Interference Analysis and Performance Evaluation for LTE TDD System

Pei Chang, Yongyu Chang, Yunan Han, Chi Zhang, Dacheng Yang
Wireless Theories and Technologies Lab
Beijing University of Posts and Telecommunications
Beijing, China
[email protected]

Abstract—LTE of the UMTS Terrestrial Radio Access and frequency division multiple access (SC-FDMA). DFT-spread
Radio Access Network is aimed at supporting high speed OFDM has a small peak-to-average power ratio than
package data service. LTE constitutes a major step toward 4G conventional OFDM, thus enabling power-efficient terminals.
network and supports both FDD mode and TDD mode. In this LTE supports operations in Frequency Division Duplex
paper, we study the interference cases of LTE TDD system, (FDD) and Time Division Duplex (TDD) modes to provide
especially the crossed timeslot interference (CTI) is discussed operators to allocate spectrum flexibly. FDD and TDD are
and evaluated in detail. According to the simulation operated in paired and unpaired spectrum respectively, and
assumption, the system performance of LTE TDD using three they are only different in physical layer. In this paper, our
classic scheduling schemes with or without consideration of analysis focuses on the TDD mode. In order to suit for
CTI is evaluated by system level simulation. We have asymmetry uplink-downlink service, TDD provides different
developed the simulation platform for both uplink and
traffic ratio by defining seven different uplink-downlink slot
downlink directions. Simulation results show that the
throughput degrades significantly due to the CTI.
configurations. According to the requirement of different
uplink and downlink service, operators can allocate radio
Keywords- LTE TD; inter-cell interference; crossed timeslot resource in a flexible manner.
interference; OFDM; A key objective with respect to deployment of LTE is to
utilize a frequency reuse one or as close to reuse one as in
practical network [4]. This requirement increases the
I. INTRODUCTION
interference, so the LTE system will not achieve the full
To suit the requirement of high speed package data capacity as the standard statement. Inter-cell interference will
service, the 3rd Generation partnership projection (3GPP) occur when the same frequency is used among the adjacent
proposed Long Term Evolution (LTE) specifications. LTE of cells. Especially in the cell edge location, the user
the UMTS Terrestrial Radio Access and Radio Access performance degrades significantly. In this paper, we will
Network is aimed at high peak data rates, low latency, give an analysis of the interference in LTE TDD system, and
improved system capacity and coverage, reduced operating system level evaluation with different scheduling schemes
costs, multi-antenna support, efficient support for packet data will be studied. The system performance with different
transmission, and flexible bandwidth operations [1], [2]. The interference situations are given by simulation results.
first release of LTE provides peak rates of 300 Mbps, and The organization of this paper is as follows. Section 
one-way delay is set to less than 5ms between base station analyzes the interference cases in LTE TDD system. Section
and terminal. Furthermore, LTE aims for a seamless  introduces three generally resource scheduling schemes.
integration with existing 2G and 3G system Finally, LTE The system level simulation models of LTE TDD system are
also constitutes a major step toward international mobile
given in section  . Section  presents the results of
telephony (IMT) –advanced [3].
simulation. Finally, our conclusions are presented in section
Orthogonal frequency division multiplexing (OFDM),
which can transmit data on a large number of parallel, .
narrow band subcarriers, is the key technique of LTE II. INTERFERENCE IN LTE TDD SYSTEM
downlink radio transmission. The design complexity of
receiver is reduced, for OFDM transmits data by adapting the Interference situation in LTE TDD system is more
number of relatively narrow band subcarriers in combination complicated than that in LTE FDD system. As will be shown
with a cyclic prefix (CP). Furthermore, OFDM supports below, we subdivide the interference into four categories.
multi-user access by allocating the subcarriers to different A. Intra-cell Interference
users in one scheduling interval.
LTE uplink employs a discrete Fourier transform (DFT)- Intra-cell interference scenarios include user equipments
spread OFDM, sometimes also referred to as single carrier (UEs) purposely scheduled on the same RBs within the cell
(MU-MIMO), or having UEs within a cell allocated with the
same RB during a network load situation [4]. In most study
cases of LTE system, that is, at least in the ideal case, the
978-1-4244-5848-6/10/$26.00 ©2010 IEEE
intra-cell interference was not considered because the

410
subcarriers were orthogonality with OFDM, and the In 3GPP specification [9], 3GPP group defined 7
scheduler did not allocate same RB to different UEs in one different DL/UL configuration schemes in TDD frame, as
cell at the same time. In the realistic system, UE is rejected illustrated in TABLE  . When two neighbor cells use
to access the system when the RB resource is exhausted. different DL/UL configuration schemes, crossed timeslot
interference may occur.
B. Self-noise
In a realistic OFDM system, imperfect carrier TABLE I. UPLINK-D OWNLINK CONFIGURATION
synchronization and channel estimation may result in ‘self-
UL/DL Period Subframe number
noise’ [5], [6]. Self-noise will degrade the orthogonality of configuration (ms) 0 1 2 3 4 5 6 7 8 9
the subcarriers, so the received samples will contain the 0 5 D S U U U D S U U U
interference signal from adjacent subcarriers. At low SINR 1 5 D S U U D D S U U D
the noise increases the loop jitter resulting in an increase to 2 5 D S U D D D S U D D
the self-noise [7]. A number of approaches to combat self- 3 10 D S U U U D D D D D
noise were proposed, and the performance of these 4 10 D S U U D D D D D D
approaches was compared in [8]. 5 10 D S U D D D D D D D
6 5 D S U U U D S U U D
C. Inter-cell Interference
In LTE system, OFDM supports intra-cell orthogonality, As shown in Fig.2, cell A uses NO.3 configuration
so the most important interference which effects system scheme of TABLE , and cell B uses NO.5 configuration
performance is inter-cell interference. Inter-cell interference scheme. Red timeslots are the crossed timeslots. In the
scenarios typically involve UEs in neighboring cells being crossed timeslots, cell A transmits DL service, and cell B
scheduled on the same RB [4]. The transmission rate of UE transmits UL service using the same frequency as the cell A
which the location is at the cell edge will be degraded mostly used. UE A and UE B are both on the edge of cell A and cell
due to inter-cell interference. As shown in Fig.1, there are B respectively, and they are very close. In this case, the co-
four types of inter-cell interference in the LTE TDD system. channel interference from UE B to UE A is very high, and
the interference is called crossed timeslot interference.
Crossed timeslot interference reduces the system
performance seriously. Some analysis of crossed timeslot
interference was given in [10].

8($ &(/ /%

Type1 Type 2
' 6 8 ' ' ' ' ' ' ' &(// $

8(% ' 6 8 8 8 ' ' ' ' ' &(// %

&( //$

Transm ission Tw
odifferent schem
es
I nterference usedby cell Aandcell B

Type3 Type4
Figure 2. Crossed Timeslot Interference
Transm ission
I nterference
III. RESOURCE ALLOCATION SCHEMES
Figure 1. Four Types of Inter-cell Interference
In LTE TDD system, OFDM subcarriers are grouped into
resource blocks (RBs) [9], [11]. RB is the minimal unit to be
Type 1: downlink UE receives interference signal coming
allocated in LTE system. A RB is defined as several
from neighbor base station.
consecutive OFDM symbols in the time domain and several
Type 2: base station receives interference signal coming
consecutive subcarriers in the frequency domain. Resource
from neighbor cell UE.
grid is composed of RBs, and the resource grid with normal
Type 3: downlink UE receives interference signal coming
CP is shown in Fig.3. Each cell has one resource grid, and
from neighbor cell UE.
the scheduler allocates RBs to different users with different
Type 4: base station receives interference signal coming
allocation scheme. Three typically allocation schemes will be
from neighbor base station.
introduced in this paper.
D. Crossed Timeslot Interference
A. Round Robin Scheme
Crossed timeslot interference is a special inter-cell
In round robin (RR) scheme, each UE has a same prior,
interference in LTE TDD system. As the type 3 in Fig.1
and the RBs are allocated to UEs in a cyclical way. In other
shown, the downlink (DL) UE in crossed timeslot receives
words, the RBs are equally allocated to the UEs, irrespective
strong interference when the neighbor cell uplink (UL) UE
uses the same timeslot. of the channel condition of UE. Let Ri = {1, 2, … , M}
denotes the number of RBs which were already allocated to

411
UE i . Let j denotes the number of UE. A new RB is
allocated to UE k is determined by this formula:
 
k = arg min {Ri } (1)   
( j =1,2,..., N )
   

ency
    

q u
RB       

Fre
      
      
      
rriers

      
a

    
b
12Suc

   

  

 
7O
FDMSym
bo
les

Tim
eSlo
t
Figure 4. Wrap-around model
Figure 3. Resource Grid with Normal CP

B. Propagation Model
B. Max C/I Scheme The propagation model is composed of path loss model,
Max C/I scheme optimizes the total system performance, shadowing model and fast fading model. Path loss is adopted
and not considers any fairness. RB is allocated to the UE COST-231 Hata model, and shadowing is a lognormal
which has best channel quality, and the UE which has a poor distribution with standard deviation of 8.9dB. Fast fading
channel condition may not be served at all. Assume that model uses the WINNER model [13]. A formula of the
Qi = {1, 2,..., N } denotes the channel quality of UE i , and relationship is follow:
RB will be allocated to the UE k is determined by this Propagation = pathloss + shadowing + fastfading
formula:
k = arg max {Qi } (2) C. Uplink Model
( i =1,2,..., N )
LTE uplink employs SC-FDMA. To model the SC-
C. Proportional Fair Scheme FDMA mechanism of uplink in LTE TDD system, N u
Proportional fair (PF) scheme considers both the total continuous RBs are allocated to each up transmission link.
system performance and the fairness of UE, thus a tradeoff Perfect knowledge of the channel condition is assumed, and
between performance and fairness is made by this scheme. the scheduler determines modulation and code scheme (MCS)
Let Qi = {1, 2,..., N } denotes the instantaneous channel by the SINR.
quality of UE i , and Ai = {1, 2,..., N } denotes the average D. Downlink Model
channel quality of UE i . The allocation priority of UE i is N d RBs are allocated to each downlink to model OFDM
Q
calculated by Pi = i . RB will be allocated to the UE k is mechanism of downlink. Unlike the uplink model, the N d
Ai downlink RBs are continuous or non-continuous, which is
determined by this formula: determined by link conditions. We assume perfect
k = arg max {Pi } (3) knowledge of the channel condition, and assume equal
( i =1,2,..., N ) power allocation similar as [14].
IV. SIMULATION MODEL E. AMC Model
The system level simulation method is used, and several According to the channel condition, MAC layer chooses
key sub-models are introduced below. the corresponding MCS level. We Use the method in [15] to
model the adaptive modulation and coding (AMC)
A. Area Model mechanism. The MCS levels which we used are shown in
As shown in Fig.4, the 7 cells wrap-around model [12] is TABLE 
used, and each cell is divided into two parts: cell central part
and cell edge part. At the beginning of simulation, UEs are
uniformly distribution among the 7 cells.

412
 TABLE II. MCS TYPE considering the CTI with same scheduling scheme is about
MCS Mod Code rate TB SINR 9,309 kbps. The throughput degrades approximately 8
1 QPSK 1/5 30 -1.27dB percent by the CTI. In despite of scheduling scheme, the
2 QPSK 1/3 50 0.00dB average throughput is increase with the UE number
3 QPSK 1/2 76 2.60dB increasing. This is owing to the multi-user diversity.
4 QPSK 2/3 101 4.67dB Fig.6 shows the average throughput per UE. When the
5 QPSK 3/4 114 5.83dB UE number is increasing, the average throughput per UE is
6 QAM16 1/2 152 7.60dB
7 QAM16 2/3 202 10.33dB
decreasing. This is because the interference and the UE
8 QAM16 3/4 228 11.80 dB number are increased together. The Max C/I scheme has an
9 QAM16 4/5 243 12.67 dB outstanding performance, and the RR scheme has the best
10 QAM64 3/5 273 14.00 dB fairness. The CTI degrades the average throughput per UE
11 QAM64 2/3 304 15.33 dB significantly. In the 150 UEs case, the average throughput
12 QAM64 4/5 364 18.27 dB per UE with Max C/I scheme in no consideration of CTI is
about 516 kbps, and the value considering the CTI with same
scheduling scheme is about 466 kbps. The throughput
F. HARQ Model degrades approximately 10 percent by the CTI.
According to [16], we use the N-channel asynchronous
transmission HARQ model, and use the chase-combining TABLE III. SIMULATION PARAMETERS
scheme by accumulating the received signal energy of frame
Parameters Assumption
retransmissions. The accumulated energy across k Carrier frequency 2GHz
transmissions is calculated by the formula: Bandwidth 10MHz
k Total Node B power 46dBm
EnergyHARQ (k ) = ∑ E (i) (4) Cell topology Wrap-around (see Fig.4)
i =1 Cell radius 500m
RB number 50
where E ( k ) is the received energy of the kth frame. CP type Short
Subcarrier bandwidth 15kHz
V. SIMULATION RESULTS RB bandwidth 180kHz
We consider the uplink transmission and the downlink HARQ scheme Chase combining (8-channel)
SINR model EESM
transmission of the LTE TDD system simultaneously. The
Traffic type FTP
parameters used in simulation are summarized in TABLE . Thermal noise density -174dBm
Based on the simulation parameter settings, we consider MCS levels 12 (see TABLE )
the system performance with CTI and without CTI Path loss COST-231 Hata model
respectively. In the scenario with CTI, frame structures NO.3, Standard deviation of shadow 8.9dB
NO.4, and NO.5, which are shown in TABLE , are used in fading
Fast fading WINNER model [13]
our simulation. In the scenario without CTI, frame structure
Scheduler RR, Max C/I, and PF
NO.3 is used. The throughputs with CTI and without CTI are
both obtained. Three scheduling schemes (RR, Max C/I, and VI. CONCLUSION
PF) are used in the simulation. The throughputs with
different scheduling schemes in different network load will In this paper, the interference cases of LTE TDD system
be presented. are analyzed firstly. Especially the CTI case is discussed in
In Fig.5, the average throughput per cell with RR, Max detail. CTI is a special inter-cell interference in LTE TDD
C/I, and PF scheduling schemes are presented, and the CTI system. The downlink UE in crossed timeslot receives strong
conditions are also considered. From the outcome, we can interference when the neighbor cell uplink UE uses the same
see that the Max C/I scheme can improve the throughput, but timeslot simultaneously.
the Max C/I scheme does not consider the user fairness. The According to the simulation model assumption, the
scheduler allocated RBs to UEs with good channel condition, system performance of LTE TDD using three classic
and the UE with bad channel condition may be rejected to scheduling schemes with consideration of CTI and with no
access the system. RR scheme scheduler guarantees fairness consideration of CTI are evaluated. Simulation results show
to every UE by allocating RBs to UEs cycling with no that the throughput degrades significantly due to the CTI,
consideration of channel condition, so the average and the degradation is different slightly with the different
throughput is small. PF scheme makes a tradeoff between network load and the different scheduling schemes. So, we
throughput and fairness. The throughput of PF is larger than must pay more attention to CTI when design a LTE TDD
RR and smaller than Max C/I. system.
From Fig.5, we can also see that the system performance The performance degradation due to CTI may be reduced
degrades significantly by the CTI. UE number (horizontal by advanced scheduling scheme. Future study should
axis) is the total number in the coverage area. In the 200 UEs consider new scheduling scheme with avoiding CTI.
case, the average throughput per cell with PF scheme in no
consideration of CTI is about 10,139 kbps, and the value

413
 x 10
4 [7] A. Al-Dweik, R. Hamila, Z. Asghar, “Dynamic feedback carrier
2 frequency offset estimator for OFDM system,” in Proc. IEEE ICSPC
RR - without CTI 2007, pp. 1043-1046, Nov. 2007.
1.8 RR - with CTI [8] J. Lim, D. Hong, “Inter-Carrier Interference Estimation in OFDM
Max C/I - without CTI Systems with Unknown Noise Distributions,” IEEE Signal Process
Average Throughput per Cell (kbps)

1.6 Max C/I - with CTI Letters, vol. 16, pp.493-496, Jun. 2009.
PF - without CTI
PF - with CTI
[9] 3rd Generation Partnership Project, Technical Specification Group
1.4 Radio Access Network; Evolved Universal Terrestrial Radio Access
(E-UTRA); Physical channels and modulation, 3GPP Std. TS 36.211
1.2 v. 8.4.0, 2008.
[10] W. Wang, K. Zhou, Y. Zhang, S. Shen, “Effects of synchronization
1 errors on the performance of TDD-OFDMA crossed timeslot,” in
proc. ChinaCom 2008, pp. 870-875, Aug.2008.
0.8
[11] E. Dahlman, S. Parkvall, J. Sköld, and P. Beming, 3G HSPA and LTE
for Mobile Broadband. New York: Academic, 2007.
0.6
[12] 3rd Generation Partnership Project, Technical Specification Group
Radio Access Network; Analysis of higher chip rates for UTRA TDD
0.4
50 100 150 200 250 300 350 evolution, 3GPP Std. TR 25.895 v. 6.0.0, 2004.
UE Number [13] IST-2003-507581 WINNER D5.4 Final report on Link Level and
System Level Channel Models v1.4, https://www.ist-winner.org.
Figure 5. Average Throughput per Cell
[14] 3rd Generation Partnership Project, Technical Specification Group
Radio Access Network; Physical layer aspects for evolved Universal
Terrestrial Radio Access (UTRA), 3GPP Std. TR 25.814 v. 7.0.0,
800
2006.
RR - without CTI [15] 3rd Generation Partnership Project, Technical Specification Group
700 RR - with CTI
Radio Access Network; Feasibility study for Orthogonal Frequency
Max C/I - without CTI
Division Multiplexing (OFDM) for UTRAN enhancement (Release 6),
Average Throughput per UE (kbps)

Max C/I - with CTI

600 3GPP Std. TR 25.892 v. 6.0.0, 2004.
PF - without CTI
PF - with CTI [16] 3rd Generation Partnership Project, Technical Specification Group
500
Radio Access Network; Physical Layer Aspects of UTRA High Speed
Downlink Packet Access, 3GPP Std. TR 25.848 v. 4.0.0, 2001.

400

300

200

100
50 100 150 200 250 300 350
UE Number

Figure 6. Average Throughput per UE

REFERENCES
[1] R. Ratasuk, A. Ghosh, W. Xiao, R. Love, R. Nory, B. Classon, “TDD
design for UMTS long term evolution,” in proc. IEEE PIMRC 2008,
PP. 1-5, Sept. 2008
[2] 3rd Generation Partnership Project, Technical Specification Group
Radio Access Network; Evolved Universal Terrestrial Radio Access
(E-UTRA) and Evolved Universal Terrestrial Radio Access Network
(E-UTRAN); Overall description; Stage 2, 3GPP Std. TS 36.300 v.
8.6.0, 2008.
[3] D. Astely, E. Dahlman, A. Furuskar, Y. Jading, M. Lindstrom, S.
Parkvall, “LTE: The Evolution of Mobile Broadband,” IEEE
Communications Magazine, pp. 44-51, April 2009
[4] G. Boudreau, J. Panicker, N. Guo, R. Chang, N. Wang, S. Vrzic,
“Interference coordination and cancellation for 4G networks,” IEEE
Communications Magazine, April 2009.
[5] J. Huang, V. Subramanian, R. Agrawal, R. Berry, “Downlink
scheduling and resource allocation for OFDM system,” IEEE Trans.
Commun., vol. 8, no. 1,pp. 288–296, Jan. 2009.
[6] H. Jin, R. Laroia, and T. Richardson, “Superposition by position,” in
Proc. IEEE ITW 2006, Mar. 2006.

414