[Paper Review] Downlink and Uplink Energy Minimization Through User Association and Beamforming in Cloud RAN

Shixin Luo, Rui Zhang+1 more|arXiv (Cornell University)|Feb 18, 2014

Advanced MIMO Systems Optimization32 references4 citations

TL;DR

This paper proposes a joint downlink (DL) and uplink (UL) user association and beamforming design in Cloud RAN to minimize total network energy consumption. By leveraging uplink-downlink duality, the problem is transformed into an equivalent DL optimization with two interrelated subproblems, enabling two efficient algorithms that significantly reduce energy use compared to benchmarks, especially in asymmetric UL/DL scenarios.

ABSTRACT

The cloud radio access network (C-RAN) concept, in which densely deployed access points (APs) are empowered by cloud computing to cooperatively support mobile users (MUs), to improve mobile data rates, has been recently proposed. However, the high density of active ("on") APs results in severe interference and also inefficient energy consumption. Moreover, the growing popularity of highly interactive applications with stringent uplink (UL) requirements, e.g. network gaming and real-time broadcasting by wireless users, means that the UL transmission is becoming more crucial and requires special attention. Therefore in this paper, we propose a joint downlink (DL) and UL MU-AP association and beamforming design to coordinate interference in the C-RAN for energy minimization, a problem which is shown to be NP hard. Due to the new consideration of UL transmission, it is shown that the two state-of-the-art approaches for finding computationally efficient solutions of joint MU-AP association and beamforming considering only the DL, i.e., group-sparse optimization and relaxed-integer programming, cannot be modified in a straightforward way to solve our problem. Leveraging on the celebrated UL-DL duality result, we show that by establishing a virtual DL transmission for the original UL transmission, the joint DL and UL optimization problem can be converted to an equivalent DL problem in C-RAN with two inter-related subproblems for the original and virtual DL transmissions, respectively. Based on this transformation, two efficient algorithms for joint DL and UL MU-AP association and beamforming design are proposed, whose performances are evaluated and compared with other benchmarking schemes through extensive simulations.

Motivation & Objective

• Address the energy efficiency challenge in Cloud RAN due to dense deployment of access points (APs), which increases interference and power consumption.
• Formulate a joint downlink and uplink user association and beamforming problem to minimize total network energy consumption while satisfying QoS requirements.
• Overcome the limitations of prior DL-only optimization methods that fail to account for uplink transmission constraints and asymmetries between DL and UL.
• Design computationally efficient algorithms that jointly optimize user-AP association, beamforming, and active AP selection for both DL and UL.
• Evaluate tradeoffs between active AP power and mobile user uplink power to achieve optimal energy efficiency in C-RAN.

Proposed method

• Leverage the uplink-downlink (UL-DL) duality principle to transform the original joint DL and UL optimization problem into an equivalent DL problem with two interrelated subproblems: one for the original DL and one for a virtual DL corresponding to the UL transmission.
• Propose two efficient algorithms based on group-sparse optimization (GSO) and relaxed-integer programming (RIP) techniques to solve the transformed problem.
• Use ℓ₁,₂ and ℓ₁,∞ norm penalties to promote sparsity in user-AP association, thereby minimizing the number of active APs and reducing energy consumption.
• Model the total energy consumption as the sum of transmit power at APs and MUs, plus static circuit power at APs, and optimize this metric under SINR constraints.
• Apply iterative optimization techniques to solve the non-convex, NP-hard problem by decomposing it into tractable subproblems using duality and penalty relaxation.
• Validate the algorithms via extensive simulations under various network conditions, including varying numbers of users and APs, and different static power levels.

Experimental results

Research questions

• RQ1 How can joint downlink and uplink user association and beamforming be optimized to minimize total energy consumption in Cloud RAN?
• RQ2 Why do existing DL-only optimization techniques such as group-sparse optimization and relaxed-integer programming fail to extend effectively to joint DL and UL scenarios?
• RQ3 Can uplink-downlink duality be exploited to transform a complex joint DL-UL problem into an equivalent, more tractable DL problem?
• RQ4 What is the performance gain of the proposed algorithms in terms of energy efficiency and power consumption tradeoffs compared to exhaustive search and joint processing benchmarks?
• RQ5 How do the choice of sparsity-inducing penalty norms (ℓ₁,₂ vs. ℓ₁,∞) affect the performance of the proposed algorithms?

Key findings

• The proposed algorithms achieve sum-power consumption performance very close to the optimal exhaustive search (ES) benchmark, especially in low-to-moderate user scenarios.
• Compared to joint processing (JP), which activates all APs, the proposed algorithms reduce total sum-power consumption by up to 60% in simulations with K=4 users and N=6 APs.
• Algorithm I (based on group-sparse optimization) consistently outperforms Algorithm II (based on relaxed-integer programming), with a performance gap of approximately 10–15% in sum-power reduction.
• The choice of sparsity penalty (ℓ₁,₂ vs. ℓ₁,∞) has minimal impact on performance, indicating robustness of the GSO-based approach.
• As the weight parameter λ increases, the proposed algorithms shift energy consumption from mobile users to active APs, achieving a favorable trade-off that approaches the optimal ES solution.
• The algorithms maintain feasibility and reliability across diverse channel realizations and are effective even under asymmetric UL/DL conditions, unlike DL-only methods.