Design and Analysis of LTE Physical Downlink Control Channel

Jialing Liu, Robert Love, Kenneth Stewart, Michael Eoin Buckley

Standards and Research Lab, Mobile Devices, Motorola Inc.
{jialingliu, robert.love, ken.stewart, e.buckley}@motorola.com).
AbstractLong Term Evolution (LTE) of UMTS Terrestrial
Radio Access and Radio Access Network is a Fourth Gener-
ation (4G) wireless broadband technology which is capable of
delivering high data transmission rates and low latency with
reduced costs, among other promises. A key research topic
is to design efcient and reliable Physical Downlink Control
Channels (PDCCHs) that carry the downlink scheduling as-
signments and uplink scheduling grants. A number of issues
regarding the design and analysis of PDCCHs are discussed in
this paper, such as the PDCCH interleaver, power boosting, power
lending, receiver handling of power boosting with unknown
boosting values, link level performance evaluation, and system
level performance evaluation. Based on mathematical analysis
and/or simulations, conclusions have been reached, including the
adoption of the simple subblock interleaver to achieve desired
interference randomization property; power boosting and power
boosting limits; and a method of handling power boosting values
unknown to the receiver in system simulations. These conclusions
have been utilized to develop the LTE standards or adopted in
LTE standards.
I. INTRODUCTION
Long Term Evolution (LTE) of Universal Mobile Telecom-
munications System (UMTS) Terrestrial Radio Access and
Radio Access Network is a Fourth Generation (4G) wireless
broadband technology. It is aimed at commercial deployment
around 2010 timeframe and will offer high peak data rates
(100 Mbps downlink and 50 Mbps uplink), low latency (10ms
round-trip delay), improved system capacity and coverage, re-
duced operating costs, multi-antenna support, efcient support
for packet data transmission, exible bandwidth operations (up
to 20 MHz), and seamless integration with existing systems
[1]. This necessitates an efcient and reliable design of control
channels. In the downlink (DL), this control signalling encom-
passes scheduling grants, a control format indicator, and Hy-
brid Automatic Repeat-ReQuest (H-ARQ) acknowledgments;
see a recent overview [2] and references therein. This paper
is focused on the design and analysis of Physical Downlink
Control Channels (PDCCHs) that carry the DL scheduling
assignments and UL (uplink) scheduling grants.
Several issues regarding the design and analysis of PDCCHs
are discussed in this paper. In Sec. II some preliminaries
about LTE and LTE control channels are presented. Sec. III
studies the PDCCH interleaver. In Sec. IV, power boosting,
power lending, receiver handling of power boosting with
unknown boosting values are investigated. Based on math-
ematical analysis and/or simulations, conclusions have been
reached, including the simple subblock interleaver to achieve
desired interference randomization property; power boosting
and power boosting limits; and a method of handling power
boosting values unknown to the receiver in system simulations.
II. LTE CONTROL CHANNEL PRELIMINARIES
LTE is OFDM based whose frame structure, general pa-
rameters, and general settings can be found in e.g. [2][5].
Each DL frame lasts 10 ms and contains 10 subframes. Each
subframe is comprised of two slots. Each slot contains 7
OFDM symbols, and the 0th and 4th OFDM symbols are pilot-
bearing. DL control signalling is located in the rst n OFDM
symbols of a subframe with n 3 and with the earliest data
transmission starting at the same OFDM symbol as the control
signalling ends.
DL control signalling is carried by three physical channels:
1) Physical Control Format Indicator Channel (PCFICH) to
indicate n, the number of OFDM symbols used for control in
this subframe, 2) Physical H-ARQ Indicator Channel (PHICH)
which carries DL ACK/NACK (Acknowledgment/Negative
Acknowledgment) associated with uplink data transmission,
and 3) Physical Downlink Control Channel (PDCCH) which
carries the DL scheduling assignments and UL scheduling
grants. Each subframe has multiple PDCCHs. A User Equip-
ment (UE), if scheduled, may be assigned one or more
PDCCHs.
The smallest element for PDCCH transmission is a Control
Channel Element (CCE). One PDCCH may contain one or
several CCEs. Each CCE has 9 so called mini-CCEs or
Resource Element Groups (REGs), and each REG has 4 non-
pilot resource elements (REs, also known as subcarriers).
QPSK is always used for PDCCH. Thus, one CCE contains 72
coded bits. Each PDCCH carries information associated with
one UE and may use different control formats with possibly
different payload sizes, and UEs monitor at least two formats
per subframe through blind detection (cf. Sec. 5.3.3.1 in [5]).
III. MAPPING CCES TO PHYSICAL RES
It is desired to map the 9 REGs of each CCE to physical
RE locations distributed throughout the time and frequency
resource. This is useful to improve 1) time and frequency
diversity, 2) inter-cell interference randomization when not
all CCEs are used, and 3) power balancing among n control
symbols when not all CCEs are used. To this end, the REGs
are interleaved at the Base Station (BS) transmitter and dein-
terleaved at the UE receiver, where the interleaver pattern also
depends on the identity of the BS (cell ID). When mapping
the interleaved REG sequence to physical REs, this is done in
a time rst fashion at the BS, that is, rst across n OFDM
symbols for a xed frequency and then across frequencies.
Three basic approaches to interleaver design may be used:
Approach 1: Block Interleaver. Approach 2: Random Permu-
tation and Number Addition. The basis for approach 2 is to
978-1-4244-2517-4/09/$20.00 2009 IEEE
use a single random permutation common to all cells. A cell-
specic permutation could then be derived from this random
permutation by adding certain values to the permutation se-
quence (with modulo N if applicable, where N is the total
number of REGs to be interleaved). Approach 3: Random
Permutation and Cyclic Sequence Shift, which cyclically shifts
the common permutation pattern by a cell-dependent offset.
Many interleaver designs were proposed and studied under
the last two approaches, such as Costas sequence interleaver,
QPP interleaver, subblock interleaver, etc.; see [6], [7].
A. Subblock interleaver with cyclic shift
The subblock interleaver already exists in the LTE speci-
cation for turbo coding purpose and can offer an appropriate
randomization property with low complexity. The subblock
interleaver is described as follows. Let L = 32. Dene
matrix Y R
KL
where K = N/L|, satisfying Y
ij
=
(i 1)L + (j 1). Then Y is column-permuted to form
Z such that Z
ij
= Y
i,j
where j

= (br(j) + L/2)%L,
br() is the bit reverse function of bitwidth 5, and % is
the modulo operation. Then let z = vec(Z) (LK N);
here vec() is the vectorization operator which stacks the
columns of a matrix below each other. Finally one obtains the
subblock common interleaver sequence p
i

N1
i=0
by removing
all negative numbers in z.
A cell-specic shift is then used to cyclically shift the
common interleaver sequence. That is, if the cell ID is J, then
a cell-specic interleaver sequence q
(J)
i

N1
i=0
with q
(J)
i
=
p
(i+J)%N
is generated. If the pre-interleaving input sequence
of quadruplets (recalling each REG has 4 QPSK symbols)
is w
i

N1
i=0
, then the interleaved sequence of quadruplets
w
i

N1
i=0
is such that w
i
= w
j
with j = q
(J)
i
.
B. Persistent collision metric
To see the interference randomization generated from
the CCE to RE mapping, it may be useful, before per-
forming system level simulations, to analyze the proper-
ties of REG sequences after the interleaving and time-
rst assignment. Suppose such a REG sequence is de-
noted as r
(J)
where J is the cell ID. Let the number
of REGs in a CCE be M (M = 9 in LTE). It is
proposed in this paper to adopt Hamming-correlation-type
persistent collision metrics as follows: Persistent self col-
lision metric C
J1,J2
=

N1
i=0
(r
(J1)
i
/M|, r
(J2)
i
/M|),
and persistent collision metric with offset C
J1,J2
(j) =

N1
i=0
(r
(J1)
i
/M|, (r
(J2)
i
/M| +j)%N/M|). Here () is
the Dirac funciton, r
(J1)
i
/M| is the CCE index of the ith REG
for cell J
1
, and N/M| is the number of CCEs in the control
region. In other words, C
J1,J2
measures the persistent collision
between the kth CCE in one cell with the kth CCE in the other
cell, for any k; and C
J1,J2
(j) measures the persistent collision
between the kth CCE in one cell with the [(k+j)%N/M|]th
CCE in the other cell, for any k
1
.
1
To be absolutely precise, the sequences with PDCCH REGs, PFICH REGs,
PHICH REGs, and punctured REGs (if any) need to be considered, which was
done in generating Fig. 1 but omitted here for notation simplicity.
The persistent collision metrics of different types of in-
terleavers were examined. Fig. 1 shows that the subblock
interleaver leads to excellent randomization for most of the
cell IDs (N = 192 and PDCCH 20% loaded).
50 100 150 200 250 300 350 400 450 500
0
5
10
15
Cell ID
C
o
l
l
i
s
i
o
n
s
Fig. 1. Persistent collision metric C
0,J
, J = 1, , 503 (example)
It is interesting to point out that the average number of
persistent collisions (averaged over all possible cyclic shifts
of the interleaver sequences) is the same for all interleavers
if PCFICH and PHICH are not considered. Namely, any
interleaver design is subject to the well-known waterbed
effect, or push-pop effect, which can be regarded as the
fundamental limitation of interference randomization. That is,
if for some cell IDs the numbers of collisions are low, then
for some other cell IDs the numbers of collisions must be
correspondingly high.
Proposition 1. If the REG sequence q
(J)
has length N, and
there are M REGs in a CCE (M = 9 in LTE), then it holds
that, for independent identically distributed uniform random
variables J
1
and J
2
,
E
J1,J2
C
J1,J2
= (M 1)
N
N 1
E
J1,J2
C
J1,J2
(j) = M 1, if j ,= 0.
Namely, the mean number of REGs of a CCE that persistently
collide with any REG of the same CCE in a neighboring cell
is constant, and so is the mean number of REGs of a CCE
which persistently collide with any REG of a different CCE
in a neighboring cell, for any interleaver design.
C. System performance
LTE system simulation was used to compare the different
interleavers. The simulation methodology includes modeling
both scheduling grants (36-bits) and scheduling assignments
(47-bits) where at most 6 scheduling grants and 7 scheduling
assignments can be sent per subframe for the 5 MHz carrier
mode modeled assuming n = 2, 2 transmit antennas, and
normal Cyclic Prex (CP). There were 13 CCEs of size 36
REs used for PDCCHs. All other REGs (8) were allocated for
PCFICH (4), PHICH(3), or punctured(1). More details about
the simulation assumptions can be found in Table V.
Interleavers are modeled in every sector (19 cell sites
with 3 sectors each) when computing interference and SINR.
The Exponential Effective SINR Mapping (EESM) link error
prediction (LEP) approach was used (not shown due to page
limitation) as well as an actual instantiation of a K = 7
convolutional decoder (which can rule out the possibility that
the EESM LEP might miss some variation effecting LEP
accuracy; however it was seen that the two approaches indeed
generated similar performance results).
TABLE I
THROUGHPUT PERFORMANCE (in kbps) OF INTERLEAVERS FOR LTE
DEPLOYMENT CASES 1 AND 3 WITH n = 2 AND DIFFERENT MAX
NUMBERS OF UES SCHEDULED PER SUBFRAME
Max
CDM
Ideal Block
App2,
QPP
App3,
QPP
App2,
Subblock
App3,
Subblock
CRRI PBRI Costas CSPI
1 5327.4 5327.4 5327.4 5327.4 5327.4 5327.4 5327.4 5327.4 5327.4 5327.4
4 6369.1 6315.1 6334.9 6330.7 6337.3 6330.7 6329.2 6329.6 6326.0 6330.1
6 6433.2 6281.8 6309.1 6301.9 6313.4 6303.0 6301.4 6302.0 6296.0 6303.4
1 152.6 153.6 153.6 153.6 153.6 153.6 153.6 153.6 153.6 153.6
4 193.9 187.5 189.8 189.1 189.3 187.9 188.4 188.5 187.8 189.0
6 229.0 192.3 192.8 193.7 194.1 193.5 192.5 193.3 193.2 192.6
1 5092.0 5093.7 5093.8 5093.8 5093.7 5093.8 5093.7 5093.9 5093.6 5093.9
4 6112.6 5999.5 6041.9 6035.6 6047.5 6035.4 6033.2 6040.0 6031.2 6033.7
6 6151.6 5951.3 6003.0 5995.8 6009.7 5993.0 5988.6 5992.3 5985.3 5988.2
1 105.9 106.3 106.3 106.3 106.3 106.3 106.3 106.3 106.2 106.3
4 144.4 136.7 138.2 138.3 138.7 137.1 138.2 138.0 138.2 139.0
6 176.7 137.5 139.0 139.3 139.4 138.8 139.4 139.2 138.6 139.6
Case
3
Avg.
Sector
tput
5%-ile
edge
UE tput
Interleaver Type
Case
1
Avg.
Sector
tput
5%-ile
edge
UE tput
0.975 0.98 0.985 0.99 0.995 1
0.01
0.02
0.03
0.04
0.05
CDF
C
C
H

F
E
R
Case 1, MAXCDM = 6, n = 2 CCH FER vs. Prob.


Block
App2, QPP
App3, QPP
App2, Subblock
App3, Subblock
App2, Random seq1
App3, Random seq1
App2, Random seq2
App3, Random seq2
App2, Random seq3
App3, Random seq3
CRRI
PBRI
Costas
CSPI
Fig. 2. PDCCH FER vs. Culmulative Probability for LTE Deployment Case
1 with n = 2 and at most 6 UEs Scheduled per subframe
One preference is for an interleaver scheme that reuses
existing hardware/rmware components without modication.
It turns out that the LTE subblock interleaver achieves some of
the best performance for throughput and PDCCH FER CDF.
Overall the proposed interleavers are similar in performance.
So it was proposed that the LTE subblock interleaver be chosen
for PDCCH REG randomization since it already exists in
the LTE specication and does not require any modication,
which was adopted in LTE (see Sec. 6.8.5 in [3]).
IV. POWER BOOSTING AND POWER LENDING
The distribution of CCE REGs across all control region in
a subframe also poses some design constraints. For one, in
case of CCE power boosting (which is widely used), REGs
in one OFDM symbol may be assigned much higher power
than those in another OFDM symbol, since the power boosted
REGs may be more concentrated in one OFDM symbol (it is
extremely difcult to obtain randomizing interleavers such that
all CCEs have approximately the same numbers of REGs in
each OFDM symbol). This is not desired since the transmitter
then needs to adjust its transmission power dynamically within
a wide range. An effective way to alleviate such a problem is
to limit the power boosting values to an appropriate range
which helps balance the powers among n symbols. It may
also occur that within one CCE, the 9 REGs have different
power boosting values. Though these power boosting values
are known to the BS, they are not transmitted to the UEs and
are unknown to UEs. Simply ignoring this intra-CCE power
offset during detection may lead to performance degradation
(and inaccurate results in case of system level simulations).
Therefore, it is desired to limit the intra-CCE power offset
and study how to deal with the unknown boosting values [8],
[9]. In addition, the PHICH requires additional power boosting
which may need to borrow power from PDCCH. The effect
of such power lending needs to be understood.
A. Handling unknown power boosting values
LTE UE receivers perform Maximal Ratio Combining
(MRC) in nominal situations. However, the convolutional
decoder for PDCCH does not have knowledge of the boosting
values used by the BS for each REG, and the decoder may
encounter codewords in which some subcarriers have different
power boosting than some other subcarriers. In practice the
MRC receiver has to assume equal gains for all REGs,
which means the actual receiver operation becomes neither
MRC nor EGC (Equal Gain Combining) in LTE. That is, the
UE receivers weighting factors account for only channel gain
variations but not the BS transmitters power boosting. This
needs to be handled properly in both link level simulations
and system level simulation, and the performance loss needs
to be studied. For simplicity, assume two measurements:
y
i
=
i
h
i
s + n
i
, where i = 1, 2,
i
is the unknown
power boosting value, h
i
is the channel gain assumed known
perfectly, s is the transmitted symbol, and n
i
is i.i.d. Gaussian
with zero mean and variance
2
. The optimal combining is
y :=

2
i=1
y
i
:=

2
i=1
h
i

y
i
(in the senses of ML, maximum
SINR, minimum distance, and MMSE), which leads to per-
formance degradation compared to

2
i=1

i
h
i

y
i
combining.
Derivations show that the combined Log-Likelihood Ratio
(LLR) is LLR =

2
i=1
LLR
i
where
LLR
i
=
2

2
i
Re( y
i
)[h
i
[
2

2
.
The proper scaling of LLR can then be accounted for for
each QPSK symbol in link level level simulations. The
effective SINR that determines the FER performance is
(

2
i=1

i
)
2
/4SINR
UE
in the AWGN case, that is, the SINR
assumed by the UE scaled by a value determined by
i
s. This
can be used to properly scale QPSK symbol SINR before
applying EESM in system level simulations. To verify the
SINR scaling formula, four link level simulations were carried
out, using 48 information bits and 288 codeword bits (144
subcarriers). The power boosting for the rst 72 subcarriers is
always 1.0, that is,
2
1
=1.0 in all four simulation. The power
boosting for the last 72 subcarriers is given as: 1)
2
2
= 1.0;
2)
2
2
= 1.5476, which should have 1 dB gain from 1); 3)

2
2
= 2.3039, which should have 2 dB gain from 1); 4)
2
2
= 3.3309, which should have 3 dB gain from 1). It can be
seen that the simulation results conrm that the convolutional
decoder is performing as expected.
-6 -5.5 -5 -4.5 -4 -3.5 -3 -2.5 -2 -1.5 -1
10
-6
10
-5
10
-4
10
-3
10
-2
10
-1
10
0
D
2
2
= 1
D
2
2
= 1.5476
D
2
2
= 2.3039
D
2
2
= 3.3309
SINR seen by UE (dB)
FER
Fig. 3. AWGN Channel Link Performance of Convolutional Decoder with
Different Unknown Intra-codeword Power Offset
Based on the above characterization, the link loss due
to unknown power boosting can be obtained analytically or
numerically. Numerical results in Table II show the increasing
link performance loss as the PDCCH CCE REG power offset
between symbol 1 and symbol 0 increases. Beyond about 4
dB the link loss starts to become signicant.
TABLE II
INTRA-CCE POWER OFFSET VS. CONVOLUTIONAL DECODER LINK LOSS
Power Offset K=7 Convolutional Coder
Sym1 vs Sym0 Link Loss
(dB) (dB)
1.90 0.06
3.62 0.17
5.22 0.35
B. Power lending to PHICH
In this subsection, power borrowing by PHICH from PD-
CCH on the rst OFDM symbol and network performance is
investigated. To compensate the power loss for PDCCH on
the rst symbol, PDCCH on the second (and third) OFDM
symbol(s) may use power higher than the case without power
borrowing, subject to symbol maximum power limitation (20
W for 5 MHz bandwidth).
System level simulations with 5MHz bandwidth and 2
control channel symbols were performed. Each symbol cannot
exceed the symbol maximum power limitation of 20 W. For
symbol 1, without any power lending, it was assumed that
8.8 W is used for reference symbols, PHICH, and PCFICH,
and 11.2 W may be used for PDCCH. Seven different power
borrowing levels were then studied: 0 dB, 0.5 dB, 1.0 dB,
1.25 dB, 1.5 dB, 2.0 dB, and 2.5 dB, corresponding to 0%,
12.2%, 25.8%, 33.3%, 41.2%, 58.5%, and 77.8% of the 11.2
W orginally allocated to PDCCH are borrowed by PHICH,
respectively. Then symbol 2 tries to compensate all the power
loss for each CCE individually, but also ensures the total power
does not exceed 20 W. Then the power boosting for each REG
is restricted to be within either +/-6 dB or +/-9 dB. The lower
boosting limit for symbol 2, however, is special and was set
differently to be -10 dB in all simulations; this is because if a
lower boosting limit of -6 dB was used then power exceeding
20 W may be possible. Finally, the intra-CCE power offset
limit was enforced for each CCE so that the REG powers
within each CCE do not vary beyond 0 dB, 2 dB, 3 dB, 4
dB, 6 dB, or 9 dB. Maximum of 6 DL assignments and 6 UL
assignments per subframe were used. The Subblock interleaver
for PDCCH was used.
The results in Tables III and IV show that there is no
noticeable loss in average cell and cell edge throughput perfor-
mance for Case 1 when the power borrowing is within 58.5%.
Higher power borrowing leads to some minor performance
loss (1% to 2%). Different intra-CCE power offset limits and
different power boosting limits have little or very minor impact
on throughput performance (< 1%), though at high power
borrowing cases such as 58.5% and 77.8% borrowing, higher
power boosting limit seems benecial. For Case 3, increasing
the power borrowing leads to only very minor degradation
of performance, up to 41.2% of power borrowing. Within this
range of power borrowing, varying the intra-CCE power offset
limit beyond 2 or 3dB and varying the power boosting limits
from +/-6 dB to +/-9 dB do not lead to meaningful changes in
either the average cell throughput or the cell edge throughput.
However, for 58.5% and 77.8% power borrowing, there is sig-
nicant throughput performance degradation. Allowing larger
intra-CCE power offset limit, which assigns more power to
symbol 2 to effectively compensate the power loss in symbol
1, prevents severe performance loss. Using +/-6 dB power
boosting limits has a better performance than using +/-9 dB
power boosting limits.
Therefore, relaxation of the intra-CCE power offset limit to
2 or 3 dB seems to be adequate for achieving non-borrowing
performance. Larger relaxation seems to have no meaningful
impact on performance except for very high power borrowing.
The histograms of the rst OFDM symbol and second OFDM
symbol PDCCH powers for Case 3, 41.2% power borrowing
and +/- 6 dB power offset, are shown in Fig. 4.
V. CONCLUSIONS
In this paper, we studied issues regarding the design and
analysis of the LTE PDCCH. Based on mathematical analysis
and/or simulations, conclusions have been reached, including
the adoption of the simple subblock interleaver to achieve
desired interference randomization property; power boosting
and power boosting limits; and a method of characterizing
TABLE III
THROUGHPUT PERFORMANCE (IN KBPS) OF PDCCH POWER BORROWING
WITH n = 2 AND +/-6 DB POWER BOOSTING LIMIT
0 6433.2 6352.5 6352.2 6353.5 6351.4 6353.0 6345.1 6297.6
2 6433.2 6352.2 6352.1 6351.6 6352.7 6348.8 6342.3 6292.5
3 6433.2 6352.5 6351.8 6351.9 6350.3 6349.6 6341.6 6287.9
4 6433.2 6350.5 6351.4 6350.5 6349.7 6348.3 6340.2 6285.3
6 6433.2 6350.5 6351.4 6350.7 6350.2 6350.2 6342.9 6293.5
0 229.0 202.2 202.4 202.3 202.2 201.6 201.2 198.3
2 229.0 202.0 201.2 201.1 203.0 202.0 202.7 198.3
3 229.0 201.1 202.4 201.2 202.2 202.2 200.6 197.7
4 229.0 201.7 201.1 202.2 201.8 201.4 201.5 198.0
6 229.0 201.7 201.1 201.9 202.5 201.0 201.4 198.0
0 6151.6 6039.4 6032.5 6010.5 5953.9 5911.4 5718.5 4924.0
2 6151.6 6042.1 6039.2 6029.3 6015.8 5989.1 5838.2 5205.5
3 6151.6 6042.2 6038.0 6027.6 6019.2 5999.3 5875.4 5318.6
4 6151.6 6041.9 6034.8 6025.4 6016.2 6003.3 5899.0 5408.3
6 6151.6 6041.9 6034.8 6026.3 6016.2 6003.2 5921.9 5567.5
0 176.7 152.5 151.1 150.6 149.6 146.3 141.4 101.6
2 176.7 151.4 150.6 151.6 150.7 150.0 145.5 114.2
3 176.7 151.4 151.3 152.2 150.0 149.5 146.4 119.3
4 176.7 150.6 152.4 150.9 151.4 150.4 147.0 124.0
6 176.7 150.6 151.2 151.6 151.3 151.0 146.7 132.2
Avg.
Sector
tput
5%-ile
edge
UE tput
Case
1
Case
3
77.8%
borrow
Avg.
Sector
tput
5%-ile
edge
UE tput
25.8%
borrow
33.3%
borrow
41.2%
borrow
58.5%
borrow
Max
Intra-
CCE
Power
Offset
Ideal
0.0%
borrow
12.2%
borrow
TABLE IV
THROUGHPUT PERFORMANCE (IN KBPS) OF PDCCH POWER BORROWING
WITH n = 2 AND +/-9 DB POWER BOOSTING LIMIT
0 6433.2 6352.5 6351.4 6351.9 6349.8 6349.5 6346.9 6328.4
2 6433.2 6352.7 6352.8 6351.9 6350.7 6350.3 6346.1 6329.1
3 6433.2 6352.6 6353.4 6352.5 6351.4 6349.2 6348.1 6331.6
4 6433.2 6352.6 6353.4 6351.0 6351.9 6349.4 6347.5 6331.2
6 6433.2 6352.6 6353.4 6350.6 6351.3 6350.0 6347.1 6330.6
9 6433.2 6352.6 6353.4 6350.6 6351.6 6349.3 6349.0 6330.8
0 229.0 202.2 203.0 201.9 202.7 202.1 201.7 201.0
2 229.0 201.6 202.3 201.4 202.6 202.2 202.9 200.7
3 229.0 201.3 202.1 201.4 201.9 201.6 201.8 200.8
4 229.0 201.3 202.0 202.2 201.6 201.7 200.8 201.9
6 229.0 201.3 202.0 202.0 201.5 201.5 201.9 201.3
9 229.0 201.3 202.0 202.0 201.9 201.8 202.0 201.3
0 6151.6 6039.4 6028.7 6003.9 5976.9 5927.6 5636.7 4276.1
2 6151.6 6038.8 6035.0 6025.0 6008.1 5978.7 5791.9 4720.1
3 6151.6 6038.8 6035.5 6023.8 6014.2 5992.5 5843.0 4907.6
4 6151.6 6038.8 6035.3 6025.5 6012.6 5996.3 5875.6 5060.8
6 6151.6 6038.8 6035.3 6025.1 6014.2 5998.9 5902.2 5286.9
9 6151.6 6038.8 6035.3 6025.1 6014.9 5997.3 5905.2 5427.1
0 176.7 152.5 151.6 150.4 150.3 148.5 138.6 86.7
2 176.7 151.3 151.2 150.9 151.1 150.9 144.3 102.2
3 176.7 151.3 151.2 151.5 150.7 150.8 146.4 111.5
4 176.7 151.3 150.5 151.6 150.6 151.4 147.2 117.3
6 176.7 151.3 150.5 149.3 150.7 151.4 147.1 124.8
9 176.7 151.3 150.5 149.3 150.7 150.2 144.7 129.1
Case
3
Case
1
Avg.
Sector
tput
5%-ile
edge
UE tput
77.8%
borrow
Avg.
Sector
tput
5%-ile
edge
UE tput
25.8%
borrow
33.3%
borrow
41.2%
borrow
58.5%
borrow
Max
Intra-
CCE
Power
Offset
Ideal
0.0%
borrow
12.2%
borrow
0
500
1000
1500
2000
2500
3000
2
.
2
5
2
.
5
2
.
7
5 3
3
.
2
5
3
.
5
3
.
7
5 4
4
.
2
5
4
.
5
4
.
7
5 5
5
.
2
5
5
.
5
5
.
7
5 6
6
.
2
5
6
.
5
6
.
7
5 7
7
.
2
5
7
.
5
S1 CCH power (Watts)
C
o
u
n
t
s

i
n

c
e
n
t
e
r

3

c
e
l
l
s
0
1000
2000
3000
4000
5000
6000
7000
8000
2 4 6 8 10 12 14 16 18 20
S2 CCH power (Watts)
C
o
u
n
t
s

i
n

c
e
n
t
e
r

3

c
e
l
l
s
Fig. 4. Histogram of two control symbol powers after PDCCH power lending
power boosting values unknown to the receiver in system
simulations effect of the unknown power boosting values in
link performance. These conclusions have been utilized to
develop the LTE standards or adopted in LTE standards. Fu-
ture research includes further optimization of PDCCH design
to support future development of LTE-Advanced technology
which offers even higher communication performance (such
as 1 Gbps DL and 500 Mbps UL peak data rates).
REFERENCES
[1] 3GPP TR 25.913, Requirements for Evolved UTRA (E-UTRA) and
Evolved UTRAN (E-UTRAN), v.7.3.0. 2006.
[2] R. Love, R. Kuchibhotla, A. Ghosh, R. Ratasuk, W. Xiao, B. Classon,
and Y. Blankenship. Downlink control channel design for 3GPP LTE.
Proc. IEEE WCNC, pages 813818, 2008.
[3] 3GPP TS 36.211, Evolved Universal Terrestrial Radio Access (E-UTRA);
Physical channels and modulation, v.8.3.0. May 2008.
[4] 3GPP TS 36.212, Evolved Universal Terrestrial Radio Access (E-UTRA);
Multiplexing and channel coding, v.8.3.0. May 2008.
[5] 3GPP TS 36.213, Evolved Universal Terrestrial Radio Access (E-
UTRA);Physical Layer Procedures , v.8.3.0. May 2008.
[6] Motorola. R1-074581, Interleaver Design for CCE to RE Group Mapping.
3GPP RAN1-51, 2007.
[7] Motorola. R1-080098, Control Symbol Power Limitation and REG
Interleaving. 3GPP RAN1-51bis, 2008.
[8] Motorola. R1-080750, PHICH Power Borrowing from PDCCH. 3GPP
RAN1-52, 2008.
[9] Motorola. R1-081131, PHICH and PDCCH Power Control requirement.
3GPP RAN1-52bis, 2008.
Note: 3GPP documents are available at ftp://ftp.3gpp.org
TABLE V
SIMULATION ASSUMPTIONS
Parameter Assumption/Value
Cellular layout Hexagonal grid, 19 sites with wraparound, 3 cells/site
Inter-site distance (ISD) 500 m (Case 1), 1732 m (Case 3)
Distance-dependent path loss L = 128.1 + 37.6log10(R), R in kilometers
Lognormal Shadowing As modeled in UMTS 30.03, B 1.4.1.4
Shadowing standard deviation 8 dB
Shadowing Correlation distance 50 m
Between sites 0.5 Shadowing
correlation Between cells per site 1.0
Penetration loss 20 dB
Carrier frequency 2 GHz
Bandwidth 5 MHz
Subcarrier spacing 15 kHz
Resource block size 180 kHz (12 subcarriers)
Cyclic Prefix overhead 7.1 % (short CP)
Subframe duration 1.0 ms
# of OFDM symbols per subframe 14
Channel model Typical Urban (TU) for PDSCH and PDCCH
UE deployment 10 per cell (uniform spatial distribution over cells)
HARQ scheme IR, Chase combining (asynchronous)
HARQ round trip delay 6 subframes (6 ms)
Max HARQ retransmissions 8
Thermal noise density -174 dBm/Hz
Antenna pattern (horizontal)
2
3
min 12 / ,
dB m
A R R R A



,
dB 3
T = 70 degrees, Am
= 20 dB (70 degree horizonal beamwidth)
Total BS TX power 43 dBm
BS antenna gain (incl. cable loss) 14 dBi
BS transmitter 1 antenna
UE speed 3 km/h
UE receiver 2 antennas
UE antenna gain 0 dBi
UE noise figure 9 dB
CQI feedback delay 2 ms
CQI subband size 180 kHz (12 subcarriers)
CQI quantization 5 bits per value/subband
CQI feedback cycle 2 ms
CQI Error 1dB for low SINR and 0.5 for high SINR
Link to system level interface K=7 Convolutional Coder (PDCCH), EESM (PDSCH)
Traffic type Full buffer
Scheduler Time and frequency selective Proportional Fair