Fronthaul-Limited Uplink OFDMA in Ultra-Dense CRAN with Hybrid Decoding

TL;DR

This paper proposes a hybrid decoding scheme for uplink OFDMA in ultra-dense CRANs, combining local decoding and joint processing to overcome fronthaul capacity limits, with optimized resource allocation.

Contribution

It introduces a novel hybrid decoding approach with joint optimization for uplink CRANs, improving throughput under fronthaul constraints.

Findings

Hybrid decoding outperforms traditional schemes.

Proposed algorithms achieve near-optimal throughput.

Significant throughput gains demonstrated in simulations.

Abstract

In an ultra-dense cloud radio access network (UD-CRAN), a large number of remote radio heads (RRHs), typically employed as simple relay nodes, are distributed in a target area, which could even outnumber their served users. However, one major challenge is that the large amount of information required to be transferred between each RRH and the central processor (CP) for joint signal processing can easily exceed the capacity of the fronthaul links connecting them. This motivates our study in this paper on a new hybrid decoding scheme where in addition to quantizing and forwarding the received signals for joint decoding at the CP, i.e. forward-and-decode (FaD) as in the conventional CRAN, the RRHs can locally decode-and-forward (DaF) the user messages to save the fronthaul capacity. In particular, we consider the uplink transmission in an orthogonal frequency division multiple access (OFDMA)-based UD-CRAN, where the proposed hybrid decoding is performed on each OFDMA sub-channel (SC). We study a joint optimization of the processing mode selections (DaF or FaD), user-SC assignments and the users' transmit power allocations over all SCs to maximize their weighted-sum-rate subject to the RRHs' individual fronthaul capacity constraints and the users' individual power constraints. Although the problem is non-convex, we propose a Lagrange duality based solution, which can be efficiently computed with good accuracy. Further, we propose a low-complexity greedy algorithm which is shown to achieve close to the optimal performance under practical setups. Simulation results show the promising throughput gains of the proposed designs with hybrid decoding, compared to the existing schemes that perform either DaF or FaD at all SCs.