You are on page 1of 6

Fractional Frequency Reuse in LTE Networks

Husam Eldin Elmutasim Osman Mohamed Elfadil, Mohammed Adil Ibrahim Ali, Mohammed Abas
Department of Electrical and Electronic Engineering
University of Khartoum
Khartoum, Sudan

Abstract— 3GPP Long Term Evolution (LTE) is a standard available to reuse these frequencies is lower than the
for wireless communications to achieve high spectral efficiency, equivalent gain achieved with this SINR improvement [2].
high peak data rates, as well as flexibility in frequency and
bandwidth. By using frequency reuse of one in LTE networks, Hence, conventional cellular systems suffer from poor
high spectral efficiency is achieved. However, the Inter-Cell spectral efficiency through high reuse factors. With a
Interference (ICI) resulted from the frequency reuse of one is a frequency reuse factor of one, as in LTE-Release 8, higher
main limitation in these networks. Fractional Frequency Reuse spectral efficiency is achieved since all the bandwidth is
(FFR) has been proposed as an Inter-Cell Interference allocated in each cell. However, there will be greater
Coordination (ICIC) technique in Orthogonal Frequency interference between neighboring cells, especially at the cell-
Division Multiple Access (OFDMA) based LTE networks. FFR edge, which reduces the SINR for cell-edge users [3].
divides the cell into an inner and an outer region, and then
different frequency reuse factors are applied in each region in Inter-Cell Interference Coordination (ICIC) is a strategy to
such a way the interference is reduced. This paper focuses on improve the performance of the network by having each cell
evaluating two FFR deployment schemes, namely, Strict FFR and allocate its resources such that interference experienced in the
Soft Frequency Reuse (SFR) and compares them with the network is minimized, while maximizing spatial reuse[3].
conventional frequency reuse scheme. A broad comparison
Fractional frequency reuse (FFR) has been proposed as an
among all these frequency reuse schemes is performed by using a
proposed mechanism which depends on Monte Carlo simulations
ICIC technique in OFDMA based wireless networks [4]. The
considering performance metrics such as Signal to Interference basic idea of FFR is to partition the cell’s bandwidth so that (i)
plus Noise Ratio (SINR), capacity and throughput. Simulation cell-edge users of adjacent cells do not interfere with each
results show that, the higher SINR achieved by Strict FFR is not other and (ii) interference received by (and created by) cell-
reflected in terms of throughput due to the small bandwidth interior users is reduced, while (iii) using more total spectrum
utilization, and Soft Frequency Reuse (SFR) can provide higher than conventional frequency reuse. In FFR the cell space is
cell-edge throughput by increasing the power control factor but divided into two regions: inner, which is close to the Base
this comes at the cost of decreased inner region throughput. Station (BS) and outer, which is situated to the borders of the
cell. The whole frequency band is divided into several sub-
Keywords—fractional frequency reuse (FFR); inter-cell bands, and each sub-band is assigned either to the inner or the
interference coordination (ICIC); orthogonal frequency-division outer region of the cell [4].
multiple access (OFDMA); long term evolutin (LTE)
There are two common FFR deployment modes: Strict FFR
I. INTRODUCTION and Soft Frequency Reuse (SFR). While FFR can be considered
Long Term Evolution (LTE) is the next major step in in the uplink or downlink, this paper focuses on the downlink
mobile radio communications, and was introduced in 3GPP since it typically supports links with greater rate requirements
Release 8. LTE uses Orthogonal Frequency Division Multiple with a low margin for interference and additionally we can,
Access (OFDMA) as its radio access technology, together unlike the uplink, neglect power control by assuming equal
with advanced antenna technologies [1]. power downlinks [4].

The scarcity of bandwidth, spectrum and power in wireless 1) Strict FFR: Strict FFR is a modification of the traditional
systems has driven the need for spectrally efficient frequency reuse used extensively in multi-cell networks [6],
communication systems. In a cellular communication system [7]. Fig. 1 illustrates Strict FFR for a hexagonal grid modeled
such as Long Term Evolution (LTE), the Inter-Cell deployment with a cell-edge reuse factor of RF=3. Users in
Interference (ICI) is one of the main factors that influences the each cell-interior are allocated a common sub-band of
data rates of the users at the cell-edge and affects the average frequencies while cell-edge users’ bandwidth is partitioned
spectral efficiency of the cell. across cells based on a reuse factor of RF. In total, Strict FFR
thus requires a total of RF + 1 sub-band. Interior users do not
When the frequencies of downlink transmission between share any spectrum with exterior users, which reduce
neighboring cells are different, like in the conventional interference for both interior users and cell-edge users [4].
cellular systems, then the inter-cell interference will be a
minor problem. This type of frequency reuse is common in
older generation of wireless networks as it improves the Signal
to Interference plus Noise Ratio (SINR). But the bandwidth
In (1), refers to the distance between user x and base
station b, is the path loss exponent, is the transmit power
of the base station b on subcarrier n, is the small-scale
Rayleigh Fading which is i.i.d. complex Gaussian value
according to ( ), and is the noise power of the
Additive White Gaussian Noise (AWGN) channel. Symbols k
and j refer to the set of all the interfering BSs (i.e. BSs that are
using the same sub-band as user x). In detail, j is the cell index
and k is the number of co-channel cells.
The interference that occurs comes from disjoint sets of
downlinks in the inner and outer region. A transmission in an
Fig. 1. Strict FFR. inner region that is assigned specific frequency band causes
interference only to inner users of other cells that are assigned
2) Soft Frequency Reuse (SFR): Fig. 2 illustrates a SFR the same band. Furthermore, it is necessary to distinguish two
deployment with a reuse factor of RF = 3 on the cell-edge. SFR categories of BSs. The first consists of all interfering BSs
employs the same cell-edge bandwidth partitioning strategy as transmitting to inner region users on the same sub-band as user
Strict FFR, but the interior users are allowed to share sub-bands x and the second consists of all interfering BSs transmitting to
with edge users in other cells. Because cell-interior users share cell-edge users on the same sub-band as user x [5].
the bandwidth with neighboring cells, they typically transmit at
In our analysis, in the case of Conventional Frequency
lower power levels than the cell-edge users [8], [9]. While SFR
Reuse and Strict FFR, we assume that equal transmit power is
is more bandwidth efficient than Strict FFR, it results in more
interference to both cell-interior and edge users [10], [4]. applied, for all BSs.
In the case of SFR inter-cell interference no longer comes
from disjoint sets of interior and exterior downlinks, but can
come from either set, and coarse power control is typical. To
accomplish this, a power control factor is introduced to
the transmit power to create two different classes,
, , where is the transmit power
of the base station if user x is an interior user and is the
transmit power of the base station if user x is a cell-edge user.
The interfering base stations are also separated into two
classes: , which consists of all interfering base stations
transmitting to cell interior users on the same sub-band as user
Fig. 2. Soft Frequency Reuse (SFR). x (at power ) and , which consists of all interfering
base stations transmitting to cell-edge users on the same sub-
II. SYSTEM MODEL band as user y (at power ). For an inner user, the
In this section we describe the theoretical approach to resulting SINR expression is given as [13]
calculate the Signal to Interference plus Noise Ratio (SINR),
capacity and throughput. We assume that the overall network is
composed of M adjacent cells. Each cell contains K users √ ( )
seeking to share N subcarriers. Users are distributed randomly (2)
inside the cell. We distinguish the case where a user is found in
the inner or in the outer region of the cell. In a typical OFDMA
cellular network, for a user x who is served by a base station b
on subcarrier n, the related SINR is given by the following Where ∑ √ ( )
equation [11]:
∑ √ ( )
√ ( )
(1) And for a cell-edge user
∑ √ ( )
√ ( ) (3)
The set represents all the interfering base stations, i.e.
base stations that are using the same sub-band as user . The
shadow fading is a random variable described by a log-
normal distribution, ( ) The path loss follows After the SINR estimation, we proceed with the throughput
the 3GPP Long Term Evolution (LTE) model, ( ) calculation. The spectral efficiency in bps/Hz of user x on
( ), where is the distance [12]. subcarrier n, can be calculated by the Shannon's formula
[14]:
( ) (4) Inner region radius is chosen to be equal to two-thirds of
Moreover, the throughput in bps of the user x can be the radius ( ( ) ) where is the total cell radius.
expressed as follows:
For each user the mechanism calculates user's SINR
according to each frequency reuse scheme, and this calculation
∑ (5) is done for 300 Monte Carlo realizations in order to average the
fading and shadowing values.
Where is the available bandwidth for each subcarrier,
and represents the subcarrier assigned to user x. When Each user's SINR is then mapped to its corresponding CQI
entity [16], in order to determine the modulation scheme which
the subcarrier n is assigned to user x. Otherwise,
tells us the number of bits per symbol used by this user. The
. spectral efficiency is then calculated from Shannon's formula.
One of the most important FFR system design parameters is Subcarrier allocation is the main part of an OFDMA RRM
the inner region radius , which determines the size of the algorithm. Here, the algorithm decides first the number of
frequency partitions. Additionally, since the cell partitions are subcarriers that must be allocated to a user and then the specific
based on the geometry of the network, knowledge of user channels from the set of available subcarriers. Subcarrier
locations is important. One practical method to determine user allocation is executed exclusively for a new user asking
classifications is for each cell to use the average received admission. Re-allocation of all OFDM subcarriers within a cell,
(SINR) of its users, which is usually a good indicator of the although it is able to exploit multi-user diversity as shown at
distance of the user from its base station. The base station then [17],[18], increases algorithmic complexity and time for
classifies users with average SINR less than a pre-determined decision making.
threshold as edge users, while users with average SINR greater
than the threshold are classified as interior users. Subcarrier allocation is divided into two steps. First, the
algorithm determines the number of subcarriers per user. The
For both FFR systems, the resource allocation between subcarrier is loaded with b bits/symbol according to the
interior and cell-edge users is proportional to the square of the modulation level. Assuming that a user asks for a bit rate
ratio of the interior radius and the cell radius R. This is optimal bps, when the subcarrier spacing is Hz the number of
when user locations are assumed to be uniformly distributed subcarriers that should be allocated to this user is,
[15]. For Strict FFR, with total subcarriers available to
the cell, the allocation of subcarriers for interior users
(10)
and exterior users is given as [13]
Then, the subcarrier allocation algorithm decides which
[ ( ) ] (6)
specific subcarriers to assign to the new user. We use random
subcarrier allocation algorithm which is a well-known
( )⁄ (7) subcarrier allocation technique, widely used in 2G wireless
For SFR the allocation of sub-bands is given as networks in order to average co-channel interference [18]. The
following shows that the algorithm randomly selects
[ ( ) ] (8) subcarriers from the set of available subcarriers and allocates
them to the user .

([ ] ) (9) 1: ← ( )

2: Allocate subcarriers to user


III. MECHANISM OVERVIEW
3: ←
In this section, we propose a mechanism for evaluating the
three types of frequency reuse schemes under consideration, where ( ) randomly selects elements from the
namely, conventional frequency resue, Strict FFR and SFR. set .
The goal of the mechanism is to produce average per-user
Since we are working under the assumption that subcarriers
SINR, average per-user spectral efficiency and average per-
are identical, it does not matter which subcarriers are allocated
user throughput as functions of distance from Base Station for
to the user. Therefore, we are only concerned about the number
each frequency reuse scheme, as well as the CDFs of the
of subcarriers allocated to each user.
mentioned parameters.
Each user randomly requests one of the service classes (64,
The proposed mechanism assumes a topology that consists
128, 256, 512 or 1024 kbps), then the corresponding number of
of a grid of M cells and K users that are randomly distributed in
requested subcarriers is calculated. Moreover, we calculate the
each cell. In case of conventional frequency reuse the whole
number of subcarriers that corresponds to the minimum service
bandwidth is used in the cell with frequency reuse 1.
class for that user.
Otherwise, the mechanism divides each cell in an inner region
and an outer region. For Strict FFR and SFR, bandwidth In case of remaining unassigned subcarriers, the excess
division is done as described in Section II. subcarriers will be assigned to the users in round robin manner.
On the other hand, if the number of the available subcarriers is
less than the number of the total requested subcarriers, the The simulation parameters that are necessary for the
mechanism will subtract subcarriers in round robin manner conduction of the experiment are presented in Table I. The
until the number of the assigned subcarriers is equal to the network is composed of 19 cells, and 24 users randomly
available subcarriers. The situation where the available distributed in each cell. Since we examine an LTE-based
subcarriers are less than the minimum number of subcarriers cellular environment all the performance requirements, link
needed by the user is assumed to not exist. and system simulation parameters (BS transmit power, Power
Noise Density, Cell radius) are in accordance with the 3GPP
Once subcarriers are allocated to each user, the specifications [16]. In detail, we consider a system with
corresponding throughput is determined. We repeat the 20MHz of bandwidth (i.e. LTE) divided into 1200 subcarriers
mechanism for 1000 Monte Carlo simulations. Then the of 15 KHz each [21]. The scenario assumed is urban macro,
average per-user SINR, average per-user spectral and average which exists in dense urban areas, served by macro cells BS of
per-user throughput are plotted as functions of average per-user 2,000 MHz frequency.
distance from Base Station, as well as their CDFs for all
frequency reuse schemes.
Strict FFR, SFR with different power control factor values
( ) and the conventional method are examined.

Initialization of Simulation
Parameters
SINR values calculation and the
averaging og these values

Subcarrier Allocation
Fig. 4. Average per-user SINR as function of average per-user distance from
the BS for different frequency reuse schemes.
Spectral Efficiency and
Throughput calculation Fig. 4 shows average per-user SINR as function of average
Averaging data and plotting the per-user distance from the BS for different frequency reuse
results schemes. The conventional method and Strict FFR have similar
SINR trend in the inner region since each user will be applied
to the same transmit and interference powers, and both achieve
Fig. 3. Flow diagram of the implemented FFR program code best SINR performance in the inner region. However, in the
outer region (cell-edge), where users in Strict FFR suffer from
A general flow diagram of the FFR program code is less interference than the conventional method, the former has
presented as shown in Fig. 3. It describes the main parts of the a better performance than the latter. SFR with along
program code and the steps that the program code performs in with the conventional method have the worst performance in
order to obtain the final results. It is important to notice that the cell-edge. For SFR, as we increasing the power control
this is only a general flow diagram and it does not include all factor, we can clearly notice that SINR performance is
the parts and functions in the actual program code implemented increasing in the cell-edge because of the increased signal
in MATLAB, it only gives a broad idea about the sequence in power. However, as the power control factor is increasing, the
which the program code operates. The written MATLAB file users in the inner are exposed to greater interference and ,
can be found on [20]. eventually, we see a considerable decrease in SINR in the inner
IV. RESULTS region. Since path loss is proportional to the distance from the
BS, the SINR, in general, is declining as we go farther from
TABLE I. SIMULATION PARAMETERS BS.

Parameter Value
Network size 19 cells
Users per cell 24
System bandwidth 20 MHz
Subcarriers 1200
Subcarriers' bandwidth 15 KHz
Cell radius 250 m
Carrier frequency 2,000 MHz
BS transmit power 46 dBm
Noise power spectral density -174 dBm/Hz
Fig. 5. Average per-user Spectral Efficiency as function of average per-user
distance from the BS for different frequency reuse schemes.
Fig. 5 shows the average per-user spectral efficiency as Fig. 7 shows CDF of average per-user SINR values for
function average per-user distance from the BS for different different frequency reuse schemes. The reason behind using
frequency reuse schemes. Since the spectral efficiency is CDFs is to show the overall performance of different frequency
proportional to the SINR (see Shannon's Formula), similar reuse schemes in spite of the region which the user belongs to.
trends to that observed in SINR for different frequency reuse It can be clearly seen that Strict FFR has the best SINR
schemes occur here. performance because of the less amount of interference facing
the cell-edge users. Strict FFR is followed closely by
conventional frequency reuse and SFR with
respectively. SFRs with increased power control factor have
worse SINR performance.

Fig. 6. Average per-user Throughput as function of average per-user distance


from the BS for different frequency reuse schemes.

Fig. 6 shows the average per-user throughput as function


average per-user distance from the BS for different frequency Fig. 8. CDF of average per-user Spectral Efficiency values for different
reuse schemes. It can be clearly seen that conventional frequency reuse schemes.
frequency reuse has the highest throughput in the inner region,
and lowest throughput in the cell-edge. For the outer region, Fig. 8 shows CDF of average per-user spectral efficiency
SFR with β=8 has the highest cell-edge throughput. And for different frequency reuse schemes. Since the spectral
obviously for less values of power control factor, less cell-edge efficiency is proportional to the SINR (see Shannon's
throughput is observed. However, although Strict FFR has the Formula), similar trends to that observed in SINR for different
best SINR performance in the cell-edge, this is not reflected in frequency reuse schemes occur here.
terms of throughput. This is due to the fact that Strict FFR has
less number of subcarriers allocated in each cell especially in
the cell-edge. On the other hand, SFR has much better
bandwidth utilization. Since the throughput is linearly
proportional to the bandwidth, and logarithmically proportional
to the SINR, a decrease in the former will surpass an increase
in the later. Therefore, Strict FFR has less throughput
performance especially in the outer region compared to SFR in
spite of its superiority in terms of SINR. Yet the throughput is
decreasing as we increase the distance from the BS, because of
the increase amount of path loss.
Although both conventional frequency and SFR have large Fig. 9. CDF of average per-user Throughput values for different frequency
bandwidth utilization, the conventional frequency performs the reuse schemes.
better in the inner region while SFR is worse especially for
increased power control factors. In contrast, the increased Fig. 9 shows CDF of average per-user throughput for
power control factors give higher cell-edge throughputs while different frequency reuse schemes. We can say that
the conventional method has worse values. conventional frequency reuse has the best performance (except
for throughput values less than 0.25 Mbps where Strict FFR
performs better) with a considerable difference from the other
methods followed by Strict FFR and SFR. As mentioned
earlier, although Strict FFR has the best SINR performance, it
cannot have the best throughput performance since the amount
of Strict FFR's allocated subcarriers per-cell is less than SFR's
and the conventional method. The logarithmic increase caused
by the increased SINR cannot contain the linear decrease
caused by reduced allocated subcarriers. The conventional
method showed much better throughput performance than the
other schemes. However, as we saw earlier in Fig 4.3 this
better throughput performance is only in the inner region, and
Fig. 7. CDF of average per-user SINR values for different frequency reuse
schemes.
the cell-edge of the conventional method has the worst [5] C. B. Dimitrios Bilios, Vasileios Kokkinos, Andreas Papazois, Georgia
throughput among all investigated schemes Tseliou, "Selecting the Optimal Fractional Frequency Reuse Scheme in
Long Term Evolution Networks," Wireless Pers Commun, 2013.
CONCLUSION [6] F. Baccelli, M. Klein, M. Lebourges, and S. Zuyev, “Stochastic
geometry and architecture of communication networks,” J
The problem of improving cell-edge throughput of LTE .Telecommunication Systems, vol. 7, no. 1, pp. 209–227, 1997.
networks has been an extensive area of research in both [7] T. Brown, “Cellular performance bounds via shotgun cellular systems,”
academia and industry. Fractional Frequency Reuse (FFR) has IEEE Journal on Sel. Areas in Communications, vol. 18, no. 11, pp.
been proposed as ICIC technique to solve the cell-edge 2443–2455, November 2000.
problem, in which each cell is divided into inner and outer [8] M. Haenggi, J. Andrews, F. Baccelli, O. Dousse, and M. Franceschetti,
region with different frequency reuse factor in each region. “Stochastic geometry and random graphs for the analysis and design of
wireless networks,” IEEE Journal on Sel. Areas in Communications, vol.
Strict FFR and SFR have been evaluated as types of FFR, in 27, no. 7, pp. 1029–1046, September 2009.
comparison with conventional frequency reuse. It is found that [9] F. B. Jeffrey G. Andrews, Radha Krishna Ganti, "A Tractable Approach
in spite of the fact that Strict FFR provides the best cell-edge to Coverage and Rate in Cellular Networks," IEEE Transactions on
users' SINR performance, this is not reflected in terms of cell- Communications, 2011.
edge throughput due to the small bandwidth utilization. SFR [10] K. Begain, G. Rozsa, A. Pfening, and M. Telek, “Performance analysis
increases cell-edge SINR and throughput by increasing the of GSM networks with intelligent underlay-overlay,” in Proc. ISCC
power control factor, but this comes at cost of corresponding 2002. Seventh Intl. Symp. on Computers and Communications,
Taormina, Italy, July 2002, pp. 135–141.
decrease of these metrics in the inner region, thus making a
[11] 3GPP TSG-RAN, TR 25.814. (2006–2005). Physical layer aspects for
tradeoff between inner and cell-edge throughput. However, evolved UTRA (Release 7), v1.3.1.
both Strict FFR and SFR provide lower performance than [12] T. S. Carmen Botella, Xiaodong Xu, Hui Zhang, "On the Performance of
conventional method in the inner region, and the improvements Joint Processing Schemes over the Cluster Area," presented at the
occur only in the cell-edge region. Vehicular Technology Conference (VTC 2010-Spring), 2010 IEEE 71st,
Taipei, Taiwan, 2010.
ACKNOWLEDGMENT [13] Jeffrey Anderson, Thomas Novlan, Illsoo Sohn, Radha Krishna Ganti,
The authors would like to thank Jean Mobark, Youmna El "Comparison of Fractional Frequency Reuse Approaches in the OFDMA
Cellular Downlink," presented at the IEEE Globecom 2010.
Bitar and Abeer Haddad from the American University of
[14] H. Lei, Zhang, L., Zhang, X., & Yang, D, "A novel multi-cell OFDMA
Beirut for their help in access to some papers in the system structure using fractional frequency reuse," presented at the IEEE
IEEEXplore library. The authors also like to thank Dr. Ghassan 18th international symposium on personal, indoor and mobile radio
Mohammed Taha Ali and Prof. Sharief Fadul Babikir from communications PIMRC’07, Athens, Greece, 2007.
Department of Electrical and Electronic Engineering, [15] Z. Bharucha and H. Haas, “The distribution of path losses for uniformly
University of Khartoum for their efforts in reviewing this distributed nodes in a circle,” Rec. Lett. Commun., vol. 2008, pp. 1–4,
paper. 2008.
[16] 3GPP TR 36.814-900, “3rd Generation Partnership Project; Technical
REFERENCES Specification Group Radio Access Network; Further Advancements for
EUTRA Physical Layer Aspects (Release 9),” March 2010.
[1] "LTE: an introduction", Ericsson 2011.
[17] D. Zarbouti, G. Tsoulos, and D. Kaklamani, “Performance Evaluation of
[2] S. Parkvall and D. Astely, “The Evolution of LTE towards IMT- OFDMA RRM Algorithms with Spectrum Reuse 1,” in Proc. European
Advanced,” Journal of Communications, vol. 4, no. 3, 2009. [Online]. Conference on Wireless Technologies (ECWT’07), Munich, Germany,
Available: Oct. 2007, pp. 205–208.
http://academypublisher.com/ojs/index.php/jcm/article/view/040314615
[18] S. Pietrzyk and G. Janssen, “Radio resource allocation for cellular
4
networks based on OFDMA with QoS guarantees,” in Proc. IEEE
[3] T. R. LAKSHMANA, "Dynamic Coordinated MultiPoint (CoMP) Global Telecommunications Conference (GLOBECOM’04), Dallas, TX,
Transmission Schemes," Master of Science in Communication Nov.- Dec. 2004, pp. 2694–2699.
Engineering, Department of Signals and Systems, CHALMERS
[19] G. Potter, “System Design Choices in Personal Communications,” IEEE
UNIVERSITY OF TECHNOLOGY, Göteborg, Sweden, 2010.
Personal Commun. Mag., vol. 2, no. 5, pp. 50–67, Oct. 1995.
[4] R. K. G. Thomas David Novlan, Arunabha Ghosh, Jeffrey G. Andrews,
[20] https://drive.google.com/file/d/0B672mIVyuZ6hcHhXMXJGWHp2UXc
"Analytical Evaluation of Fractional Frequency Reuse for OFDMA
/view?usp=sharing
Cellular Networks," IEEE Transactions on Wireless Communications,
2011. [21] 3GPP. (2009). LTE Resource Guide. Anritsu Company.

You might also like