Abstract
This paper investigates the compress-and-forward scheme for an uplink cloud radio access network (C-RAN) model, where multi-antenna base-stations (BSs) are connected to a cloud-computing based central processor (CP) via capacity-limited fronthaul links. The BSs compress the received signals with Wyner-Ziv coding and send the representation bits to the CP; the CP performs the decoding of all the users' messages. Under this setup, this paper makes progress toward the optimal structure of the fronthaul compression and CP decoding strategies for the compress-and-forward scheme in C-RAN. On the CP decoding strategy design, this paper shows that under a sum fronthaul capacity constraint, a generalized successive decoding strategy of the quantization and user message codewords that allows arbitrary interleaved order at the CP achieves the same rate region as the optimal joint decoding. Further, it is shown that a practical strategy of successively decoding the quantization codewords first, then the user messages, achieves the same maximum sum rate as joint decoding under individual fronthaul constraints. On the joint optimization of user transmission and BS quantization strategies, this paper shows that if the input distributions are assumed to be Gaussian, then under joint decoding, the optimal quantization scheme for maximizing the achievable rate region is Gaussian. Moreover, Gaussian input and Gaussian quantization with joint decoding achieve to within a constant gap of the capacity region of the Gaussian multiple-input multiple-output (MIMO) uplink C-RAN model. Finally, this paper addresses the computational aspect of optimizing uplink MIMO C-RAN by showing that under fixed Gaussian input, the sum rate maximization problem over the Gaussian quantization noise covariance matrices can be formulated as convex optimization problems, thereby facilitating its efficient solution.
Introduction
In the evolving landscape of wireless communications, Cloud Radio Access Network (C-RAN) has emerged as a key architecture for meeting the demands of 5G and beyond. In a C-RAN, simple base stations (BSs) act as "remote radio heads," capturing signals from mobile users and forwarding them to a powerful, cloud-based Central Processor (CP). This centralized approach allows for advanced interference management and cooperative processing. However, the performance of C-RAN is fundamentally limited by the capacity of the "fronthaul" links that connect the BSs to the CP.
To address this bottleneck, BSs must compress their received signals before transmission. The standard approach is the "compress-and-forward" scheme, where BSs utilize Wyner-Ziv coding to exploit spatial correlations between signals received at different locations. This paper tackles three critical challenges in this domain: the complexity of decoding strategies at the CP, the optimality of Gaussian signaling and quantization, and the computational tractability of network optimization.
System Model
The study considers an uplink MIMO C-RAN model consisting of $K$ multi-antenna users and $L$ multi-antenna base stations. Each BS is connected to the CP via a dedicated, capacity-limited fronthaul link. The signals received at the BSs are corrupted by additive noise and inter-user interference, as depicted in the system model below.
Decoding Strategies: Joint vs. Successive
A primary concern in C-RAN design is the decoding complexity at the CP. While joint decoding (JD) of all quantization bits and user messages is theoretically optimal, it is computationally intensive. The authors investigate whether simpler successive decoding (SD) strategies can achieve comparable performance.
Using tools from submodular optimization and the properties of polyhedra, the paper demonstrates that a "generalized successive decoding" strategy—which allows for an arbitrary interleaved decoding order of quantization and user codewords—achieves the exact same rate region as joint decoding under a sum fronthaul capacity constraint. Furthermore, the authors prove that a practical successive decoding strategy (decoding all quantization codewords first, then all user messages) is sufficient to maximize the total sum rate under individual fronthaul constraints. This result significantly simplifies the receiver architecture without sacrificing throughput.
Optimality of Gaussian Signaling
While Gaussian distributions are often the default in wireless research, their optimality in multi-user, multi-relay C-RAN environments was not previously established. This paper provides a rigorous proof using Fisher Information and the de Bruijn identity: if the user inputs are Gaussian, then Gaussian quantization is the mathematically optimal compression strategy for maximizing the achievable rate region under joint decoding.
Moreover, the authors show that the combination of Gaussian input and Gaussian quantization achieves a rate within a "constant gap" of the true information-theoretic capacity of the Gaussian MIMO uplink C-RAN. This gap, which depends on the number of antennas and BSs, remains constant regardless of the signal-to-noise ratio (SNR), providing strong theoretical justification for the use of Gaussian strategies in practical designs.
Computational Tractability via Convex Optimization
Optimizing the quantization parameters in C-RAN typically involves solving non-convex problems that are difficult to scale. The authors overcome this by introducing a convex reparameterization. By shifting the optimization variable from the quantization noise covariance matrix to a related matrix $B_\ell$, the sum-rate maximization problem (for a fixed Gaussian input) is transformed into a convex optimization problem. This allows the network parameters to be optimized efficiently using standard numerical solvers.
Conclusion and Future Work
This research establishes a theoretical foundation for the design of efficient uplink C-RANs. By proving the equivalence of successive and joint decoding for sum-rate maximization and establishing the optimality and tractability of Gaussian strategies, the paper bridges the gap between information theory and practical network engineering.
However, some challenges remain. The current convex optimization framework assumes fixed user transmit covariances; jointly optimizing both user transmission and BS quantization remains a non-convex challenge. Additionally, the model assumes perfect Channel State Information (CSI), which is difficult to maintain in highly dynamic real-world environments. Future work will likely focus on extending these results to scenarios with imperfect CSI and broader rate region characterizations under individual fronthaul constraints.
Summary updated