Optimal design of energy-efficient HetNets: joint precoding and load balancing

Jingya Li, Emil Björnson, Tommy Svensson, Thomas Eriksson, Mérouane Debbah

TL;DR

This paper tackles the challenge of optimizing energy efficiency in heterogeneous wireless networks (HetNets) by jointly designing how data is transmitted (precoding) and how users are distributed among base stations (load balancing). They propose a method to intelligently put base stations into a low-power "sleep" mode, significantly reducing overall energy consumption without sacrificing user quality of service.

What Problem Does This Paper Address?

Future wireless networks are increasingly adopting a "heterogeneous" architecture, where a mix of large macro base stations (MBSs) and numerous small, low-power base stations (SBSs) coexist. This dense deployment of SBSs can improve network capacity and user experience by bringing coverage closer to users. However, this also leads to a significant increase in the total number of active base stations, and consequently, a substantial rise in overall energy consumption. This energy consumption has two main components: the dynamic power used for transmitting signals, and the static "circuit" power consumed by the base station hardware even when idle or in a low-power state. The challenge is to design network operation strategies that minimize total energy consumption while still meeting user demands for data rates (Quality of Service or QoS). Simply having all base stations active, even if they are low-power, can be less energy-efficient than strategically activating only a subset of them. This paper aims to find the optimal way to manage these base stations, deciding which ones should be active and how to best transmit data to users, to achieve maximum energy savings.

Methods

The authors formulate the problem as a joint optimization of precoding vectors (how to shape the transmitted signals) and load balancing (which users are served by which base stations). They also incorporate the decision of whether each base station should be in an "active" or "sleep" mode.

  • Optimal Solution: They first show that for any fixed combination of base station modes (active/sleep), the problem has a "hidden convexity" structure, allowing for an optimal solution. The global optimum is then found by exhaustively searching through all 2<sup>M</sup> possible mode combinations (where M is the number of base stations). This exhaustive search provides an upper bound for performance.
  • Heuristic Algorithm: To overcome the computational complexity of the exhaustive search, they propose an iterative heuristic algorithm. This algorithm uses convex approximations of the non-convex power consumption functions. In each iteration, it solves a relaxed convex problem, gradually driving base stations with low transmit power towards the sleep mode. This approach aims to find a local optimum efficiently.
  • Key Techniques: The paper utilizes concepts from convex optimization, semi-definite relaxation (SDR), and iterative approximation techniques. They also model imperfect channel state information, which is a realistic scenario in wireless networks.

Datasets

Not discussed in the paper. The evaluation is based on simulations using a defined propagation model and network deployment.

Key Insights and Ideas

  • Sleep Mode is Crucial: The most significant insight is that strategically putting base stations into a sleep mode, through proper load balancing, is a highly effective strategy for achieving substantial energy savings in heterogeneous networks.
  • Joint Optimization is Key: Optimizing precoding, load balancing, and base station activation simultaneously is essential for maximizing energy efficiency. Treating these aspects in isolation would likely lead to suboptimal results.
  • Trade-off Between Circuit and Dynamic Power: The paper highlights the critical trade-off between dynamic transmit power (reduced by shorter distances with SBSs) and circuit power (increased by the sheer number of base stations). Activating fewer base stations can significantly reduce circuit power, often outweighing any increase in dynamic power.
  • QoS Drives Activation: The probability of a base station being active is directly influenced by the target Quality of Service (QoS) requirements. Higher QoS demands necessitate more active base stations to ensure reliable service.

Key Results

  • The proposed "Optimal" and "Heuristic" schemes significantly reduce total power consumption compared to an "All Active" scheme, especially at lower target spectral efficiency requirements.
  • For instance, at a target spectral efficiency of 1 bit/s/Hz, the "Optimal" scheme can achieve approximately 20% lower total power consumption than the "All Active" scheme.
  • The "Optimal" scheme generally outperforms the "Heuristic" scheme in terms of energy savings, though the heuristic provides a good approximation with much lower complexity.
  • The BS activation probability increases with the target spectral efficiency, as expected, to meet higher QoS demands. The "Optimal" scheme tends to have lower activation probabilities for SBSs compared to the "Heuristic" scheme at lower QoS levels.

Limitations and Future Work

  • Computational Complexity of Optimal Solution: The exhaustive search for the global optimum is computationally prohibitive for large networks (2<sup>M</sup> combinations). This is why the heuristic algorithm is crucial.
  • Heuristic is Local Optimum: The proposed heuristic algorithm guarantees a local optimum, not necessarily the global optimum.
  • Future Work: The paper suggests exploring more advanced heuristic algorithms that can achieve near-optimal performance with even lower complexity. Further investigation into the impact of different BS types and deployment scenarios on energy efficiency would also be valuable. The paper also mentions that the BS activation probability depends on the ratio between active and sleep mode circuit power, suggesting further analysis in this area.

Summary updated

Abstract

This paper considers the downlink of a heterogeneous network, where multiple base stations (BSs) can serve the users by non-coherent multiflow beamforming. We assume imperfect channel state information at both BSs and users. The objective is to jointly optimize the precoding, load balancing, and BS operation mode (active or sleep) for improving the energy efficiency of the network. The considered problem is to minimize the weighted total power consumption (both circuit power and dynamic transmit power), while satisfying per-user quality of service constraints and per-BS transmit power constraints. This problem is non-convex, but we prove that for each combination of BS modes, the considered problem has a hidden convexity structure. Thus, the global optimal solution is obtained by an exhaustive search over all possible BS mode combinations. Furthermore, by iterative convex approximations of the non-convex power consumption functions, a heuristic algorithm is proposed to obtain a local optimal solution with low complexity. Simulation results illustrate that our proposed algorithms significantly reduce the total power consumption, compared to the scheme where all BSs are continuously active. This implies that putting a BS into sleep mode by proper load balancing is an important solution for energy savings in heterogeneous networks.
BibTeX
@inproceedings{li2015optimal,
  author = {Jingya Li and Emil Björnson and Tommy Svensson and Thomas Eriksson and Mérouane Debbah},
  title = {{Optimal design of energy-efficient HetNets: joint precoding and load balancing}},
  booktitle = {ICC},
  year = {2015},
  doi = {10.1109/icc.2015.7249059},
}
Figure