Source-linked AI summary
A Survey of Energy-Efficient Techniques for 5G Networks and Challenges Ahead
Stefano Buzzi, Chih-Lin I, Thierry E. Klein, H. Vincent Poor, Chenyang Yang, Alessio Zappone
TL;DR
5G networks face unprecedented device and traffic growth while increasing capacity through transmit-power scaling would be economically and environmentally unsustainable. This survey reviews energy-efficient wireless communications across resource allocation, network technologies, hardware, energy transfer, and emerging techniques, and concludes that energy efficiency has become a key design measure while major challenges remain before the 1000-times improvement goal can be reached.
Problem
5G must support unprecedented device and traffic growth without relying on transmit-power scaling that produces unacceptable operating costs and increasing CO2 emissions.
Method
The survey synthesizes seminal and recent energy-efficient wireless communications research across resource allocation, hardware, network technologies, energy harvesting and transfer, and emerging techniques.
Results
Energy efficiency has gained its own role as a performance measure and design constraint for communication networks, but technical, regulatory, policy, and business challenges remain before the 1000-times improvement goal can be reached.
Takeaways & Limitations
A holistic approach combining multiple energy-efficient techniques is necessary because separate treatment of resource allocation, deployment, harvesting, and transfer is unlikely to achieve the desired improvement.
Takeaways & Limitations
Quantifying the energy-efficiency impact of caching and mobile computing requires new energy-consumption models that account for their relevant consumption.
Abstract
from arXiv · showhide
After about a decade of intense research, spurred by both economic and operational considerations, and by environmental concerns, energy efficiency has now become a key pillar in the design of communication networks. With the advent of the fifth generation of wireless networks, with millions more base stations and billions of connected devices, the need for energy-efficient system design and operation will be even more compelling. This survey provides an overview of energy-efficient wireless communications, reviews seminal and recent contribution to the state-of-the-art, including the papers published in this special issue, and discusses the most relevant research challenges to be addressed in the future.
I. INTRODUCTION
5G must deliver dramatically greater capacity for unprecedented device and traffic growth without relying on unsustainable increases in transmit power and energy consumption. The survey organizes energy-efficiency approaches into four categories and reviews the state of the art across them.
- I. INTRODUCTION: 5G networks must support more than 50 billion connected devices and 1000 times the capacity of present cellular systems.Projected services include smart homes, cities, cars, telesurgery, and advanced security, with traffic reaching tens of Exabytes per month.
- I. INTRODUCTION: Scaling transmit power to increase capacity is economically unsustainable because it would produce unacceptable operating costs.Present wireless techniques cannot provide the desired capacity increase merely by increasing transmit powers.
- I. INTRODUCTION: Wireless networks also require energy efficiency because ICT systems produce 5% of global CO2 emissions and the wireless share is expected to reach 75%.The GSMA additionally demands more than a 40% reduction in CO2 emissions per connection by 2020.
- I. INTRODUCTION: The survey identifies resource allocation, network planning and deployment, energy harvesting and transfer, and hardware solutions as four broad energy-efficiency categories.Examples include energy-efficiency-oriented resource allocation, traffic-adaptive base-station switching, environmental energy harvesting, and cloud-based radio access networks.
- I. INTRODUCTION: The paper surveys the state of the art in these four categories, with special focus on contributions published in the issue.The stated goal is to review energy-efficient wireless communication techniques and their role in addressing the 5G energy challenge.
II. RESOURCE ALLOCATION
Energy-efficient resource allocation optimizes reliably transmitted information per Joule rather than throughput alone. It models consumed energy through radiated and static power, selects benefit metrics according to channel information and service requirements, and supports network-wide or node-level objectives.
- Energy-efficient allocation optimizes reliably transmitted information per Joule, requiring radio resources to target energy efficiency rather than throughput alone.
- Consumed energy includes radiated power, amplifier losses, and static hardware dissipation; under common assumptions, it is modeled using radiated power and a static term.The static term covers transmitter and receiver circuit blocks.
- Benefit functions measure reliably transmitted data using capacity, achievable rate, throughput, or outage capacity, depending on the system and available channel information.Outage capacity is suited to slow-fading scenarios, while ergodic metrics apply when only statistical channel information is available.
- Energy efficiency is nonmonotone in transmit power and reaches its maximum at a finite power level, unlike traditional metrics that increase monotonically with power.
- Increasing static power shifts the energy-efficiency maximizer upward; when static power dominates radiated power, maximizing energy efficiency approaches numerator maximization.
- Network objectives include Global Energy Efficiency, which maximizes aggregate benefit per total power, and multi-objective combinations that tune individual node efficiencies.Weighted sum, product, and minimum objectives characterize parts of the energy-efficient Pareto boundary, while weighted minimum energy efficiency can characterize the complete boundary.
- Energy-efficiency optimization can incorporate maximum-power, rate, delay, and interference-temperature constraints, whereas energy minimization requires QoS constraints to avoid the trivial zero-power solution.Energy minimization targets minimum energy for required performance but does not directly optimize the benefit-cost ratio.
III. NETWORK PLANNING AND DEPLOYMENT
The survey identifies disruptive technologies for planning, deployment, and operation as responses to the sheer number of connected devices expected in 5G networks.
- Several potentially disruptive technologies have been proposed to address the sheer number of connected devices in 5G network planning, deployment, and operation.
A. Dense networks
Dense networks address the growing number of devices by deploying more infrastructure equipment, particularly through heterogeneous densification approaches.
- A. Dense networks: Dense networks respond to the increasing number of devices by increasing the amount of deployed infrastructure equipment.
- A. Dense networks: Dense heterogeneous networks place many heterogeneous nodes, from macro base stations to femtocells and relays, per unit area in demand-based deployments.
1) Dense Heterogeneous Networks:
Dense heterogeneous networks increase infrastructure-node density through demand-based deployment, creating irregular layouts whose energy-efficiency benefits involve a densification trade-off.
- 1) Dense Heterogeneous Networks:: Dense heterogeneous networks deploy many macro base stations, femtocells, and relays opportunistically according to demand, producing irregular network layouts.This contrasts with traditional deployments that divide a macro-cell into relatively few smaller areas.
- 1) Dense Heterogeneous Networks:: Node densification reduces communication distances and can increase data rates at lower transmit powers.
- 1) Dense Heterogeneous Networks:: Densification also creates additional interference that may degrade network energy efficiency.
- 1) Dense Heterogeneous Networks:: The energy-efficiency benefit of densification saturates as infrastructure-node density increases, indicating an optimal density level.
2) Massive MIMO:
Massive MIMO replaces small antenna arrays with hundreds of low-cost antenna elements, offering radiated-power savings while introducing analysis, estimation, and hardware challenges.
- 2) Massive MIMO:: Very large antenna arrays require limiting-behavior analysis, while channel estimation is complicated by pilot contamination and hardware impairments.
- 2) Massive MIMO:: Radiated power decreases proportionally to the square root of the number of deployed antennas while information rate remains unchanged in an ideal single-cell system.The stated result excludes hardware-consumed power.
- 2) Massive MIMO:: Network energy efficiency is maximized at a finite number of deployed antennas when hardware power is included.
- 2) Massive MIMO:: Recent special-issue contributions address self-organizing cells, traffic-responsive deployment, and energy-efficient resource management in heterogeneous networks.
B. Offloading techniques
Offloading techniques shift traffic or communication functions across nearby devices, alternative radio technologies, caches, visible light, and mmWave links to support capacity and energy-efficiency goals.
- B. Offloading techniques: D2D communications let nearby devices communicate directly under base-station instruction, potentially using much lower transmit power than cellular transmission.
- B. Offloading techniques: Visible-light communications use LEDs for short-range indoor links, offering high energy efficiency, large bandwidth, and high data rates.Demonstrated rates include 3.5 Gbit/s at 2 m and 1.1 Gbit/s at 10 m with 5 mW optical output power.
- B. Offloading techniques: Local caching stores popular content at base stations during light-load periods, reducing repeated backhaul transmissions and potentially improving core-network energy efficiency.
- B. Offloading techniques: mmWave bands above 10 GHz provide short-range communications up to 100–200 m and can offload traffic from sub-6 GHz cellular frequencies.
- B. Offloading techniques: Energy-aware D2D research studies savings and trade-offs involving bandwidth, buffer size, and service delay using mixed-integer linear programming.
- B. Offloading techniques: Energy-efficient caching gains are larger with stringent backhaul capacity, low interference, skewed content popularity, and pico-cell caching.
IV. ENERGY HARVESTING AND TRANSFER
Energy harvesting enables wireless networks to use renewable, clean, or recycled energy, while energy randomness motivates offline, stochastic, learning-based, and energy-sharing approaches.
- IV. ENERGY HARVESTING AND TRANSFER: Environmental harvesting obtains clean energy from natural sources such as sunlight and wind, whereas RF harvesting recycles energy from radio signals and interference.
- IV. ENERGY HARVESTING AND TRANSFER: Random energy availability is a central design challenge for harvesting-powered communication systems.
- IV. ENERGY HARVESTING AND TRANSFER: Energy used at time t cannot exceed the energy harvested up to time t.
- IV. ENERGY HARVESTING AND TRANSFER: Early environmental-harvesting studies used offline policies that assumed future harvested energy was known, including directional waterfilling and finite-battery extensions.
- IV. ENERGY HARVESTING AND TRANSFER: Online designs avoid specific-time harvest knowledge by using stochastic optimization with known energy statistics or learning-theory approaches.
- IV. ENERGY HARVESTING AND TRANSFER: Combining RF harvesting with wireless power transfer allows nodes to share energy, redistribute total network energy, and prolong low-battery-node lifetimes.
V. HARDWARE SOLUTIONS
Energy-efficient hardware spans RF-chain and transceiver simplification, hybrid beamforming, and cloud-based RAN architectures. The survey also highlights traffic-aware C-RAN management, which can switch low-load base stations off to reduce energy consumption.
- Energy-efficient hardware includes greener RF chains, simplified transmitter/receiver structures, and cloud RAN architectures using software-based network functions.
- One-bit quantization and hybrid analog/digital beamformers are proposed to improve hardware energy efficiency, especially in systems with many antennas.
- Large mmWave antenna arrays make fully digital beamforming costly in complexity and energy, motivating hybrid analog/digital structures.
- SIC-based hybrid precoding with a sub-connected architecture is near-optimal in simulations and more energy-efficient than spatially sparse and fully digital precoding.
- C-RAN transfers base-station functions to remote data centers for software implementation, supporting more energy-efficient network operation.
- Learning-based traffic prediction lets C-RAN switch inactive or low-load base stations off, reducing overall energy consumption on real traffic traces.
VI. FUTURE RESEARCH CHALLENGES
The survey turns from reviewed 5G energy-efficient techniques to the next research steps needed for energy-efficient networks. It presents future challenges as the subject of the following discussion.
- After reviewing current 5G energy-efficient techniques, the survey asks what research steps should follow toward energy-efficient 5G networks.
A. The need for a holistic approach
The survey argues that energy-efficient technologies have mostly been studied separately, whereas achieving the desired improvement requires examining their integration. It also notes that interference-limited 5G makes direct fractional-programming applications prohibitively complex.
- A. The need for a holistic approach: Resource allocation, deployment and planning, energy harvesting, and energy transfer have mostly been studied separately rather than through a single integrated approach.
- A. The need for a holistic approach: The survey questions whether separate techniques can achieve a thousand-fold energy-efficiency increase and states that a holistic approach is likely necessary.
- A. The need for a holistic approach: Fractional programming is identified as suitable for energy-efficient resource allocation, but its direct application is typically prohibitively complex in interference-limited networks.
- A. The need for a holistic approach: 5G networks are expected to be interference-limited because orthogonal transmission and linear interference neutralization are impractical for serving massive numbers of nodes.
C. Dealing with randomness
Future wireless systems will contain substantial randomness in topology, traffic, and energy availability, but existing statistical and learning approaches have limited energy-efficiency analysis. Emerging techniques such as caching and mobile computing also require richer energy models, while broader technical and policy challenges remain.
- C. Dealing with randomness: Randomness will pervade future network topologies, traffic evolution, and energy availability, requiring new statistical models for energy-efficient design.
- C. Dealing with randomness: Random matrix theory and stochastic geometry are promising tools, but their energy-efficiency impact remains insufficiently investigated.
- C. Dealing with randomness: Learning techniques can let devices respond to observations in a self-organizing manner, yet their effects on energy-efficient network design remain little studied.
- C. Dealing with randomness: Caching can reduce energy consumption by distributing frequently accessed content and avoiding backhaul transmissions, while mobile computing can prolong the lifetime of battery-limited nodes.
- C. Dealing with randomness: Quantifying emerging techniques requires models that include backhaul overhead, feedback signaling, and digital-signal-processor computation energy.
- C. Dealing with randomness: Reaching the ambitious 1000-times energy-efficiency goal still involves technical, regulatory, policy, and business challenges.