Abstract:Cell-free massive multiple-input multiple-output (CFmMIMO) is a paradigm that can improve users' spectral efficiency (SE) far beyond traditional cellular networks. Increased spatial diversity in CFmMIMO is achieved by spreading the antennas into small access points (APs), which cooperate to serve the users. Sequential fronthaul topologies in CFmMIMO, such as the daisy chain and multi-branch tree topology, have gained considerable attention recently. In such a processing architecture, each AP must store its received signal vector in the memory until it receives the relevant information from the previous AP in the sequence to refine the estimate of the users' signal vector in the uplink. In this paper, we adopt vector-wise and element-wise compression on the raw or pre-processed received signal vectors to store them in the memory. We investigate the impact of the limited memory capacity in the APs on the optimal number of APs. We show that with no memory constraint, having single-antenna APs is optimal, especially as the number of users grows. However, a limited memory at the APs restricts the depth of the sequential processing pipeline. Furthermore, we investigate the relation between the memory capacity at the APs and the rate of the fronthaul link.

What problem does this paper attempt to address?

The main problem that this paper attempts to solve is in the cell - free massive MIMO (CFmMIMO) network, when access points (APs) adopt a sequential fronthaul architecture and have limited memory, how to optimize the compression and storage of the received signal vectors and study its impact on the user spectral efficiency (SE). Specifically, the paper focuses on the following aspects: 1. **Memory limitations in the sequential fronthaul architecture**: - In the sequential fronthaul architecture, each AP must store its received signal vector in memory until it receives relevant information from the previous AP in the sequence to improve the estimation of the user signal vector. This leads to limitations in memory capacity, especially when the AP is at the end of the sequence and needs to store a large number of received signal vectors. 2. **Selection of compression methods**: - The paper explores two compression methods: vector - wise compression and element - wise compression. To improve memory utilization efficiency, vector - wise compression is superior to element - wise compression, but in practice, element - wise compression is more often used to reduce complexity. For this reason, the paper proposes a pre - processing method, that is, using principal component analysis (PCA) to decorrelate the local received signal vectors, and then performing element - wise compression on the elements of the pre - processed vectors. 3. **Memory models and allocation strategies**: - The paper considers two memory models: fixed - per - AP memory and total - memory - fixed constraint. It studies the relationship between memory capacity and the maximum fronthaul link rate under these two models. 4. **Optimal number of APs and sequential fronthaul length**: - Through simulation results, it studies the impact of AP memory limitations on the optimal number of APs and compares the sequential fronthaul processing lengths under different memory allocation schemes (such as equal - allocation or linearly - increasing allocation). 5. **Cost of spatial diversity**: - Different from most of the existing advanced research, the paper points out that in the sequential processing topology, spatial diversity is not obtained for free. Since the end - AP needs to store a large number of received signal vectors, this introduces significant compression noise, which offsets the effect of increasing macro - diversity. In summary, this paper aims to solve the problem of limited AP memory in the CFmMIMO network under the sequential fronthaul architecture. By optimizing the compression methods and memory allocation strategies, it maximizes the average spectral efficiency of users.

Cell-free Massive MIMO with Sequential Fronthaul Architecture and Limited Memory Access Points

Sequential Processing in Cell-free Massive MIMO Uplink with Limited Memory Access Points

Compressive Massive Random Access for Massive Machine-Type Communications (mMTC).

Cell-free ISAC Massive MIMO Systems with Capacity-Constrained Fronthaul Links

Transmission strategies for cell-free massive mimo with limited capacity fronthaul links

Cell-Free Millimeter-Wave Massive MIMO Systems with Limited Fronthaul Capacity

Performance and Architectural Tradeoffs in Scalable Cell-Free Massive MIMO

Sequential In-Network Processing for Cell-Free Massive MIMO with Capacity-Constrained Parallel Radio Stripes

Cell-Free Massive MIMO with Mixed-Resolution ADCs and I/Q Imbalance Over Rician Spatially Correlated Channels

Performance Analysis on Cell-Free Massive MIMO With Capacity-Constrained Fronthauls and Variable-Resolution ADCs

Cell-Free Massive MIMO versus Small Cells

Hierarchical Cell-Free Massive MIMO for High Capacity with Simple Implementation

A Review on Cell-Free Massive MIMO Systems

Cluster-Based Cell-Free Massive MIMO Systems: A Novel Framework to Enhance Spectral Efficiency with Low Complexity

Enabling Cell-Free Massive MIMO Systems with Wireless Millimeter Wave Fronthaul

Joint User Association and Power Control for Cell-Free Massive MIMO

Front Haul Multi-Homing for Cell-Free Massive MIMO: A Low-Latency Approach With Distributed and Hierarchical CPU Deployment

Joint Sequential Fronthaul Quantization and Hardware Complexity Reduction in Uplink Cell-Free Massive MIMO Networks

Foundations of User-Centric Cell-Free Massive MIMO

Co-existing/Cooperating Multicell Massive MIMO and Cell-Free Massive MIMO Deployments: Heuristic Designs and Performance Analysis