FREQUENCY AND TIME DOMAIN PACKET SCHEDULING BASED ON CHANNEL PREDICTION WITH IMPERFECT CQI IN LTE

International Journal of Wireless & Mobile Networks (IJWMN) Vol. 5, No. 4, August 2013
DOI : 10.5121/ijwmn.2013.5412 157
FREQUENCY AND TIME DOMAIN PACKET
SCHEDULING BASED ON CHANNEL PREDICTION
WITH IMPERFECT CQI IN LTE
Yongxin Wang1
, Kumbesan Sandrasegaran2
, Xinning Zhu3
, Jingjing Fei4
Xiaoying
Kong5
and Cheng-Chung Lin
6
1
FEIT, University of Technology, Sydney, Australia
Yongxin.Wang-1@student.uts.edu
2
Kumbesan.Sandrasegaran@uts.edu.au
3
Beijing University of Post and Telecommunications, Beijing, China
zhuxn@bupt.edu.cn
4
CSE, University of New South Wales, Sydney, Australia
jingjingf@cse.unsw.edu.au
5
Xiaoying.Kong@uts.edu.au
6
Cheng-Chung.Lin@eng.uts.edu.au
ABSTRACT
Channel-dependent scheduling of transmission of data packets in a wireless system is based on
measurement and feedback of the channel quality. To alleviate the performance degradation due to
simultaneous multiple imperfect channel quality information (CQI), a simple and efficient packet
scheduling (PS) algorithm is developed in downlink LTE system for real time traffic. A frequency domain
channel predictor based on Kalman filter is first developed to restore the true CQI from erroneous channel
quality feedback. Then, a time domain grouping technique employing the joint of Proportional Fair (PF)
and Modified Largest Weighted Delay First (M-LWDF) algorithms is used. It was proved this proposal
achieves better performance in terms of system throughput and packet loss ratio by simulation results.
KEYWORDS
LTE, packet scheduling, channel estimation, Kalman filter, imperfect CQI
1. INTRODUCTION
To provide a long-term development of the third generation (3G) services and better quality of
wireless communication in terms of high performance and capacity, simplicity and wide range of
terminals, “Super 3G”[1]. standard that expands upon the 3rd generation partnership project
(3GPP) specification was proposed in 2004, and it was called long-term evolution (LTE) within
3GPP. Orthogonal frequency division multiple access (OFDMA) technology was chosen in the
downlink LTE instead of other access technologies (e.g. Frequency Division Multiple Access -
FDMA, Time Division Multiple Access - TDMA, and Code Division Multiple Access - CDMA)
due to its robustness to inter-symbol interference (ISI) and its immunity to frequency-selective
fading nature of the mobile channels[2].The use of the OFDMA is best suited for simultaneous
support for multiple high data rate real time applications.

158
In downlink LTE, packet scheduling is the term used when a packet scheduler select a user, based
on a priority such as good channel quality and packet delay information, to transmit its packets on
the radio resources. The LTE standard defines a resource allocation structure in both time and
frequency domains. The minimum downlink LTE transmission unit that can be allocated to a user
is called a physical resource block (PRB) which includes two resource blocks (RB). In the
frequency domain (FD), a PRB consists of 12 consecutive sub-carries of 180 kHz total bandwidth
and is of 1 ms duration in time domain (TD). This feature encourages the use of both time and
frequency domain packet scheduling algorithms.
There are many well-known packet scheduling algorithms designed in time domain, such as PF
algorithm[3] and M-LWDF algorithm[4]. The PF scheduler provides a good trade-off between
throughput maximization and fairness guarantee by trying to serve a user whose channel quality
achieves its peak maximum to average ratio in any given scheduling interval. However, because
of a lack of consideration of the buffer information of each user, such as the delay of the head of
line (HoL, the HoL packet of a user is the packet that has resided the longest in its buffer at the
enhanced Node B (eNB)), the PF scheduling algorithm is not expected to support real time
services. The M-LWDF algorithm was developed to support real time traffic in code division
multiple access-high data rate (CDMA-HDR) system. This algorithm is supposed to have a
reasonably overall performance due to the consideration of packet delay information, the
instantaneous data rate and average throughput. The adaptation of these two packet scheduling
algorithms into frequency domain for downlink LTE can be found in[5].
Various frequency domain packet scheduling algorithms that are able to allocate resources to
multiple users in one transmission time interval (TTI) have been proposed. These algorithms
calculate the channel quality of each user on every PRB, and select a user with the highest priority
on that PRB. Recent FDPS study in [6] has shown that FDPS can provide around 40%
improvement in both average system capacity and cell throughput over time domain only
scheduling. In[7-10], algorithms that are based on resource allocation and resource assignment
techniques were developed. The resource allocation technique calculates the number of RBs that
is to be allocated to each user in each TTI. If the PRBs are insufficient to satisfy all the users’
allocation requirement, the priority of each user will be calculated to decide who can be selected
to obtain the competitive resource. Thereafter, the resource assignment technique allocates the
PRBs to the user who has the highest priority for that PRB based on a scheduling algorithm.
These resource allocation-assignment scheduling algorithms are quite complex due to the fact that
it involves many steps on the RB basis, resulting in a large number of iterations. .
To balance the computational complexity and scheduling efficiency, joint time frequency
schedulers are investigated in[11-13]. These schedulers operate in two steps, where the first step
performs a pre-selection based on a TD scheduling algorithm to limit the number of multiplexed
users that go into the frequency scheduler. The second step consists of a resource allocation to
users from the TD scheduler according to a frequency scheduling algorithm. The computational
complexity can be reduced by the employment of the TD scheduler which either does not
consider each user’s channel quality or only take the channel quality across the whole bandwidth
into consideration. Furthermore, the algorithm can benefit from the limitation of the number of
input users to the FD scheduler. In fact, these algorithms are still complex due to the simultaneous
employment of two packet schedulers.
However, all the above mentioned packet scheduling algorithms assume that the measured CQI of
each user are ideally fed back to the eNB without any error, which is impossible in a practical
wireless channel. In the downlink LTE system, feedback and processing delays cause a mismatch
between the current channel state and the CQI received by the scheduler[14], along with
interference, multi-path fading, and shadowing. These might cause severe performance

159
degradation especially for a channel-dependent packet scheduler that highly relies on an accurate
CQI report. In [15], performance of adaptive OFDM systems due to outdated CQI was improved
by channel prediction while [16] studied the effect of multiple estimations to mitigate the
degradation due to CQI delay as well as OFDM channel estimation error. The problem caused by
imperfect CQI in OFDMA downlink system has been studied in[17,18]. In [17], optimal resource
allocation algorithms employing a dual optimization framework were developed, in which the
complexity for discrete rate scenario was shown by simulation results to be much less than
continuous rate scenario. In [18], a delay-sensitive cross-layer design framework with
heterogeneous delay requirements was proposed so as to guarantee a fixed target outage
probability in optimal power and subcarrier allocation under heterogeneous users’ delay
constraints
Generally, imperfect channel conditions can lead to three types of inaccurate CQI: erroneous CQI,
outdated CQI and unavailable CQI when the delay of the outdated CQI is longer than the
scheduling interval. Outdated CQI is the most common situation due to the fact that feedback and
processing delay lead to a mismatch between the current channel state and the received CQI. The
effect of outdated CQI has been analysed in [14, 19, 20]. In [14], several CQI estimators with
CQI delay were studied and numerical results showed that the proposed prediction method based
on Stochastic Approximation had the best performance and a lower computational complexity. In
[19], the Probability Distribution Function (PDF) of the CQI corresponding to an predicted CQI
was proved to improve the Multi carrier Proportional Fair (MPF) algorithm in terms of system
throughput in downlink OFDMA systems. In [20], outdated and unavailable CQI scenarios were
simulated by sporadic traffic. Results showed that the proposed resource allocation scheme based
on an optimal HARQ and adaptive modulation and coding policy improves scheduling efficiency
for LTE-advanced networks with delay constraint. In [3], simulation results presented that the
improvement method can alleviate the detrimental effect caused by CQI delay.
Several multi-user systems predictors were investigated in [21]. Wiener filter for iterative channel
estimation in OFDMA systems, channel estimation performance without memory (Block-Least
Squares Estimation, Block-LSE), and channel estimation with memory employing Kalman filter
are discussed in [22], [23], and [24], respectively.
Among the existing packet scheduling algorithms in LTE downlink system, only limited number
of them have an overall acceptable performance under simultaneous multiple imperfect channel
states. However, due to the large-scale computational problem, the practical implementation of
the algorithms is difficult. The aim of this paper is to propose an efficient and simple packet
scheduling algorithm that provides good system performance when there are simultaneous
multiple CQI problems. This scheduling algorithm involves a FD predictor based on Kalman
filter and a simple TD grouping technique which allows alternate use of two different FD
schedulers. Simulation results show that the packet scheduling is optimized with the combination
techniques, which results in a performance improvement.
The rest of this paper is organized as follows: The downlink LTE system model is provided in
Section 2. The FD channel quality predictor based on Kalman filter is described in Section 3
followed by introduction of the TD grouping technique in Section 4. Section 5 gives an overview
of the proposed algorithm. Simulation results are presented in Section 6. Finally, Section 7
concludes the paper.
2. DOWNLINK LTE SYSTEM MODEL
The LTE deploys a simple architecture that only consists of eNBs which perform packet
scheduling along with other radio resource management (RRM) functions. In this paper, 7

160
hexagonal cells of 5 MHz bandwidth with 25 PRBs and 2 GHz carrier frequency is modelled.
Each PRB is of 1 ms duration and contains 14 OFDMA symbols and a total of 168 resource
elements (REs) with the usage of a normal cyclic prefix (CP). Table 1 gives the main downlink
LTE system parameters.
It is assumed in this paper, each eNB is located at the centre of each cell with 100 meter radius.
Users are uniformly distributed within each cell and inter-cell interference remains constant. Each
user moves within the cell at a constant speed in a constant direction which is initialized as a
random value and is wrapped-around whenever it reaches the cell boundary [25]. Users only have
real time applications.
Users compute their instantaneous channel quality (e.g. signal-to-noise-ratio, SNR) on each RB
which is mapped into a CQI, and report it to the serving eNB at each TTI. The reported CQI value
is then used to make a scheduling decision and determine the data rate of a user for packet
transmission. In this paper, it is assumed that all the CQI reports have 3 ms delay [26] and
become unavailable at regular intervals (i.e. the CQIs are set to 0 every 10 ms). Equation (1) gives
the mathematical expression of the instantaneous SINR experienced by user i on RB j at time t:
, ( ) = , ( ) ∗ (1)
where , ( ) is the efficiency (in bits/RE) of PRB j of user i at time t. is the
total number of Res specified for downlink data transmission.
Table 1. Main downlink LTE system parameters
System Parameters Values
Cellular layout 7 hexagonal cells
Bandwidth 5 MHz
Carrier frequency 2 GHz
Mode of operation FDD
Number of RBs 25
Number of sub-carriers per RB 12
Total number of Sub-carriers 300
Sub-carrier spacing 15 kHz
Scheduling interval (TTI) 1 ms
Number of OFDMA symbols per TTI 14 (Normal CP)
Total number of REs 168
Total eNB transmit power 43.01 dBm
Erroneous CQI Delay and unavailable
User speed 120km/h
3. FD CHANNEL QUALITY PREDICTOR BASED ON KALMAN FILTER
In 1960, Kalman filter was proposed as an efficient recursive solution of the least-squares method
in [24]. The Kalman filter is the best possible (optimal) estimator for a wide variety of problems,
due to its simple and clear mathematical model with the robustness and efficiency in the
resolution of various problems.

161
3.1. Prediction Process
As an optimal autoregressive prediction method, Kalman filter has been applied to channel
condition prediction and it enables to forecast current channel state from the state in the last time
slot. Specially, the classical Kalman filter method includes two steps: estimation and correction.
In the estimation step, the current CQI value is estimated from the CQI in last TTI while the
estimated current CQI conditioned on last TTI is improved according to the observed CQI value
that is received from each user’s CQI report in the correction step. The equations are given below:
Estimation step:
( | − 1) = Φ ( − 1| − 1) (2)
( | − 1) = Φ ( − 1| − 1)Φ + (3)
where ( | − 1) is the estimate of X(t) based on the time t-1. is transition matrix that connects
two consecutive states, X(t) and X(t-1), of the system, ( | − 1) is the covariance of ( | − 1),
and ( − 1| − 1) is the covariance of ( − 1| − 1). The random variable Q denotes noise
error matrix. The aim of calculating ( | − 1) is to estimate the accuracy of ( | − 1). If
( | − 1) → 0, which means the estimate ( | − 1) is so close to the realistic CQI value that
only limited correction need to be made, otherwise, if ( | − 1) → 1, which means the estimate
( | − 1) is considered to be unreliable, and it will reduce to the observation.
Correction step:
( ) = ( ) + (4)
( ) = ( | − 1) [ ( | − 1) + ] (5)
( ) = ( | − 1) + ( )[ ( ) − ( | − 1)] (6)
( ) = [ − ( ) ] ( | − 1) (7)
Equation (4) calculates the observed value ( ) from the realistic channel, where H is the channel
gain, R is observed error matrix. In our situation, Z(t) can be obtained directly from the CQI value
reported by the users.
Equation (5) gives the Kalman gain ( ), where the superscript T represents transpose. Kalman
gain can be regarded as a weight factor that “decides” the amount of improvement to the estimate
( | − 1)
From the Equation (2) - (5), we can get the optimal estimation of current state ( ) from
Equation (6), which is an update from ( | − 1).
In addition, we need to update the covariance to use in the next packet scheduling interval, and
this process is given by Equation (7), where I is the identity matrix.
The procedure is shown in Figure. 1.

162
Figure 1: the procedure of Kalman filter
3.2. Initialization of Kalman Filter
Although few papers and books talk about initialization of Kalman filter, it is rather important to
mention that if either the matrix Q or R are disarranged once, they keep disarranged and will not
be updated and the Kalman process might give bad predictions [27].
At first, the expressions of state matrix X(t) and observed matrix Z(t) can be derived from the
Taylor series:
( ) = (0) + ̇(0) + ̈(0) + ⋯ (8)
where ̇(0) indicates the first derivative of x at initial time t=0, ̈(0) indicates the second
derivative of x at initial time t=0 etc. The Taylor series is generally reduced to:
( ) = (0) + ̇(0) + ̈(0) (9)
In channel condition prediction, x(0) is the Signal-to-Interference-plus-Noise-Ratio (SINR) of
user i on RB j at scheduling interval 0, denoted by , we can consider ̇(0) as the change rate of
SINR, represented as and ̈(0) is the gradient of , denoted by . , , are the observed
, , , respectively. On this basis, the state equation and observation equation can be
respectively established as follows, where T denotes the transpose.
( ) = ( , , )
( ) = ( , , )
Then, the transition matrix can be determined as:
Φ =
1 2
0 1
0 0 1
Kalman Filter
Estimation Step
Estimate current
state X(t|t-1)
calculate
covariance P(t|t-
1)
Correction Step
calculate
observed value
Z(t)
calculate Kalman
gain K(t)
calculate optimal
estimate X(t)
calculate
covariance P(t)

163
Thereafter, since the observed matrix is directly obtained from the state matrix, the channel gain
matrix can be determined as the identity matrix:
=
1 0 0
0 1 0
0 0 1
After the determination of state matrix, the observed matrix, the transition matrix and the channel
gain matrix, covariance matrix of the measurement error R can be initialized Let , ( ), , ( ),
, ( ) denote the random variable that describes the measurement error. The error measurement
covariance is
= [( , ( ), , ( ), , ( ) ) ( , ( ), , ( ), , ( ) )] (10)
The elements of the principal diagonal are the covariance of , ( ), , ( ), , ( ) respectively.
Other elements are the covariance of each two variables, since each variable is completely
independent. The matrix is:
=
, 0 0
0 , 0
0 0 ,
Finally, the state matrix ( − 1| − 1) can be simply set to zero. However, it is obvious that it is
not zero, hence, we get an error in the first estimation and we can set the covariance matrix of the
estimation error ( − 1| − 1) as:
=
0 0
0 0
0 0
4. TD GROUPING TECHNIQUE
Papers [13, 28] proved that packet scheduler can benefit from the use of two scheduling
algorithms. However, the simultaneous use of two different scheduling algorithms can lead to an
increase in scheduling processing time and computational complexity. A scheduling technique
that employs FD- PF and FD-M-LWDF alternately in adjacent TTIs is proposed
FD-PF packet scheduling algorithm selects a user that maximizes , ( ) on PRB j at TTI i:
, ( ) = , ( )
( )
(11)
( + 1) = 1 − ( ) + ∗ ( + 1) (12)
( + 1) = , ( + 1) ∗ , ( + 1)
, ( + 1) =
1 ℎ + 1
0 ℎ + 1
(13)

164
where , ( ) is the priority of user i on PRB j at scheduling interval t, , ( ) is the instantaneous
data rate of user i on PRB j at scheduling interval t, ( ) is the average throughput of user i at
scheduling interval t, ( + 1) is the indicator function of the event that packets of user i are
selected for transmission at scheduling interval t+1 and is a time constant. The design objective
of FD-PF algorithm is to maximize the long term throughput of the user whose current achievable
data rate is better compared to the average throughput.
FD- M-LWDF packet scheduling algorithm selects a user that maximizes , ( ) in Equation (14)
to receive its packet in each scheduling interval.
, ( ) = ∗ ( ) ∗
, ( )
( )
(14)
= −
( )
(15)
where , ( ) is the priority of user i on PRB j at scheduling interval t and is the QoS
requirement of user i, ( ) is the delay of the HoL packet of user i at scheduling interval t.
Moreover, , ( ) represents the instantaneous data rate of user i on PRB j at scheduling interval t
and ( ) is the average throughput of user i at scheduling interval t. In addition, indicates the
service-dependent PLR threshold of user i and is the service-dependent buffer delay threshold
of user i. FD-M-LWDF algorithm was proposed to support QoS of multiple real time data users
sharing a wireless channel. This algorithm is based on an overall consideration of packet delay
information, the instantaneous data rate and average throughput and supposed to have a
reasonably good overall performance.
The TD grouping technique divides the whole scheduling TTI into two groups: odd numbered
TTI group and even numbered TTI group. FD-PF and FD-M-LWDF algorithm are employed in
the two groups alternately. Each packet scheduling interval only employs one single scheduling
algorithm which reduces the computational complexity compared with other joint TD-FD packet
scheduling algorithms that prioritize users twice. Due to the fact that this algorithm uses two
different scheduling algorithms in long-term, it is supposed to have reasonable performance
improvement.
5. THE JOINT FD PREDICTION AND TD GROUPING PACKET SCHEDULING
ALGORITHM
The main principle of the proposed algorithm is: firstly, at each TTI, the optimal estimated CQI
value of each user is predicted by Kalman filter on each PRB, moreover the highest supported
data rate of each user on each PRB can be estimated, and then the scheduling algorithm employed
at this TTI is decided by the TD grouping technique which makes the scheduling decision based
on the estimated CQIs.
Based on the above analysis, the proposed packet scheduling algorithm is described in Figure 2.

165
Figure 2: Flow chart of the proposed algorithm
6. SIMULATION RESULTS
The metrics such as system throughput and service packet loss ratio (PLR) are used in this paper
to evaluate packet scheduling performance. System throughput is defined as the total size of
successfully received packets (in bits) at all users in the downlink divided by the simulation time.
The system throughput has following mathematical expression:
ℎ ℎ = ∑ ∑ ( ) (16)
where ( ) is the total size of correctly received packets (in bits) of user i at time t, T is the
total simulation time and N is the total number of users.
Service PLR is an important quality of service (QoS) metric, to satisfy QoS requirements, it has to
be kept below a threshold throughout a session. For real time applications, the threshold is 10
(i.e. QoS constraints of the GBR (guaranteed bit rate) services). The service PLR is defined as the
ratio of the total size of discarded packets of a service to the total size of all packets of a service
that have resided in the eNB buffer. The mathematical expression for the service PLR is given
below:
=
∑ ∑ ( )
∑ ∑ ( )
(17)
where PLR is the packet loss rate and ( ) represent the total size of discarded packets
(in bits) of user i of a service at time t. ( ) is the total size of all packets (in bits) that have
arrived into the eNB buffer of user i of a service at time t, N is the total number of users and T is
the total simulation time.

166
The simulation environment settings described in Table 1 is simulated. The FD-PF and FD-M-
LWDF schedulers mainly serve as reference schedulers. Figure 3 and Figure 4 show the PLR of
the FD-PF, FD-M-LWDF and the proposed algorithm with increasing number of users in the
system (system capacity) under perfect CQI and imperfect CQI, respectively. It was observed in
Figure 3 that the FD-PF algorithm has the worst PLR followed by the proposed algorithm and the
FD-M-LWDF algorithm. To satisfy the QoS of GBR services, the maximum number of users that
FD-PF algorithm can support is less than 30. Both of the proposed algorithm and the FD-M-
LWDF algorithm can support 30 users. The use of FD-M-LWDF scheduler in the TD grouping
based algorithm has significantly reduced the PLR compared with the FD-PF only scheduling and
achieved the same system capacity with FD-M-LWDF algorithm, it is due to the fact that they
consider packet delay information when making scheduling decision. Table 2 illustrates the
proposed algorithm has 116.94% and 6.31% improvement in service PLR over FD-PF and FD-M-
LWDF algorithm at 70 users, respectively.
Figure 4 shows that the proposed algorithm outperforms the other two algorithms and has the
lowest PLR when the reported CQIs have 3 ms delay and become unavailable at a regular interval
(10 ms). This is due to the use of the Kalman filter based CQI estimator. Packet scheduling
decisions are made based on the estimated CQI instead of the erroneous CQI. It can be seen in the
figure that both FD-PF and FD-M-LWDF scheduler have a substantial increase in PLR, which
leads to a degradation in system capacities that fall below 30 users. This illustrates that the FD-PF
and FD-M-LWDF scheduler are not tolerant of the imperfect CQI.
The system throughput of the three algorithms with increasing system capacity is shown in Figure
5 and Figure 6. A perfect CQI scenario is depicted in Figure 5 while Figure 6 demonstrates a
practical system model with imperfect CQI. It can be seen from Figure 5 that the differences of
the three algorithms are not obvious until the system capacity is above 50 users. FD-PF has the
worst throughput performance while the proposed algorithm and the FD-M-LWDF algorithm
have similar system throughput.
Similar to Figure 4, Figure 6 shows the proposed algorithm outperforms the two reference
algorithm when a practical channel condition with imperfect CQI is considered. An improvement
of 8.56% and 4.42% in system throughput over FD-PF and FD-M-LWDF algorithm with 70 users
is illustrated in Table 3, respectively.
Figure 3 PLR comparison with perfect CQI

167
Figure 4 PLR comparison with imperfect CQI
Figure 5 System throughput comparison with perfect CQI
Figure 6 System throughput comparison with imperfect CQI

168
Table 2 PLR at 70 users
PLR
Performance
improvement (%)
FD-PF algorithm 11.202 116.94
FD-M-LWDF algorithm 5.4898 6.31
Proposed algorithm 5.1637 -
Table 3 System throughput at 70 users
System throughput
(Mbps)
Performance
improvement (%)
FD-PF algorithm 5.6921 8.56
FD-M-LWDF algorithm 5.9174 4.42
Proposed algorithm 6.1792 -
7. CONCLUSIONS
This paper investigates the performance of three packet scheduling algorithms in the ideal and
practical downlink 3GPP LTE system, respectively. A novel packet scheduling algorithm based
on FD prediction and TD grouping technique for real time applications in the downlink LTE
system is proposed in this paper. The simulation results show that the proposed algorithm
outperforms FD-PF and M-LWDF algorithm in terms of service PLR and system throughput
when the CQI reports has delay and errors. This simple and efficient algorithm which is robust to
simultaneous multiple channel defects are proved to satisfy QoS requirements for real time users.
REFERENCES
[1] K. Kinoshita, "Current Status of “FDMA” 3G service and DoCoMo’s B3G Activities," in ICB3G –
2004, May 2004, pp. 13-21.
[2] 3GPP, Physical Layer Aspects for Evolved Universal Terrestrial Radio Access
(UTRA) (Release 7), 2009.
[3] A. Jalali, R. Padovani, and R. Pankaj, "Data throughput of CDMA-HDR a high efficiency-high data
rate personal communication wireless system," in Vehicular Technology Conference Proceedings,
2000. VTC 2000-Spring Tokyo. 2000 IEEE 51st, 2000, pp. 1854-1858.
[4] M. Stambaugh, "HARQ Process Boosts LTE Communications," Agilent Technologies Sep, 2008.
[5] H. A. M. Ramli, R. Basukala, K. Sandrasegaran, and R. Patachaianand, "Performance of well known
packet scheduling algorithms in the downlink 3GPP LTE system," in Communications (MICC), 2009
IEEE 9th Malaysia International Conference on, 2009, pp. 815-820.
[6] A. Pokhariyal, T. E. Kolding, and P. E. Mogensen, "Performance of Downlink Frequency Domain
Packet Scheduling for the UTRAN Long Term Evolution," in Personal, Indoor and Mobile Radio
Communications, 2006 IEEE 17th International Symposium on, 2006, pp. 1-5.
[7] J. Gross, J. Klaue, H. Karl, and A. Wolisz, "Subcarrier allocation for variable bit rate video streams in
wireless OFDM systems," in Vehicular Technology Conference, 2003. VTC 2003-Fall. 2003 IEEE
58th, 2003, pp. 2481-2485 Vol.4.
[8] N. Ruangchaijatupon and J. Yusheng, "Simple Proportional Fairness Scheduling for OFDMA Frame-
Based Wireless Systems," in Wireless Communications and Networking Conference, 2008. WCNC
2008. IEEE, 2008, pp. 1593-1597.
[9] P. Jeongsik, H. Sungho, and C. Ho-Shin, "A Packet Scheduling Scheme to Support Real-Time Traffic
in OFDMA Systems," in Vehicular Technology Conference, 2007. VTC2007-Spring. IEEE 65th,
2007, pp. 2766-2770.

169
[10] X. Liu, L. Guangyi, Y. Wang, and P. Zhang, "Downlink Packet Scheduling for Real-Time Traffic in
Multi-User OFDMA System," in Vehicular Technology Conference, 2006. VTC-2006 Fall. 2006
IEEE 64th, 2006, pp. 1-5.
[11] A. Wang, L. Xiao, S. Zhou, X. Xu, and Y. Yao, "Dynamic resource management in the fourth
generation wireless systems," in Communication Technology Proceedings, 2003. ICCT 2003.
International Conference on, 2003, pp. 1095-1098 vol.2.
[12] A. Pokhariyal, K. I. Pedersen, G. Monghal, I. Z. Kovacs, C. Rosa, T. E. Kolding, and P. E. Mogensen,
"HARQ Aware Frequency Domain Packet Scheduler with Different Degrees of Fairness for the
UTRAN Long Term Evolution," in Vehicular Technology Conference, 2007. VTC2007-Spring. IEEE
65th, 2007, pp. 2761-2765.
[13] B. Kian Chung, S. Armour, and A. Doufexi, "Joint Time-Frequency Domain Proportional Fair
Scheduler with HARQ for 3GPP LTE Systems," in Vehicular Technology Conference, 2008. VTC
2008-Fall. IEEE 68th, 2008, pp. 1-5.
[14] R. A. Akl, S. Valentin, G. Wunder, and S. Stanczak, "Compensating for CQI Aging By Channel
Prediction: The LTE Downlink."
[15] M. R. Souryal and R. L. Pickholtz, "Adaptive modulation with imperfect channel information in
OFDM," in Communications, 2001. ICC 2001. IEEE International Conference on, 2001, pp. 1861-
1865 vol.6.
[16] W. Yafeng and Y. Hongwen, "Retransmission priority scheduling algorithm for forward link packet
data service," in Communication Technology Proceedings, 2003. ICCT 2003. International
Conference on, 2003, pp. 926-930.
[17] I. C. Wong and B. Evans, "Optimal resource allocation in the OFDMA downlink with imperfect
channel knowledge," Communications, IEEE Transactions on, vol. 57, pp. 232-241, 2009.
[18] D. S. W. Hui and V. K. N. Lau, "Delay-sensitive cross-layer designs for OFDMA systems with
outdated CSIT," in Wireless Communications and Networking Conference, 2007. WCNC 2007.
IEEE, 2007, pp. 264-269.
[19] H. Ayoub and M. Assaad, "Scheduling in OFDMA Systems With Outdated Channel Knowledge," in
Communications (ICC), 2010 IEEE International Conference on, 2010, pp. 1-5.
[20] T. Villa, R. Merz, R. Knopp, and U. Takyar, "Adaptive modulation and coding with hybrid-ARQ for
latency-constrained networks," in European Wireless, 2012. EW. 18th European Wireless
Conference, 2012, pp. 1-8.
[21] J.-H. Jeon and Y.-H. Lee, "Prediction of channel information in multi-user OFDM systems," in
Circuits and Systems, 2006. MWSCAS'06. 49th IEEE International Midwest Symposium on, 2006,
pp. 468-472.
[22] J. Bonnet and G. Auer, "Chunk-based channel estimation for uplink OFDM," in Vehicular
Technology Conference, 2006. VTC 2006-Spring. IEEE 63rd, 2006, pp. 1555-1559.
[23] D. Aronsson, T. Svensson, and M. Sternad, "Performance evaluation of memory-less and Kalman-
based channel estimation for OFDMA," in Vehicular Technology Conference, 2009. VTC Spring
2009. IEEE 69th, 2009, pp. 1-5.
[24] R. E. Kalman, "A new approach to linear filtering and prediction problems," Journal of basic
Engineering, vol. 82, pp. 35-45, 1960.
[25] A. G. Orozco-Lugo, F. A. Cruz Prez, and G. Hernandez Valdez, "Investigating the boundary effect of
a multimedia TDMA personal mobile communication network simulation," in Vehicular Technology
Conference, 2001. VTC 2001 Fall. IEEE VTS 54th, 2001, pp. 2740-2744 vol.4.
[26] K. I. Pedersen, G. Monghal, I. Z. Kovacs, T. E. Kolding, A. Pokhariyal, F. Frederiksen, and P.
Mogensen, "Frequency Domain Scheduling for OFDMA with Limited and Noisy Channel Feedback,"
in Vehicular Technology Conference, 2007. VTC-2007 Fall. 2007 IEEE 66th, 2007, pp. 1792-1796.
[27] M. Kohler, Using the Kalman filter to track human interactive motion: modelling and initialization of
the Kalman filter for translational motion: Citeseer, 1997.
[28] G. Monghal, K. I. Pedersen, I. Z. Kovacs, and P. E. Mogensen, "QoS Oriented Time and Frequency
Domain Packet Schedulers for The UTRAN Long Term Evolution," in Vehicular Technology
Conference, 2008. VTC Spring 2008. IEEE, 2008, pp. 2532-2536.

170
Authors
Yongixn Wang is currently a Master of Engineering by research Candidate in the Faculty of Engineering
and Information Technology, University of Technology, Sydney (UTS), Australia. She received a Master of
Engineering Degree in Software Engineering from Huazhong University of Science and Technology (2007)
and Bachelor of Science Degree in Electrical and Electronic Engineering from Wuhan University of
Science and Technology (2004). Her current research interests focus on packet scheduling in radio resource
management for the future wireless networks
Kumbesan Sandrasegaran (Sandy) holds a PhD in Electrical Engineering from McGill University
(Canada) (1994), a Master of Science Degree in Telecommunication Engineering from Essex University
(UK) (1988) and a Bachelor of Science (Honours) Degree in Electrical Engineering (First Class) (UZ)
(1985). He was a recipient of the Canadian Commonwealth Fellowship (1990-1994) and British Council
Scholarship (1987-1988). He is a Professional Engineer (Pr. Eng) and has more than 20 years experience
working either as a practitioner, researcher, consultant and educator in telecommunication networks.
During this time, he has focused on the planning, modelling, simulation, optimisation, security, and
management of telecommunication networks.
Xinning Zhu is currently a visiting scholar in the Faculty of Engineering and Information Technology,
University of Technology, Sydney (UTS), Australia. She received her PhD, MS and BS degree in
Communication and Information System from Beijing University of Posts and Telecommunications
(BUPT) in 2010, 1995 and 1992. She is an associate professor at School of Information and
Communication Engineering, BUPT, China. Her current research interests focus on interference
management and mobility management in radio resource management for heterogeneous networks
Jingjing Fei is currently a Master of Engineering by research Candidate in the Computer Science and
Engineering, University of New South Wales (UNSW), Australia. He received a Master of Engineering
Degree in Software Engineering from Huazhong University of Science and Technology (2007) and
Bachelor of Science Degree in Computer Science and Technology from Wuhan University of Technology
(2004). His current research interests focus on wireless sensor networks.
Xiaoying Kong has broad interests in control engineering and software engineering. Her recent research
work has been in GPS, inertial navigation systems, data fusion, robotics, sensor networks, Internet
monitoring systems, agile software development methodologies, web technologies, web design
methodologies, web architecture framework, analytical models of software development, and value-based
software engineering. She has many years work experience in aeronautical industry, semiconductor
industry, and software industry.
Cheng-Chung Lin is currently a PhD Candidate in the Faculty of Engineering and Information
Technology, University of Technology, Sydney (UTS), Australia. He received a graduate certificate
(GradCert) in advanced computing (2006) and a Master of Information Technology in internetworking in
School of Computer Science and Engineering from University of New South Wales, Australia (2007). His
current research interests focus on handover and packet scheduling in radio resource management for the
future wireless networks

FREQUENCY AND TIME DOMAIN PACKET SCHEDULING BASED ON CHANNEL PREDICTION WITH IMPERFECT CQI IN LTE

More Related Content

What's hot (20)

Viewers also liked (6)

Similar to FREQUENCY AND TIME DOMAIN PACKET SCHEDULING BASED ON CHANNEL PREDICTION WITH IMPERFECT CQI IN LTE (20)

Recently uploaded (20)

FREQUENCY AND TIME DOMAIN PACKET SCHEDULING BASED ON CHANNEL PREDICTION WITH IMPERFECT CQI IN LTE