[Paper Review] Downlink Performance and Capacity of Distributed Antenna Systems

Sina Firouzabadi, Andrea Goldsmith|arXiv (Cornell University)|Sep 14, 2011

Advanced MIMO Systems Optimization9 references18 citations

TL;DR

This paper establishes the ergodic downlink capacity of distributed antenna systems (DAS) with per-antenna power constraints, deriving closed-form capacity expressions under two CSI scenarios: perfect CSIT and only receiver CSI. It proves that beamforming is optimal with CSIT, while equal power allocation across antennas is optimal without CSIT, and demonstrates that optimal antenna placement significantly enhances system capacity and energy efficiency.

ABSTRACT

This paper investigates the performance of the downlink channel in distributed antenna systems. We first establish the ergodic capacity of distributed antennas, under different channel side information (CSI) assumptions. We consider a generalized distributed antenna system with $N$ distributed ports, each of which is equipped with an array of $L$ transmit antennas and constrained by a fixed transmit power. For this system we calculate the downlink capacity to a single antenna receiver, under different assumptions about the availability of the channel states at the transmitter. Having established this information theoretic analysis of the ergodic capacity of distributed antenna systems, this paper also investigates the effect of antenna placement on the performance of such systems. In particular, we investigate the optimal placement of the transmit antennas in distributed antenna systems. We present a fairly general framework for this optimization with no constraint on the location of the antennas. Based on stochastic approximation theory, we adopt a formulation that is suitable for node placement optimization in various wireless network scenarios. We show that optimal placement of antennas inside the coverage region can significantly improve the power efficiency of wireless networks.

Motivation & Objective

• To analyze the ergodic downlink capacity of distributed antenna systems under per-antenna power constraints, which are common in practical deployments.
• To investigate the impact of channel state information at the transmitter (CSIT) on system capacity, distinguishing between perfect CSIT and only receiver CSI.
• To determine the optimal placement of distributed antenna ports in a general, unconstrained geographic framework to maximize system performance and energy efficiency.
• To quantify the performance gains in capacity and power efficiency achievable through optimal antenna placement compared to conventional centralized antenna systems.

Proposed method

• Formulates a generalized DAS model with N distributed ports, each having L transmit antennas and a dedicated power budget S_m.
• Derives closed-form ergodic capacity expressions by solving a stochastic optimization problem under two CSI assumptions: perfect CSIT and only receiver CSI.
• Applies stochastic approximation theory to optimize antenna placement without topological constraints, enabling general deployment scenarios.
• Uses Jensen’s inequality and covariance bounds to derive upper bounds on the achievable SNR, proving optimality conditions for power allocation.
• Demonstrates that with perfect CSIT, beamforming per port maximizes capacity, while without CSIT, equal power allocation across antennas in each port is optimal.
• Validates the optimality of the derived solutions by showing that any deviation from symmetric power allocation leads to a contradiction in the objective function value.

Experimental results

Research questions

• RQ1 What is the ergodic downlink capacity of a distributed antenna system with per-antenna power constraints and different levels of channel state information at the transmitter?
• RQ2 How does the optimal transmission strategy—beamforming versus equal power allocation—depend on the availability of CSI at the transmitter?
• RQ3 What is the impact of optimal antenna placement on the capacity and energy efficiency of distributed antenna systems?
• RQ4 Can a closed-form solution be derived for the downlink capacity of DAS under per-antenna power constraints, and if so, under what conditions?
• RQ5 How does interference from neighboring cells affect the ergodic capacity when treated as noise in the DAS model?

Key findings

• With perfect CSIT, the optimal transmission strategy is beamforming at each distributed antenna port, maximizing the signal-to-noise ratio and achieving the highest possible ergodic capacity.
• When CSI is only available at the receiver, the optimal strategy is to allocate equal power across all antennas within each port, and cooperation between ports does not increase capacity.
• The ergodic capacity is derived in closed form for both CSIT and non-CSIT scenarios, providing a theoretical foundation for system design.
• Optimal placement of distributed antennas significantly improves system capacity and power efficiency, with gains that are highly sensitive to antenna location.
• The capacity gain from optimal placement can be substantial, especially in interference-limited scenarios, and is maximized when the system is designed with a joint optimization of power allocation and location.
• The derived capacity expressions are validated through mathematical proof using stochastic approximation and inequality constraints, showing that symmetric power allocation is optimal when CSIT is absent.