Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Unicast-Multicast Cell-Free Massive MIMO: Gradient-Based Resource Allocation

Mustafa S. Abbas, Zahra Mobini, Hien Quoc Ngo, Michail Matthaiou

TL;DR

Simulation results show that the joint optimization strategy enhances notably the sum SE (SSE) (up to 58%) compared to baseline schemes, while maintaining low complexity.

Abstract

We consider a cell-free massive multiple-input multiple-output (CF-mMIMO) system with joint unicast and multi-group multicast transmissions. We derive exact closed-form expressions for the downlink achievable spectral efficiency (SE) of both unicast and multicast users. Based on these expressions, we formulate a joint optimization problem of access point (AP) selection and power control subject to quality of service (QoS) requirements of all unicast and multicast users and per-AP maximum transmit power constraint. The challenging formulated problem is transformed into a tractable form and a novel accelerated projected gradient (APG)-based algorithm is developed to solve the optimization problem. Simulation results show that our joint optimization strategy enhances notably the sum SE (SSE) (up to 58%) compared to baseline schemes, while maintaining low complexity.

Unicast-Multicast Cell-Free Massive MIMO: Gradient-Based Resource Allocation

TL;DR

Simulation results show that the joint optimization strategy enhances notably the sum SE (SSE) (up to 58%) compared to baseline schemes, while maintaining low complexity.

Abstract

We consider a cell-free massive multiple-input multiple-output (CF-mMIMO) system with joint unicast and multi-group multicast transmissions. We derive exact closed-form expressions for the downlink achievable spectral efficiency (SE) of both unicast and multicast users. Based on these expressions, we formulate a joint optimization problem of access point (AP) selection and power control subject to quality of service (QoS) requirements of all unicast and multicast users and per-AP maximum transmit power constraint. The challenging formulated problem is transformed into a tractable form and a novel accelerated projected gradient (APG)-based algorithm is developed to solve the optimization problem. Simulation results show that our joint optimization strategy enhances notably the sum SE (SSE) (up to 58%) compared to baseline schemes, while maintaining low complexity.
Paper Structure (10 sections, 1 theorem, 50 equations, 2 figures, 1 algorithm)

This paper contains 10 sections, 1 theorem, 50 equations, 2 figures, 1 algorithm.

Table of Contents

  1. Introduction
  2. System Model
  3. Uplink Training
  4. Downlink Data Transmission and SE Analysis
  5. Problem Formulation and APG Optimization
  6. APG Method and Problem Reformulation
  7. Gradient of $g(\mathbf{\vartheta})$
  8. Projection onto $\widehat{\mathcal{C}}$
  9. Numerical Results
  10. Conclusion

Key Result

Proposition 1

The SE expressions for the $u$-th unicast user and $k_m$-th multicast users are given by $\mathrm{SE}_{u} = \frac{T-\tau}{T} \log_{2}({1+\mathrm{SINR}_u} )$ and $\mathrm{SE}_{m,k} = \frac{T-\tau}{T} \log_{2}({1+ \mathrm{SINR}_{m,k}} ),$ respectively, where the closed-form expressions for the receiv

Figures (2)

  • Figure 1: CDF of the SSE where the total number of unicast and multicast users is $28$ ($U=16$, $M=3$, $K_m=4$, $N=100$ and $\bar{SE}_{QoS}=SE_{QoS}=0.5$ bit/s/Hz).
  • Figure 2: Average SSE versus the number of APs for different number of multicast users ($U=5$ and $\bar{SE}_{QoS}=SE_{QoS}=0.4$ bit/s/Hz).

Theorems & Definitions (2)

  • Proposition 1
  • Proof 1