Proactive Algorithm to Reduce Power Consumption

and Handovers in Heterogeneous Mobile Networks

Michel Nahas, Mohammad Sabati Rana Kanso, Mohammad Azizi

Electrical Engineering Department School of Engineering
Australian College of Kuwait Lebanese International University
P.O. Box 1411, Safat 13015, Kuwait P.O. Box: 146404 Mazraa, Beirut, Lebanon
[email protected]

Energy consumption of communication and cellular networks

To save the energy of the whole network, the phenomenon of
has recently received increased attention in the research traffic load fluctuation implies that some BSs can be switched
community. Moreover, a large number of base stations (BSs) off when the traffic load is light.
have been deployed in order to meet these demands on cellular Nowadays, green communication is one of the main design
traffic. This leads to considerable increase in the operational cost goals of future mobile networks and current research aims to
and contributes as a major portion of the energy consumption of enable sustainable growth of broadband wireless infrastructure,
cellular networks. Therefore, network operators are considering different solutions have been proposed so far for improving the
how to reduce the energy consumption and design “Green energy efficiency of wireless networks while maintaining a
Communication”. The concept of Green Communication is to given Quality-of-Service (QoS). Today, the energy demand for
manage energy consumption and maintain reasonable costs while
keeping high quality of service and benefits. In this paper, we
communication is being increased by larger and larger number
present some techniques for reducing power consumption by of communication devices and infrastructures. In addition, due
developing solutions in terms of the network architecture and the to the rapidly growing of the mobile network, the footprint of
base stations components. We use some of these solutions to CO emission from massive energy production is being a
develop and implement a novel algorithm that combines Cell critical issue that comes up with threating the global warming.
Zooming and Heterogeneous Networks with many types of base Research shows that 3% of world’s electrical energy is taken
station while taking into consideration user’s mobility. The by Information and Communication Technology (ICT) sector
efficiency and the power management of our algorithm, and it is responsible for 2% of world’s CO emission [2].
especially in high traffic networks, are achieved and approved by
This problem trends the network operators to focus on
simulations.
energy consumption, design, develop and to investigate various
Index Terms— Cellular networks, Green communications, models and associated science needed for combined energy
Energy saving, Sleep mode, Cell zooming, Heterogeneous efficiency and network optimization.
networks. In this paper, we present a brief survey of some recent
methods suggested to improve the power efficiency of cellular
networks, and discuss various remarkable techniques toward
I. INTRODUCTION green mobile networks. We focus on existing green
communication tools based on the idea of changing the
In recent years, traffic demand in cellular networks has coverage area of cells in function of MSs distribution (cell
significantly grown and throughout the expansion of mobile zooming) or switching BSs off (sleep mode), and the ability of
networks, the energy efficiency has always been a major having several connection options with the presence of smaller
concern. However, there is an increasing worldwide concern size cells (micro, pico and femto) inside a macro cell.
about potential damages to the environment, such as carbon Additionally, we present our optimization algorithm called
dioxide (CO ) emissions that leads to severe threats for global ProGreen which focuses on user equipment position and
warming. This concern compelled many researchers to study mobility and adjusts base stations power level based on these
different ways on how to minimize energy consumption of factors. ProGreen will reduce the power consumed in a
cellular systems and network operators to continuously explore heterogeneous network while decreasing the number of
future technologies in order to bring improvements in the entire handovers (HOs) of users between the different BSs of the
network infrastructure. The large number of base stations (BSs) network.
contributes a major portion of the energy consumption of The rest of this paper is organized as follow, in section II
cellular networks. When a BS is in its working mode, the we present the background of our research and a review of the
energy consumption of processing circuits and air conditioner state of art techniques. While in section III, our proposed
takes up about 60% of the total consumption [1]. However, system model is introduced and then in section IV, the phases
most of the efforts for energy saving in cellular networks still of the proposed ProGreen algorithm are explained. In section
focus on reducing the transmit power of BSs and mobile V, we present the simulations results and finally we conclude
stations (MSs). in section VI.
 II. BACKGROUND AND PREVIOUS WORK B. Previous Works
In this section, we review concepts such as cell zooming Recently, many research papers have investigated network
and sleep mode techniques which are applied on the network architecture based solutions to reduce power consumption in
architecture level. The cell sizes are adaptively adjusted cellular networks. Intelligent network deployment strategies
according to traffic load, user requirements and channel where using small, low power base station are believed to
conditions by implementing some changes to existing network decrease the power consumption compared to high power
hierarchy. These are often believed as powerful strategies to macro base stations. In [4], different heterogeneous networks
reduce energy consumption of cellular radio networks. are investigated in the context of energy efficiency for both
uniform and non-uniform scenarios, taking into account the
A. Network Architecture Based Solutions effect of cell radius, various BSs types and number of
When a BS is in its working mode, the energy consumption additional BSs on network power consumption.
takes up about 60% of the total consumption [3]. Therefore, to In [5], a semi-dynamic green resource management
save the energy of the whole network, the phenomenon of mechanism was proposed to minimize BSs energy
traffic load fluctuation implies that some BSs can be switched consumption and also to satisfy rate requirement of each MS
off when the traffic load is light. Therefore, many switching for the downlink of a cloud-assisted HetNet. The authors made
on/off schemes have been proposed such as sleep mode, cell use of the scalability of system configurations such that green
zooming and virtual coverage. resource management can be employed by flexibly switching
off some unnecessary hardware components, especially for off-
1) Sleep Mode peak traffic scenarios. A green communication in a multi-
Sleep mode is a mechanism applied to allow the hardware operator heterogeneous wireless medium was investigated in
components in the BS to be switched off in idle conditions, [6]. The proposed decentralized downlink radio resource
such that the energy consumption is modulated over the allocation mechanism guaranteed mutual benefits in power
variations in traffic load. This is done in a way that takes into saving among different service providers. Cooperation among
consideration the same quality of service level to mobile different networks is favored only if it results in a win-win
stations covered in the base station area. situation among the service providers.
The authors in [7] aimed to optimize energy efficiency of
2) Cell Zooming HetNets by introducing several levels of sleep depths.
Cell zooming is a technique in which base stations Coverage probability and average achievable rate in
dynamically adjust their coverage radius and hence their heterogeneous wireless networks are derived with different
transmit powers based on user locations. The transmit power is sleep modes for small cells using stochastic geometry-based.
set to the minimum required level depending on user locations In [8], cluster-based approach for maximizing the energy
and signal to interference and noise ratio (SINR) required. efficiency of wireless small cell networks was proposed. A
Implementing cell zooming in cellular networks needs to dynamic mechanism is proposed to group locally coupled
introduce some new components and corresponding small cell base stations into clusters based on location and
functionalities to current network architecture. The Control traffic load. Within each formed cluster, BSs coordinate their
Server (CS), controls the procedure of cell zooming. The CS is transmission parameters to minimize a cost function which
a virtual entity in the network, which can be either captures the tradeoffs between energy efficiency and flow level
implemented in the gateway or distributed in the BSs or performance, while satisfying their users’ quality-of-service
between several BSs. requirements. Clusters also will compete with one another in
Energy consumption reduction in cellular networks can be order to improve the overall network’s energy efficiency.
achieved by zooming in or zooming out the cells based on the
current traffic distribution. By adjusting the transmit power of III. PROPOSED SYSTEM MODEL
the base station, power can be saved and can be varied Let us consider a cellular network, where cells are
dynamically. deployed in a circular form, with a central macro cell
consisting of micro cells and pico cells. Base stations are
3) Heterogeneous Networks
located at the center of the cells and have omni-directional
Heterogeneous networks are intelligent networks where
antennas. At the normal case, all micro BSs are using the same
macro and micro BSs operate and cooperate in order to achieve
transmit power and the same radius of coverage, same for pico
an optimal power control and QoS. The idea being that a BS
BSs. Micro BSs are uniformly distributed in the cells, but pico
closer to mobile stations lowers the required transmit power
BSs are distributed randomly in a specified region.
due to advantageous path loss conditions. The main concept of
heterogeneous network (HetNet) is the operators’ deployment A. Power Consumption Models
of the cooperation between macro layer and “intelligent pico For a better understanding on how the power will be
cells” or femto cells to function together with the Macro layer. optimized, and since the power computations for large
In order to provide the best capacity and coverage possible, networks is expensive or unfeasible, we present models of
a combination of large and small cells (macro/ micro/ femto/ power consumption for HetNets base stations (macro, micro
pico) is required. and pico BSs).
 The parameter αB accounts for power consumption that
1) Macro Base Station Modeling scales with the radiated power due to cooling, power supply
Macro base stations provide a large area of coverage within and signal processing as well as amplifier and feeder loss, this
a mobile network. Therefore the macro base station power parameter is similar to PPEQ in macro BSs. The parameter bB
consumption mainly depends on the size of the area covered models the power which is consumed independently of the
as well as the degree of coverage required. They usually cover transmit power due to power supply, cooling and signal
areas with radius from about 1000m to 5000m with more than processing. This parameter is equivalent to PSTQUV in macro
90% coverage [9]. The relation between power consumption base station. is a variable parameter that varies with the
and the radiated power and the number of users or the cell radius of the cell. Table III shows the relation between the
load using the linear model of the form: radiated power, cell zooming and the power consumed:

_ = + ; (1) Table III. Different Zoom Level, Radiated Power and Coverage for the Micro
Base Station.
= + . ; (2)
Coverage 50m 100m 200m 400m 500m
! "# $%& ' (
= )."
; (3)
PTX 5mw 210mw 4.1w 8.4w 10.25 W
Ptotsect ! 1+ αmacro ( Zooming
_ = Pfixed + + αmacro . Ptax . (4) 25% 50% 100% 200% 250%
0.12 level

Where denotes the fixed power consumed that scales

3) Pico Base Station Modeling
with the cell load and it equals to 395.56 W and
Besides macro and micro BSs, pico cells are designed to
= 354.44 W. _ denotes the total power consumed by serve very small area such as part of a building, a street
the base station which is equal to the sum of and corner, malls and railway stations. These BSs require much
that denotes the power consumption that scales with the cell smaller power since they normally cover area with a cell
load. denotes the power amplifier that has a significant radius of 100m. Power consumption model of pico BSs is
portion of the energy consumed, since sufficient power is represented in Eq. 5 where a pico BS with a cell radius of 100
needed to reach distant terminals. ? ? denotes the m, we consider aB = 5.5 and bB = 38 W:
average power radiated to cover a specific area. This
parameter varies proportionally with the radius of the covered _ = !5.5 × ?Z + 38(, (6)
area.
where ?Z is the radiated power of pico BS.
Table I shows the power consumption percentage of each
element in the BS.
IV. PROGREEN ALGORITHM
Table I. Power Consumption Percentage of Macro Base Station Components.
Our ProGreen algorithm takes into consideration three
Air Signal Power Power major functions that help in the simulation which are the
Component
Conditioning Processing Amplifier Supply
Power
speed, the direction and the time remaining in a cell.
10%-25% 5%-10% 50%-80% 5%-10%
consumption
A. Speed of the Mobile Station
The speed of a mobile station is a very important scalar in
Table II shows all the values needed for calculating the ProGreen algorithm; it defines the rate of change of the
P@ABC_DE@FA in the macro base station [10]: user’s position with respect to time and it is used with a
certain threshold, 5 m/s. The speed function takes many
Table II. Different Zoom Level, Radiated Power and Coverage for the Macro parameters, the old and new locations in Cartesian system (x,
Base Station. y) of the MS, the time difference between these positions and
Coverage 400 m 600m 800m 1km the previous speed of the MS if it exists.
_! ` ab e e
GHIHJKLM cd( # !f` a bf cd (
17w 25w 33w 40w \] = ^. 5 + 0.5\ g . (7)
Zooming level 40 % 60% 80% 100%
The equation above calculates the average cumulative
speed of the user where \] is the velocity of the user and since
2) Micro Base Station Modeling
it takes the previous speed which is also calculated by the
Micro BSs usually cover smaller area with a cell radius
same equation. In the first speed calculation for the new user
between 100m and 250m [11], therefore these micro BSs
\ g is treated as zero.
require much smaller power consumption than macro cells.
The power consumption model of micro base stations is given B. Direction of the Mobile Station
by Same as for the speed, the direction of a MS plays an
= !N . + O (. (5) important role in connecting the user to the most suitable BS.
_
The algorithm uses Eq. 8 to determine the direction of the user D. Proactive Algorithm: ProGreen
which is an angle!h( between 0 and 360 degrees. ProGreen algorithm focuses on all the users’ locations and
h = 0.5 ∗ h) + 0.5 ∗ jNkb" !l −l ,o −o (. (8) movements inside a network in order to serve them in a best
m g m g
way with lowest possible power consumption and is based on
The direction function takes the old and the new locations estimating the MS velocity from the received signal in cellular
of the user to determine which direction the MS is going to. It networks. Such interest is due to the fact that significant
also takes the previous angle h) of the user if it exists and improvements in the performance of wireless communication
calculates the average between them. systems are possible if the MS speed is known. Therefore,
when a new user enters the network; this algorithm will be
C. Time Remaining of the Mobile Station in the Cell triggered and will start its first phase, which is calculating the
The connection time in the algorithm is a very important speed of the MS. Then the second phase where a comparison
value besides the amount of the consumed power. Velocity of will take place so the MS can be connected to the best BS.
the user and the direction of the MS are used to compute the
time remaining of any user in the current cell. Once the time 1) First Phase
remaining in the cell is known, it is compared to a certain In the first phase, a new method is proposed for estimating
threshold and the algorithm decides whether to connect the the velocity of mobile stations, the basic idea is to avoid
MS to the current BS or to any other BS. several types of handoff calls that unnecessarily occupy extra
The function takes into consideration the cell radius, the power consumption by considering the relative mobility of the
location of the MS towards the base station and user’s user.
velocity. It first computes the angle between the MS and the The MS speed is determined using the well-known
base station according to their locations in the cell and the relationship between the distances taking into consideration
distance in Eq. 9 is calculated using this angle. Then, the time the direction and the differences in time. However, an
remaining in the cell will be the distance divided by the user important feature within a cellular network is that the location
speed. Fig. 1 shows the movement of the user in the cell. of a user can be determined.
As long as the cell-phone is powered on, the location of
the MS can always be traced to at least the cell from which it
is receiving signal from the cellular network. Such network-
based methods of estimating the location of a user is useful to
calculate the exact distances that the user is moving. This
paper investigates the accuracy of using cellular network BS
information for estimating the velocity of user, through
quantities analysis of BS data, where we can get the location
of each user at a specific time. Thus, depending on the data
that we get from the base station and using Eq. 7 we can get
an approximate value of velocity for a certain user.
According to the user’s speed, we classify the users into
two categories. First category is below a velocity threshold
where there are three optional BSs to connect to it
(macro/micro/pico). For instance, we assume the velocity
threshold is 5m/s for urban areas. Second category is above
Fig. 1. Example of Movement of Mobile Station in the Cell.
the velocity threshold where there are only two optional BSs
(macro/micro) because pico cells coverage is very small
In the figure, p is the location of the BS, the MS is located comparing to high speeds (i.e., v>5m/s). The basic idea behind
at C, and its location is considered to be known. q1 is the cell this discrimination is to reduce the possibility of handoff calls.
radius. The line CF represents the present trajectory of MS. Thus, calculating the average velocity of the MS as
Since we know distances d1, R1 and angle θ, the equation primary criteria will help in reducing the number of available
below can be used to obtain CF which is the remaining (optional) BSs, and this will lead to power consumption
distance in the cell [12]: reduction through minimizing the handoff calls. Fig. 2 below
illustrates steps of this phase in a flow chart.
rs = t1 u^v θ ± _!q1 − t1 vyk o(. (9)
2) Second Phase
After calculation of the user speed and classifying him into
Finally, the time remaining in the cell will be as follows: one of the two categories explained above, the MS will
z{ proceed to the second phase where it will be connected to a
j= . (10) BS with the lowest power consumption with respect to the
remaining time of the MS in the cell. In this phase of the
algorithm all BSs will be sorted in an ascending order based
on the extra power consumed per user for each available BSs,
where the energy consumption calculation will take place • Step 6: Else, go to Step 2 and choose the next base station
using Eq. 5. where the current base station is eliminated from the list
of sorted base stations.
Here, we will use BS information for estimating the
direction and speed of MS. According to the direction and
speed of MS we can obtain the time remaining of a MS in a
cell. Once the time remaining in the cell is known, it is
compared to a certain threshold and the algorithm decides
whether to connect the MS to the current base station or to
zoom out the selected cell if possible.

Fig. 2. The First Phase of the Proactive Algorithm.

A comparison between connecting the MS to the

macro/micro/pico will be performed in order to take the
optimum decision for the whole network; the BS with the
minimum value will be selected. A BS will enter the sorting
procedure if and only if there is a possibility to cover the
location of the MS at its maximum zooming level, and the BS
capacity status is able to accept any new user. Based on this
decision, the BS will remain in the available BSs list or will be
eliminated and the algorithm will move to its next state.
One of the major points to note in this phase is the
computation of the distance between the BS and the MS
Fig. 3. The Second Phase of the Proactive Algorithm.
before sorting the BSs whether or not that new MS is able to
connect to it in an optimized energy consumption connection.
Since choosing the distance between the MS and the BS as a If it is possible for the selected cell to zoom out, the
primary criterion will help in reducing the number of BSs in algorithm will start from the beginning (second phase only),
the list which in turn leads to additional power consumption and BSs will be sorted again in an ascending order based on
reduction. the extra power consumed per user for each available BSs
Defining this point as primary criteria for a BS will taking into consideration the new coverage of the selected cell
optimize the performance of the proposed algorithm after zooming out. This new zooming level of the selected cell
regardless of the BS cell load since it is shown in the previous may lead to power consumption for the whole network higher
section that the micro BS power consumption will increase than the power consumption caused by connecting the MS to
more when it zooms out than when its cell load increases. adjacent BS. Therefore comparing extra power consumed per
Thus, the distance measured between the MS and the BS at its user after zooming out the current BS is mandatory to have an
current zooming level has significant rule to obtain an optimal decision.
optimized energy consumption connection. In case there is no possibility for the selected cell to cover
Fig. 3 shows the steps of this phase: the location of the MS and it reaches its maximum zooming
• Step 1: Compute the extra power consumed per MS for level, the selected BS will be eliminated from the list of
all the available cells. available BSs and the algorithm will move on the next
• Step 2: Choose the BS with the minimum power minimum extra power consumed per user.
consumption.
• Step 3: Calculate the time remaining of the user in the V. SIMULATION AND RESULTS
selected cell. We consider a macro BS with radius q = 1 km in its
• Step 4: Compare the time remaining with a threshold Tth normal state (without zooming) and a maximum number of
(i.e., 5 seconds). If the remaining time is above the users | =500. The micro BSs each of radius q = 200 m
threshold, connect the MS to this cell. in the normal state and a maximum number of users } =50
• Step 5: Else, find out whether the selected cell is able to each, and the pico BSs each of radius q = 100 m in the
zoom out. If yes, go to Step 1. normal state and a maximum number of users ƞ =20 each.
Furthermore, we assume that users are randomly distributed consumption for different number of pico BSs is plotted in
inside the network. Two different scenarios will be discussed Fig. 5.
hereafter to show how “ProGreen” can reduce the total energy
consumption of the different BSs in the network. It can be seen that our “ProGreen” algorithm consumes
Using Matlab (R2014a), we will compare our new less power than the “Nearest” algorithm, because the
algorithm “ProGreen” which is based on choosing the lowest “ProGreen” algorithm depends on connecting the MS to the
power consumption of the available BSs in the network taking base station with the minimum power. In the second
into consideration the velocity, direction and time remaining simulation, it is clear that when increasing the number of pico
of the MS in a cell, with another the “Nearest algorithm”, cells in the network less power is achieved, that is because
which connects each MS to the nearest BS (macro, micro or pico BS consumes less power than micro and macro BSs.
pico) regardless from the MS’s speed or direction.

A. Scenario 1: Simulations without Mobility

We simulate “Nearest” and “ProGreen” algorithms and
compare the power consumed by each algorithm for different
number of k users up to k = 200. In this scenario, there is no
need to test the velocity and remaining time of the MSs
because they are not moving. Simulation has been performed
in order to ensure the consistency and the sustainability of
“ProGreen” algorithm over the “Nearest” algorithm. The
configuration parameters of this simulation were adjusted to
repeat the algorithm for 200 users where for each user the total
power consumption of the system is recorded, and the average
of these total power consumptions for the two algorithms is
plotted in Fig. 4.

Fig. 5. Power Consumption of ProGreen in Terms of the Number of Pico BSs

B. Scenario 2: Simulations with Mobility

We simulate for a number of k=200 users, by changing the
positions of the 200 MSs and comparing the number of
handovers for the “Nearest” and “ProGreen” algorithms for
different average MSs velocities ranging from for • = 0m/s to
20m/s. The simulation starts by connecting the 200 MSs to the
BS similarly to Scenario 1 for each algorithm then we begin to
move the same MSs and determine whether a handover to
another BS for each MS is needed or not.
In the case of the “Nearest” algorithm, we take the average
velocity of MSs into consideration because this algorithm
connects the MSs to the nearest BS according to their new
locations. Therefore, there is no need to calculate the
remaining time for the MS in the connected cell. The zooming
level of micro and macro cells is also important for this
Fig. 4. Power Consumption for ProGreen and "Nearest" algorithm with 50 algorithm.
Pico Cells For the “ProGreen” algorithm, we take into account the
speed, the direction and the remaining time of the MS in the
As shown in Fig. 4, the power consumption at 100 users is cell. These parameters track and estimate the future movement
925W for the “Nearest” algorithm, and for “ProGreen” of the MS. After knowing the new position, the speed, the
algorithm the power is 660W. Therefore, “ProGreen” direction of the MS and the remaining time can be calculated.
algorithm reduced the power consumption with savings of If the remaining time of the MS in the current cell is less than
265W (decrease of 28% approximately). One additional five seconds (threshold remaining time), a handover will
simulation has been performed in this case in order to ensure occur. In this case, “ProGreen” algorithm also considers the
the reduction of total power consumption of the cellular zooming level for micro and macro cells same as for the
network using the “ProGreen” algorithm by increasing the “Nearest” algorithm. Fig. 6 and Fig. 7 show the average
number of pico BSs, and the average of these total power number of handovers for both algorithms when increasing the
velocity from 0m/s to 20m/s for 100 MSs and 200 MSs mobility in the network and benefits by previously developed
respectively. When simulating over 100 MSs, the average techniques such as cell zooming and cells sleep mode. Our
number of handovers for the “Nearest” algorithm is 18 HOs algorithm achieved good reduction in the power consumed by
when v = 20m/s, while the average number of handovers for the different BSs and also a decrease in the number of needed
the “ProGreen” algorithm is 15 HOs for the same velocity. handovers in a network with micro and pico cells installed
For 200 MSs, Fig. 7 shows that the average number of within the macro base station.
handovers for the “Nearest” algorithm is 33 HOs when v = 20
m/s, while for “ProGreen” algorithm the average number of VII. REFERENCES
handovers is only 25 HOs for the same velocity. [1] K. Al Haj Ismaiil, B. Assaf, M. Ghantous and M. Nahas, "Reducing
Power Consumption of Cellular Networks by Using Various Cell Types
and Cell Zooming," in Third International Conference on e-
Technologies and Networks for Development (ICeND), Beirut, Lebanon,
April 2014.
[2] S. Boiardi, A. Capone and B. Sansó, "Radio Planning of Energy-Aware
Cellular Networks," Computer Networks: The International Journal of
Computer and Telecommunications Networking, vol. 57, no. 13, pp.
2564-2577, September 2013.
[3] J. T. Louhi, "Energy efficiency of modern cellular base stations," in
IEEE 29th International Telecommunications Energy Conference, Rome,
Italy, October 2007.
[4] A. J. Fehske, F. Richter and G. P. Fettweis, "Energy Efficiency
Improvements through Micro Sites in Cellular Radio Mobile Network,"
in IEEE GLOBECOM Workshops, Honolulu, USA, December 2009.
[5] P. Cao, W. Liu, J. S. Thompson, C. Yang and E. A. Jorswieck,
"Semidynamic Green Resource Management in Downlink
Heterogeneous Networks by Group Sparse Power Control," IEEE
Journal on Selected Areas in Communications, vol. 34, no. 5, pp. 1250 -
1266, May 2016.
Fig. 6. Average Number of Handovers in Terms of Velocity for 100 Mobiles [6] M. Ismail, K. Qaraqe and E. Serpedin, "Cooperation Incentives and
Downlink Radio Resource Allocation for Green Communications in a
Heterogeneous Wireless Environment," IEEE Transactions on Vehicular
Technology, vol. 65, no. 3, pp. 1627 - 1638, March 2016.
[7] C. Liu, B. Natarajan and H. Xia, "Small Cell Base Station Sleep
Strategies for Energy Efficiency," IEEE Transactions on Vehicular
Technology, vol. 65, no. 3, pp. 1652 - 1661, March 2016.
[8] S. Samarakoon, M. Bennis, W. Saad and M. Latva-aho, "Dynamic
Clustering and on/off Strategies for Wireless Small Cell Networks,"
IEEE Transactions on Wireless Communications, vol. 15, no. 3, pp. 2164
- 2178, March 2016.
[9] H. Claussen, B. L. S. Alcatel-Lucent, L. T. W. Ho and F. Pivit, "Effects
of joint macrocell and residential picocell deployment on the network
energy efficiency," in IEEE 19th Personal, Indoor and Mobile Radio
Conference (PIMRC), Cannes, France, September 2008.
[10] A. Chatzipapas, S. Alouf and V. Mancuso, "On the Minimization of
Power Consumption in Base Stations Using On/Off Power Amplifiers,"
in IEEE Online Conference on Green Communications (GreenCom),
September 2011.
[11] M. Nahas, M. Ghantous, K. Al Haj Ismaiil and B. Assaf, "For Better
Energy Consumption and Management in Future Cellular Networks," in
International Conference on Renewable Energies for Developing
countries (REDEC), Beirut, Lebanon, July 2014.
Fig. 7. Average Number of Handovers in Terms of Velocity for 200 Mobiles
[12] F. Sabeen Shaikh, "Proactive Policy Management for Heterogeneous
Networks," in 3rd International Conference on Information and
VI. CONCLUSIONS Communication Technologies: From Theory to Applications, Damascus,
Syria, April 2008.
Currently, the information and communication technology
ICT sector is facing a serious challenge related to the
increasing power consumption due to the continuous
deployment of new base stations in order to respond to the
traffic demand around the world. In this paper, we have
reviewed the concept of heterogeneous networks and some
techniques that treat these challenges such as sleep modes and
cell zooming. Then, we have proposed our optimization
algorithm “ProGreen” that takes into consideration the user