|
Full title—System-Level Performance Analysis of Uplink Cell-Free Massive MIMO with L-MMSE Combining using Stochastic Geometry
This work investigates system-level performance of uplink cell-free massive MIMO (CF-mMIMO) networks with the local minimum-mean-squared-error (L-MMSE) combining, which is known to offer significant performance improvements over conjugate beamforming combining. Despite its advantages, the system-level performance of uplink CF-mMIMO networks with L-MMSE has been rarely studied due to its mathematical intractability.
The work analytically conducts the performance evaluation using stochastic geometry, which enables us to derive closed-form expressions for some performance metrics, i.e., coverage probability and potential throughput, in both interference-limited and noise-limited scenarios. The derived expressions explicitly elucidate how the system parameters influence the system-level performance.
Specifically, it is clearly demonstrated that for a given total number of antennas within a CF-mMIMO network, deploying as many access points with a single antenna as possible is optimal for both coverage probability and potential throughput in the noise-limited scenario, whereas the antenna distribution has no impact on the performance metrics in the interference-limited scenario. Moreover, this work reveals that in the noise-limited scenario, the potential throughput increases linearly with user density, while in the interference-limited scenario, it grows but eventually converges to a constant value. Comprehensive Monte-Carlo simulations validate our analytic derivations and insightful findings from the analysis.
Full Article: IEEE Transactions on Vehicular Technology, Early Access
| 
