Energy Efficient Wireless Communication Survey

John Henroid (A paper written under the guidance of Prof. Raj Jain) DownloadPDF


Information and communication technology (ICT) is one of fastest growing areas is wireless communications. The number of mobile devices in particular have grown at an exponential rate. Current network architectures are insufficient for supporting the growth and effectively adjusting to areas with different traffic demands. Small scale network deployments that can adapt to traffic at different levels has shown to reduce energy consumed while still maintaining high quality of service (QoS). Since greenhouse gasses have increased with the communication networks, energy efficiency will be a big concern as wireless adoption increase in developing countries. This paper examines different strategies of maximizing energy efficiency to handle the growth of wireless traffic.

Keywords: Energy Efficiency, Small Scale Networks, Wireless Communication, Massive MIMO, Adaptive Control

Table of Contents:

1. Introduction

Research has shown that there is expected to be a 10x increase of wireless data from 2015 to 2021 [Ericsson15]. Trends show that wired communications are slowing down while wireless traffic is rising at an exponential rate [Cisco16]. As use of mobile devices increase, the main concern is how long the device lasts and not access to the network. Energy efficiency is usually defined as the number of bits that can be sent over a unit of power consumption which is usually quantified by bits per Joule. The determining factor of energy efficiency for mobile devices is the power needed to transmit data.

We will begin by introducing the different factors that have shown improvement in energy efficiency. In Section 2 we will discuss the main energy consumer of energy in wireless systems, the radio. In Section 3 we will introduce new ways of increasing efficiency through new network topologies. In Section 4 we will present ideas on antenna diversity through massive multi-input multi-output schemes (MIMO). In Section 5 we will put forth adaptive control schemes in regards to energy efficiency. In Section 6 we will show how some of the best ideas with wireless energy efficiency can be implemented in smart grids.

2. Wireless Radios

At the simplest level, a typical wireless device consists of a radio unit and a baseband unit. The radio unit is responsible for sending and receiving data over a radio while the baseband unit is responsible for processing the data that is to be sent or received. It has been shown that the radio consumes around 57% of the energy in wireless communication due to the need of power amplification of signals [Hu14]. Reducing the number of radios in a wireless network is one way to improve the energy efficiency of a network. A typical radio network can be seen in Figure 1. Every node has a baseband and a radio unit to communicate with a gateway.
Standard Radio Architecture Model
Figure 1: Standard Radio Architecture Model [Hu14]

In Figure 2 there are proposed ideas to separate the baseband unit from the radio unit. There can be pools of multiple baseband units that share a radio which increases the utilization rate of the radio. This also has a side benefit of decreasing the interference since the number of radios is fewer. While this technology is not widely adopted it has a lot of promise by breaking up the network into smaller cells. As the number of cells increase there would be greater route diversity and improve the quality of the wireless network.

Cooperative Topology
Figure 2: Cooperative Topology [Hu14]

3. Small Cell Networks

A hot topic of research is to move away from gateway nodes and change to a densely populated network of multi-tiered devices. The current trend is to have a mostly homogenous network of a single access point that runs all nodes on the network communicate through it. The access point is usually connected to a distributed system that connects to the rest of the internet. As wireless devices become more widely used this network these access points are always one and using energy constantly regardless of the level of traffic. As the rate of wireless adoption rises new network topologies can yield better energy efficient strategies. Current cellular networks are flat with many overlapping coverage areas and all capable of servicing the same amount of traffic regardless of the demands. A network can be built with different sized cells and multiple radio technologies that satisfy the needs of those cells.

3.1 Heterogeneous Networks

Current wireless networks have few paths for a node to send and transmit information. The lack of route diversity can cause signal degradation multipath loss. Increasing the number of nodes on a network allows the reduction of path loss and by having the transmitters and receivers closer together. Small scale networks have also been called Heterogeneous Networks (HetNet).

A heterogeneous network is one that has many types of nodes with different power and data capabilities. The nodes that are deployed will depend on the traffic demands on the area. Currently nodes are constantly powered or operating in low traffic areas and wasting a lot of energy while there is nothing to transmit. Nodes that are responsible for back bone communications or relaying data lots of data require more energy but will able to have a larger radius of communication. In lower traffic areas there will be lower power nodes that will be sufficient for traffic demands.

HetNets are divided into different cell types being capable of serving different ranges of traffic in a multi-tiered architecture. Figure 3 shows the different cells and the levels of where they reside in the architecture. Larger cells are capable of handling larger areas and require more power. The lower nodes have lower power constraints but have a denser network than the higher nodes.

Level and Types of HetNet Cells
Figure 3: Level and Types of HetNet Cells [Navaratnarajah13]

Cells at the lower levels have a smaller range and require less power. As data propagates on the network then it will eventually get passed to the next level on a larger cell until it reaches its destination. It has been shown that these levels with different power requirements yield better energy efficiency and shows some promise to replace current networks. Replacing the current nodes with smaller cells like femto cells, which are small low power devices design to work in the home, would reduce the power consumed as well as the operational costs for a network.

3.2 Interference

There is a drawback with heterogeneous networks related to the density of the nodes. At the inter cell level there can be interference with neighboring nodes due to all the cells using the same spectrum. Frequency division strategies have been developed to either divide groups of cells into using parts of the spectrum allowing to share the entire spectrum. There have also been time division strategies to allow coordination of transmission schedules between cells. Figure 4 shows a typical HetNet topology showing smaller nodes connected to an access point which can communicate to other access points. There is some overlap between the access point nodes which shows the potential interferences between nodes. The denser the network the greater the possibility of having issues with data transmission and receiving due to interference.
HetNet Topology
Figure 4: HetNet Topology [He14]

While this is a promising solution, there is a lot of work to be done with deploying a usable HetNet. There also needs to be improvement at the highest levels of the wireless communication infrastructure to deal with the increased traffic needs. The current number of nodes at the macro and micro level has not yet achieved the density required [Ericsson14]. Also the existing macro cells need to accommodate more antenna diversity and allow for more of the spectrum to be used.

4. Massive MIMO

Though a less common solution, another idea for reducing energy is Massive MIMO antenna deployment through large antenna arrays. Massive MIMO takes advantages of many antennas for a much smaller base of users. In contrast to small cell networks which has high base station density, the massive MIMO solution has high antenna density to serve users.

4.1 Issues

While this approach allows for higher bandwidth and spatial diversity, it does not necessarily reduce the operating cost of the radios. Studies have been done to see if Massive MIMO is a suitable replacement for wired broadband data. It has been suggested that certain MIMO technologies can be adaptive as well and switch to single input multiple output (SIMO) during off peak hours to conserve energy. Algorithms have been developed to detect low traffic and it has been studied that limiting the number of active radios improves energy efficiency [Xu12].

4.2 Comparison to HetNet

Research has shown while both small cell and massive MIMO improve energy efficiency there is a threshold for both [Liu13]. By default there is larger spectral efficiency with massive MIMO. As the number of cells in a small cell network grows the spectral efficiency decreases since the spectrum of the nodes overlaps. To prevent these kinds of issues, the nodes can be put to sleep which improves the energy efficiency There is a point where too many cells in a sleep state can be detrimental to the reliability of the network. If too many nodes are sleeping then that reduces the route diversity and can cause the network to fail. This is could be harmful for time critical functions that need a reliable network to send commands or status to other points in the network. If too many nodes are awake during a low traffic period then energy is being wasted. Current trends show that between HetNets and massive MIMO, HetNets are gaining more traction for being the infrastructure needed to handle the growth of wireless traffic [Liu13].

5. Adaptive Control

Adaptive methodologies can be implemented to improve the energy efficiency of wireless nodes and optimally utilize resources. It is possible that these new technologies can satisfy the growth of wireless data and optimally distribute network resources like frequency spectrum and reduce energy consumption. The most prevalent types of adaptive control can been seen with cognitive radios adapting the radio frequency for transmission and receiving as well as controlling the sleep time of devices.

5.1 Cognitive Radios

Allowing all nodes to have the same resources is not optimal. It wastes energy at the off peak times and depending on how many resources are available may not be sufficient for on peak times. Since the license exempt spectrum is very crowded with wireless devices there may be a lot of interference to wireless communication in the household.

The area of cognitive radios has been an area of study for improving the spectral and energy efficiency of wireless networks. Cognitive radios are ones that can sense their environment and adapt itself with resource constraints. If it detects that a channel is currently occupied it might be able to hop to a new frequency for data transmission. It has been shown that adaptive resource allocation can produce optimal and suboptimal solutions and allow for better utilization of resources [Wang13]. Combined with a small cell densely populated network the adaption of radio spectrum is crucial and has been shown that there are many benefits to implement cognitive radios at the femto cell level [Xie12]. At this level cells are battery operated and increased energy efficiency can extend the life of these devices.

5.2 Sleep Modes

As previously mentioned another common approach to achieving energy efficiency is to put inactive nodes to sleep. It has been shown that 25-30% energy savings can be achieved during low traffic times [Feng13]. Adding QoS data during low traffic periods can add a significant overhead. As the density of nodes in a network increases, this energy savings can be significant. Deeper sleep modes can save even more energy but reduce the functionality of the nodes as well as require a longer wakeup time to become active. There is a tradeoff about how long a node should be asleep to maximize its energy efficiency and not degrade the functionality of the network [Budzisz14].

There are many strategies for controlling the sleep cycle of nodes. The typical static sleep schedule involves a fixed timer and has a node wake up after an interval. This is only useful for set known intervals. A current area of interest is dynamic sleeping modes which are difficult to predict an optimal sleep cycle because it is based on the traffic and the density of the network to service requests.

The current areas of interest for dynamic strategies are controlled from an access point, user, or the network. In node controlled the access point does pilot sensing to detect if nodes are attached and wakes them up. In user controlled, the device sends a wakeup message to the access point. The last option is to have a network controlled sleep mode where a wakeup message is sent to an access point over the backbone of the network through an algorithm to decide when devices should be woken up.

5.3 Delay Tolerant Systems

Considering the type of data is important for determining how to send it. Data collection of nodes on a network may not be a time critical to send out. Collecting and transmitting data at off peak hours has certain energy efficient benefits as well. Furthermore, node aggregation can improve the utilization at low peak times while during heavy demand times resources can be assigned in an adaptive manner [Erol-Kantarci13]. If wireless devices can adaptively determine the current state of the traffic is high it could possibly prevent a retransmission. Huge energy savings can be realized for devices that have delay tolerance.

6 Smart Grids

It has been shown that the carbon footprint increases with wireless use [Lambert12]. To manage the cost and effects of climate change there is a need to reduce overall consumption but be able to control the energy efficiency of wireless devices. Current legacy grids can do this to an extent but are insufficient for dealing with the demands of the future. Some of the best ideas related to energy efficiency can be implemented with smart grids.

6.1 Overview

Smart grids technologies can help reduce operational costs but require a significant change to data collection which can be facilitated by a better wireless network infrastructure. By implementing a HetNet smart grids are a useful way to help reduce energy consumption. They can determine energy needs for the house from smart meters, neighborhood, or city level all managed and getting input from the wireless network. The needs of the network are hard to quantify and can shift throughout the day and can be a very hard problem to adapt to the dynamic nature of energy consumption. A smart grid can allocate resources as necessary to the needs of the network. Improvements to the wireless communication system that transmit this data and improving the reliability of the network are important issues to facilitate the adoption of smarter energy consumption.

6.2 Management

To achieve better smart grid management the time of use is recorded through a network of smart meters. With this information real time demand can be tracked and even be used to predict energy needs in the future. Smart Grids are divided into the residential, neighborhood, and city level. The residential communication involves technologies like Zigbee over 802.15.4 and Wireless with 802.11 to get data from smart meters to improve energy efficiency. At the neighborhood level cellular or 802.11 technologies are used to consolidate data about energy distributed to multiple houses by one transformer [Erol-Kantarci14]. In regards to hierarchical wireless topologies the 802.11s standard for mesh networking is a consideration to allow for nodes to sleep during inactive periods. At the city level the cellular networks would be used to connect neighborhoods to utility facilities. With a better infrastructure we can adapt to the growth of wireless adoption and the energy demand in the world while reducing our global carbon footprint.

7. Summary

In this paper we have shown different approaches to promote higher energy efficiency in wireless communications. As wireless data grows new strategies will have to be developed to service the world's needs. There are many challenges ahead to reorganize the wireless communication landscape from the residential level, cities, and expansion into more rural areas.

Currently there is a trend to make things devices smarter with changes home automation through adaptive smart meters. The biggest hurdle seems to be creating an infrastructure past the home and into the neighborhoods and cities. Increasing energy efficiency is a big challenge to reduce the world's carbon footprint. While there appears to be a lot of promising solutions there is still a lot of work to be done.

What is clear is that adaptive control for dynamic resource allocation with wireless networks is necessary to achieve high energy efficiency. It has shown up in most topics discussed with Massive MIMO, cognitive radios, and small cell networks. Considering the numerous entities that can exist on a wireless network at any one time it is a tough problem to solve. There are active and sleeping devices that constantly change the available resources and it is hard to pinpoint at any given time the state of the network. Trying to maximize the spectral and route diversity as well as optimize the energy consumption of all nodes in a wireless network is key to reduce the global carbon footprint.

8. Reference

[Ericssson15] "Ericsson mobility report,"

[Cisco16] "Cisco Visual Networking Index"

[Hu14] Hu, R.Q., Qian, Yi, "An Energy Efficient and Spectrum Efficient Wireless Heterogeneous Network Framework for 5G Systems" in IEEE Communications Magazine, Vol. 52, No. 5, May 2014, Pages 94 - 101,

[Navaratnarajah13] Navaratnarajah, S., Saeed, A., Dianati, M., Imran, M.A. "Energy efficiency in heterogeneous wireless access networks" in IEEE Wireless Communications, Vol. 20, No. 5, October 2013, Pages 37 - 43,

[He14] He, Shiwen, et al "Leakage-Aware Energy-Efficient Beamforming for Heterogeneous Multicell Multiuser Systems" in IEEE Journal on Selected Areas in Communications, Vol. 32, No. 6, June 2014, Pages 1268 - 1281,

[Ericsson14] "Ericsson HetNet Infographic",

[Xu12] Xu, Z. et al "Energy-Efficient Configuration of Spatial and Frequency Resources in MIMO-OFDMA Systems" in IEEE Transactions on Communications, Vol. 61, No. 2, October 2012, Pages 564 - 575,

[Liu13] Liu W., Han S., Yang C., Sun C. "Massive MIMO or small cell network: Who is more energy efficient?" in IEEE Wireless Communications and Networking Conference Workshops (WCNCW), 2013, 7-10 April 2013, Pages 24 - 29,

[Wang13] Wang, S., Ge, M., Zhao, W., "Energy-efficient resource allocation for OFDM-based cognitive radio networks" in IEEE Transactions on Communications, Vol. 61, No. 8, June 2013, Pages 3181 - 3191,

[Xie12] R. Xie, et al "Energy-Efficient Resource Allocation for Heterogeneous Cognitive Radio Networks with Femtocells" in IEEE Transactions on Wireless Communications, Vol. 11, No. 11, October 2012, Pages 3910 - 3920,

[Feng13] Feng D., et al. "A survey of energy-efficient wireless communications" in IEEE Communications Surveys & Tutorials, Vol. 15, No. 1, February 2013, Pages 167 - 178,

[Budzisz14] Budzisz, L. et al "Dynamic Resource Provisioning for Energy Efficiency in Wireless Access Networks: A Survey and an Outlook" in IEEE Communications Surveys & Tutorials, Vol. 16, No. 4, June 2014, Pages 2259 - 2285,

[Erol-Kantarci14] Erol-Kantarci, M., Mouftah, H.T. "Energy-Efficient Information and Communication Infrastructures in the Smart Grid: A Survey on Interactions and Open Issues" in IEEE Communications Surveys & Tutorials, Vol. 17, No. 1, July 2014, Pages 179 - 197,

[Lambert12] S. Lambert et al., "Worldwide electricity consumption of communication networks," Opt. Exp., vol. 20, no. 26, pp. B513-B524, Dec. 2012.,

8. Acronyms

Last Modified: April 17, 2016
This and other papers on current issues in Wireless and Mobile Networking are available online at
Back to Raj Jain's Home Page