Link tải luận văn miễn phí cho ae Kết Nối
Abstract—In this study we analyze the downlink OFDMA system level
performance for three different channel quality indicator (CQI) reporting schemes. The effect of terminal measurement and estimation errors,
quantization from formatting and compression, and uplink reporting delays and detection errors are included. We find that a simple thresholdbased CQI scheme provides an attractive trade-off between downlink system level performance and uplink CQI signaling overhead, as compared
to using a Best-M scheme. When applied to the UTRAN LTE system in
a 10 MHz bandwidth, we find that a frequency domain packet scheduling
gain of 40% is achievable with a CQI word size of only 30-bits. Finally,
the effect of applying a so-called outer loop link adaptation algorithm is
reported.
I. INTRODUCTION
In this study we are addressing the downlink performance
of cellular systems using orthogonal frequency division multiple access (OFDMA) in combination with adaptive modulation coding and frequency domain packet scheduling (FDPS).
Application of FDPS with optimal allocation of sub-carriers,
modulation, and coding scheme to the different users is one
of the techniques that can boost the performance of such systems [1]-[2]. However, the cost of achieving these gains is increased signaling overhead, and especially the required channel
feedback from the terminals to the base station. Ideally, each
terminal should measure the signal-to-interference-noise-ratio
(SINR) for each sub-carrier, and report that information back
to the base station at a high reporting rate. However, this would
obviously result in excessive control signaling overhead, especially for configurations with many users per cell. In this study
we therefore investigate several candidate channel quality indicator (CQI) schemes in order to evaluate how much we can
reduce the feedback information while still being able to gain
from FDPS and fast link adaptation via variable modulation
and coding. In this analysis we include the effect of the timefrequency variant radio channel, terminal measurement and estimation imperfections, simple compression techniques before
sending the information from the terminal to the base station,
feedback delays, etc. Examples of previous CQI design and
performance studies for OFDMA systems include [3]-[7].
As a case study, we use the UTRAN long term evolution
(LTE) [8] as an example with the system parameter settings
listed in [9]. Hence, the assumed sub-carrier spacing is 15 kHz,
where sets of 12 adjacent sub-carriers are grouped into physical resource blocks (PRB) of 180 kHz. The frequency domain
scheduling resolution for multiplexing users is one PRB, so this
implicitly limits the required channel feedback to one quantity
per PRB at most.
The rest of the paper is organized as follows: The overall
modeling framework and the proposed CQI schemes are described in Section II. The simulation methodology is outlined
in Section III, while the performance results are reported in Section IV. Concluding remarks are given in Section V.
II. CHANNEL FEEDBACK SCHEMES
A. Overall control loop
The control loop between the serving cell (base station) and
the user equipment (UE) is illustrated in Fig. 1. The UE measures the frequency selective channel quality in the downlink
every Tcqi seconds. The measurement result is afterwards formatted and compressed to a finite number of bits (called the
CQI word) before being reported in the uplink. The reporting
is subject to a delay of Tdelay seconds, including both the time
it takes to send the CQI report and the time it takes to decode
the CQI report at the serving cell. The received CQI word at
the serving cell is subject to errors due to imperfect decoding
of the received signal. Finally, the received CQI words from
all the UEs in the cell are used for radio resource managament
(RRM) decisions such as packet scheduling and link adaptation.
RRM
functions
CQI word
detection
Time-variant
radio channel
Uplink
reporting
delay
Thermal
noise
Measurement
unit
Formatting
and
compression
Time-variant
other-cell
interference
+
Serving cell UE
CQI word
Fig. 1. Simplified block diagram showing the control loop for downlink packet
scheduling and link adaptation based on noisy and partial feedback information from the terminals (UEs).
B. Measurement model
For the sake of simplicity, we assume that the raw measured
CQI is a post detection SINR. The SINR is measured periodically for K sub-bands, each with a bandwidth of ∆fcqi and an
averaging time of ∆tcqi. The values of ∆fcqi and ∆tcqi shall be
sufficiently small to capture both the frequency selectivity and
the time-variant behavior of the downlink channel quality. On
the contrary, if the two parameters are set at too small values,
the SINR measurement will be subject to excessive errors due
to scarcity of pilot reference symbols for SINR estimation. In
this study we rely on the conclusions from [5] and [10], where
it is found that setting ∆fcqi = 360 kHz and ∆tcqi = 2 ms is
a reasonable choice for LTE. Hence, ∆fcqi corresponds to 24
sub-carriers, or equivalently two PRBs. Given these settings, it
has been found from link simulations that the measured SINR
expressed in decibels can be approximated with the ideal value
and an additive independent identically distributed zero mean
Gaussian error with σcqi = 1 dB standard deviation.
C. Formatting and compression schemes
Assuming full CQI reporting, a quantized phiên bản of all
the measured SINRs accross the entire system bandwidth are
reported in the uplink. Assuming K measurement intervals
and quantization to Q-bits, this results in a CQI word size of
WF ullCQI = K · Q bits per reporting. The dynamic range
of the LTE modulation and coding schemes from QPSK with
excessive coding to 64QAM with marginal coding is approximately 25 dB [11], which means that Q = 5 bits are required to
achieve a ∆cqi ≤ 1 dB quantization step. Hence, for a 10 MHz
system bandwidth with K = 25, WF ullCQI = 125-bits per
CQI word. The latter represents a fairly large uplink signaling
overhead (125 kbps for CQI reporting every 1 ms), which can
be further reduced by considering the following two frequency
domain compression schemes.
The basic principle of the Threshold-based CQI scheme is illustrated in Fig. 2 [5]. Here the highest measured CQI value
is identified, and all the other measurements within a threshold window of X dB are averaged (the dark grey bars in Fig.
2). Both the average value and a bit mask indicating which
measurements are within the threshold window are reported to
the serving cell as part of the CQI word. Hence, the thresholdbased scheme requires Wthreshold = Q+K bits. The threshold
value, X, is considered to be a parameter set by the network and
signaled to the UE.
The last CQI compression scheme we consider is called the
Best-M scheme [6]. Here the M raw CQI measurements with
the highest SINR are reported together with a bit mask with the
frequency position of those measurements. The M parameter
is considered to be a management parameter set by the network
and signaled to the UE. The Best-M scheme requires a CQI
word size of1,
WBestM = Q · M + Ceil log2 MK . (1)
Given the three considered CQI reporting schemes, the CQI
word size is compared in Fig. 3 for a 10 MHz system bandwidth with K = 25 measurement intervals and Q = 5-bits. The
CQI word size is clearly reduced for both the Best-M scheme
and the Threshold-based scheme as compared to the full CQI
reporting. However, the penalty for the reduced CQI word size
is a more coarse frequency selective CQI measurement, which
limits the ability of the RRM algorithms to take advantage of
the frequency selective channel variations for the different UEs.
The CQI word size for Best-M is depending on the M parameter setting, which indicates that the required uplink CQI bandwidth varies depending on this parameter configuration. On the
contrary, the CQI word size for the Threshold-based scheme is
independent of the threshold setting (X).
Full CQI reporting
Best-M scheme
Threshold-based scheme
0 5 10 15 20 25
M paramet
Do Drive thay đổi chính sách, nên một số link cũ yêu cầu duyệt download. các bạn chỉ cần làm theo hướng dẫn.
Password giải nén nếu cần: ket-noi.com | Bấm trực tiếp vào Link để tải:
Abstract—In this study we analyze the downlink OFDMA system level
performance for three different channel quality indicator (CQI) reporting schemes. The effect of terminal measurement and estimation errors,
quantization from formatting and compression, and uplink reporting delays and detection errors are included. We find that a simple thresholdbased CQI scheme provides an attractive trade-off between downlink system level performance and uplink CQI signaling overhead, as compared
to using a Best-M scheme. When applied to the UTRAN LTE system in
a 10 MHz bandwidth, we find that a frequency domain packet scheduling
gain of 40% is achievable with a CQI word size of only 30-bits. Finally,
the effect of applying a so-called outer loop link adaptation algorithm is
reported.
I. INTRODUCTION
In this study we are addressing the downlink performance
of cellular systems using orthogonal frequency division multiple access (OFDMA) in combination with adaptive modulation coding and frequency domain packet scheduling (FDPS).
Application of FDPS with optimal allocation of sub-carriers,
modulation, and coding scheme to the different users is one
of the techniques that can boost the performance of such systems [1]-[2]. However, the cost of achieving these gains is increased signaling overhead, and especially the required channel
feedback from the terminals to the base station. Ideally, each
terminal should measure the signal-to-interference-noise-ratio
(SINR) for each sub-carrier, and report that information back
to the base station at a high reporting rate. However, this would
obviously result in excessive control signaling overhead, especially for configurations with many users per cell. In this study
we therefore investigate several candidate channel quality indicator (CQI) schemes in order to evaluate how much we can
reduce the feedback information while still being able to gain
from FDPS and fast link adaptation via variable modulation
and coding. In this analysis we include the effect of the timefrequency variant radio channel, terminal measurement and estimation imperfections, simple compression techniques before
sending the information from the terminal to the base station,
feedback delays, etc. Examples of previous CQI design and
performance studies for OFDMA systems include [3]-[7].
As a case study, we use the UTRAN long term evolution
(LTE) [8] as an example with the system parameter settings
listed in [9]. Hence, the assumed sub-carrier spacing is 15 kHz,
where sets of 12 adjacent sub-carriers are grouped into physical resource blocks (PRB) of 180 kHz. The frequency domain
scheduling resolution for multiplexing users is one PRB, so this
implicitly limits the required channel feedback to one quantity
per PRB at most.
The rest of the paper is organized as follows: The overall
modeling framework and the proposed CQI schemes are described in Section II. The simulation methodology is outlined
in Section III, while the performance results are reported in Section IV. Concluding remarks are given in Section V.
II. CHANNEL FEEDBACK SCHEMES
A. Overall control loop
The control loop between the serving cell (base station) and
the user equipment (UE) is illustrated in Fig. 1. The UE measures the frequency selective channel quality in the downlink
every Tcqi seconds. The measurement result is afterwards formatted and compressed to a finite number of bits (called the
CQI word) before being reported in the uplink. The reporting
is subject to a delay of Tdelay seconds, including both the time
it takes to send the CQI report and the time it takes to decode
the CQI report at the serving cell. The received CQI word at
the serving cell is subject to errors due to imperfect decoding
of the received signal. Finally, the received CQI words from
all the UEs in the cell are used for radio resource managament
(RRM) decisions such as packet scheduling and link adaptation.
RRM
functions
CQI word
detection
Time-variant
radio channel
Uplink
reporting
delay
Thermal
noise
Measurement
unit
Formatting
and
compression
Time-variant
other-cell
interference
+
Serving cell UE
CQI word
Fig. 1. Simplified block diagram showing the control loop for downlink packet
scheduling and link adaptation based on noisy and partial feedback information from the terminals (UEs).
B. Measurement model
For the sake of simplicity, we assume that the raw measured
CQI is a post detection SINR. The SINR is measured periodically for K sub-bands, each with a bandwidth of ∆fcqi and an
averaging time of ∆tcqi. The values of ∆fcqi and ∆tcqi shall be
sufficiently small to capture both the frequency selectivity and
the time-variant behavior of the downlink channel quality. On
the contrary, if the two parameters are set at too small values,
the SINR measurement will be subject to excessive errors due
to scarcity of pilot reference symbols for SINR estimation. In
this study we rely on the conclusions from [5] and [10], where
it is found that setting ∆fcqi = 360 kHz and ∆tcqi = 2 ms is
a reasonable choice for LTE. Hence, ∆fcqi corresponds to 24
sub-carriers, or equivalently two PRBs. Given these settings, it
has been found from link simulations that the measured SINR
expressed in decibels can be approximated with the ideal value
and an additive independent identically distributed zero mean
Gaussian error with σcqi = 1 dB standard deviation.
C. Formatting and compression schemes
Assuming full CQI reporting, a quantized phiên bản of all
the measured SINRs accross the entire system bandwidth are
reported in the uplink. Assuming K measurement intervals
and quantization to Q-bits, this results in a CQI word size of
WF ullCQI = K · Q bits per reporting. The dynamic range
of the LTE modulation and coding schemes from QPSK with
excessive coding to 64QAM with marginal coding is approximately 25 dB [11], which means that Q = 5 bits are required to
achieve a ∆cqi ≤ 1 dB quantization step. Hence, for a 10 MHz
system bandwidth with K = 25, WF ullCQI = 125-bits per
CQI word. The latter represents a fairly large uplink signaling
overhead (125 kbps for CQI reporting every 1 ms), which can
be further reduced by considering the following two frequency
domain compression schemes.
The basic principle of the Threshold-based CQI scheme is illustrated in Fig. 2 [5]. Here the highest measured CQI value
is identified, and all the other measurements within a threshold window of X dB are averaged (the dark grey bars in Fig.
2). Both the average value and a bit mask indicating which
measurements are within the threshold window are reported to
the serving cell as part of the CQI word. Hence, the thresholdbased scheme requires Wthreshold = Q+K bits. The threshold
value, X, is considered to be a parameter set by the network and
signaled to the UE.
The last CQI compression scheme we consider is called the
Best-M scheme [6]. Here the M raw CQI measurements with
the highest SINR are reported together with a bit mask with the
frequency position of those measurements. The M parameter
is considered to be a management parameter set by the network
and signaled to the UE. The Best-M scheme requires a CQI
word size of1,
WBestM = Q · M + Ceil log2 MK . (1)
Given the three considered CQI reporting schemes, the CQI
word size is compared in Fig. 3 for a 10 MHz system bandwidth with K = 25 measurement intervals and Q = 5-bits. The
CQI word size is clearly reduced for both the Best-M scheme
and the Threshold-based scheme as compared to the full CQI
reporting. However, the penalty for the reduced CQI word size
is a more coarse frequency selective CQI measurement, which
limits the ability of the RRM algorithms to take advantage of
the frequency selective channel variations for the different UEs.
The CQI word size for Best-M is depending on the M parameter setting, which indicates that the required uplink CQI bandwidth varies depending on this parameter configuration. On the
contrary, the CQI word size for the Threshold-based scheme is
independent of the threshold setting (X).
Full CQI reporting
Best-M scheme
Threshold-based scheme
0 5 10 15 20 25
M paramet
Do Drive thay đổi chính sách, nên một số link cũ yêu cầu duyệt download. các bạn chỉ cần làm theo hướng dẫn.
Password giải nén nếu cần: ket-noi.com | Bấm trực tiếp vào Link để tải:
You must be registered for see links