity and reliability.

Summary updated

BibTeX
@article{li2016success,
  author = {Chang Li and Jun Zhang and Jeffrey G. Andrews and Khaled Ben Letaief},
  title = {{Success Probability and Area Spectral Efficiency in Multiuser MIMO HetNets}},
  journal = {IEEE Trans. Communications},
  year = {2016},
  doi = {10.1109/tcomm.2016.2531719},
}
Figure