Beruflich Dokumente
Kultur Dokumente
A1A
T
A1
(2)
The cell load threshold is set to be . Then we can judge the
load state of cells as follows:
less-loaded state if T
A1
(1 ),
overloaded state if T
A1
> (1 ).
(3)
In this paper, we suppose cell A
1
is overloaded and cell A
2
is less-loaded. Once the unequal distribution is detected by
cell A
1
, it will instruct the UE to scan neighboring target cell
A
2
and report the results in order to nd out the collection
of UEs that can handover to , which we dene as U
A1,A2
.If
the U
A1,A2
is not null, cell A
1
will initiate a HANDOVER
REQUEST to cell A
2
.
B. Coverage Adjustment
Suppose that the component carriers of lightly-loaded cell
A
2
is ranged from the lowest frequency to the highest fre-
quency
C
A2
= {f
1
, f
2
, ..., f
K
} (4)
The equal power of each CC is allocated at rst. Because
the channel fading of high frequency is larger than that of low
frequency, the rand-size relationship between the CCs is given
by:
h
f1
> h
f2
> ... > h
fK
(5)
Apparently, f
1
has the maximum coverage. And the order of
carrier coverage adjustment is from f
2
to f
K
The specic
policy of coverage adjustment is described as follows:
Step1: Initialize i=1, transmitting power and load of f
i
are
P
fi
A2
(0) and L
fi
A2
(0) respectively. Judge the load state of carrier
f
i
according to its current load L
fi
A2
(0) and carrier load thresh-
old L
th
. If L
fi
A2
(0) < L
th
, then dont need any adjustment
and return the HANDOVER REQUEST ACKNOWLEDGE to
cell A
1
. Otherwise, forward to step 2.
Step2: Let i=i+1 and dene the P
MAX
as the transmitting
power threshold. p is an appropriate increment of transmitting
power. Here are the two restrictions:
L
fi
A2
< L
th
(6)
p +P
fi
A2
P
MAX
(7)
Only when both of the two restrictions are met, we set the
transmitting power of carrier f
i
as:
P
fi
A2
= P
fi
A2
+p (8)
And go straight to step3. When (5) and (6) cant be met
simultaneously, we let i=i+1 again. If i K , go back to
step2 if not, skip to step5.
Step3: Target cell A
2
instructs UEs to measure the signal
interference noise ratio (SINR) on every CC after transmitting
adjustment. Then
fi
A2,
is obtained, which means the SINR
of UE on carrier f
i
in cell A
2
.
Step4: Transfer the UEs of cell A
2
which occupied the
resource of f
i
originally, to carrier f
j
when the SINR meets
the minimum QoS of UEs. The collection of UEs that can be
transferred from f
i
to f
j
in cell A
2
is:
U
(fi,fj)
A2
= {|
fi
A2,
>
th
, U
fi
A2
, f
i
C
A2
, f
j
C
A2
}
(9)
Step5: Cell b recalculates the load state on each carrier.
Y
fi
A2,
represents the number of resource block occupied by
UE in cell A
2
. Assume the UEs are transferred from f
i
to f
j
. Then the load on f
i
is represented by :
L
fi
A2
= L
fi
A2
(0)
U
(f
i
,f
j
)
A
2
Y
fi
A2,
(10)
And the load on f
j
is given by:
L
fj
A2
= L
fj
A2
(0) +
U
(f
i
,f
j
)
A
2
Y
fi
A2,
(11)
After the coverage adjustment, Cell A
2
returns a HAN-
DOVER REQUEST ACKNOWLEDGE to the cell A
1
which
includes the information of available resource for handover.
C. Handover Implementation
Based on the returned HANDOVER REQUEST AC-
KNOWLEDGE, cell A
1
can have the collection of available
resources of cell A
2
, which meet the condition of transferring.
A2,
= {f
i
|
fi
A2,
>
th
, f
i
C
A2
, U
A1,A2
} (12)
Then the cell A
1
instructs the UEs to handover to the cell A
2
,
and accordingly achieves the load balancing between cells.
IV. PERFORMANCE EVALUATION
In this section we describe our simulation environment and
performance metrics, and then analyze simulations results.
A. Simulation Environment
In order to verify the validity of DACC-LBS, we complete a
simulation of downlink based on LTE-Advanced system with
CA. The 3GPP case 1 extended
[6]
is adopted. There are four
component carriers to be aggregated, and f
1
= 860MHz,
f
2
= 2300MHz, f
3
= 3400MHz, f
4
= 4300MHz.The
specic parameters are showed in the Table 1.
TABLE I
SIMULATION PARAMETERS
Simulation Parameters Settings
cell Layout Hexagonal grid, 3 sectors per site
Inter-site distance 500 m
Number of cells 7
Shadowing standard deviation 8 dB
Thermal Noise Spectral Density 174 dbm/Hz
Penetration Loss 20 db
Component Carrier Bandwidth 20MHz
Total TX power 46 dBm
Antenna pattern 1*1
Trafc model Full buffer
B. Performance Metrics
We use two metrics to evaluate the performances of DACC-
LBS :
1) Balancing factor
Balancing Factor is commonly used as the indicator of
evaluating the performance of load balancing schemes
[7]
. The
success handover ratio can be also manifested by the value
of balancing factor. S
j
represents the throughput of sector j.
Then we have:
=
{
K
j=1
S
j
}
2
K
K
j=1
S
j
2
(13)
The range of the Balancing Factor is: 1/k 1. When
approaches 1, it indicates that the load distribution among cells
is extremely balanced. But when trends to 1/K, it suggests
that system is extremely non-equilibrium.
2) Average throughput per sector
System capacity is an important indicator that measures the
performance of load balancing schemes and also can reect
the utilization ratio of wireless resource. In this paper, the
average throughput per sector represents the system capacity.
Suppose that the throughput per UE is T
i
and the total number
of sectors is K. Then the average throughput per sector is:
s
averag
=
(
N
i=1
)
k
(14)
C. Simulation Results and Analysis
To verify the performance of DACC-LBS, we compare it
with TH-LBS. Fig. 3 illustrates the balancing factor varies with
the number of UEs per sector. The curves of the two schemes
are decreased when the number of UEs is less than 20. It
can be explained by that the trafc load is a little unbalanced
but the load state of cells are still less-loaded. In this case,
both TH-LBS and DACC-LBS are not triggered. With the
increasing number of the UEs, the uneven distribution of the
load among the cells is aggravating. Adopting TH-LBS can
balance the load among the cells to a certain extent, for the
reason that some UEs in overloaded cells can handover to the
less-loaded cells. But it will also cause the maximum coverage
carrier to be overloaded and cant share the load in case that
more UEs are in the system. Consequently, it maintains the
load balancing factor of the system around 0.86. While DACC-
LBS could stay around 0.95 for that it can utilize the idle
resource in the whole less-loaded cell to balance the trafc in
the system. Apparently, DACC-LBS improves impressive gain
in the load balancing factor by 11%.
15 20 25 30 35 40 45 50 55 60 65 70
0.5
0.6
0.7
0.8
0.9
1
Number of UEs Per Sector
B
a
l
a
n
c
i
n
g
F
a
c
t
o
r
DACCLBS
THLBS
Fig. 3. Comparison of Balancing Factor
15 20 25 30 35 40 45 50 55 60 65 70
2
2.5
3
3.5
4
4.5
5
5.5
x 10
7
Number of UEs Per Sector
T
h
r
o
u
g
h
p
u
t
P
e
r
S
e
c
t
o
r
/
b
p
s
DACCLBS
THLBS
Fig. 4. Comparison of average throughput per sector
Then, the system capacity is analyzed. Each UE with
2Mbps rate requirement is assumed. Fig. 4 shows the average
throughput per sector varies with the number of UEs from
15 to 70. From simulation results, DACC-LBS can improve
the average throughput by 20%, compared with TH-LBS.
When the trafc distributed unequally, TH-LBS can transfer
only a few UEs to less-loaded cell, which is limited by the
resource of maximum coverage carrier. But with the DACC-
LBS, the idle resources of all CCs are utilized to serve for the
UEs data demand. Therefore the utilization ratio of wireless
resource is improved by DACC-LBS. The problems can be
better understood with Fig. 5 which manifests the number of
successful transferred UEs of two load balancing schemes. The
more UEs are in the system, the more possible overloaded
15 20 25 30 35 40 45 50 55 60 65 70
0
5
10
15
20
Number of UEs Per Sector
N
u
m
b
e
r
o
f
S
u
c
c
e
s
s
H
a
n
d
o
v
e
r
U
E
s
DACCLBS
THLBS
Fig. 5. Comparison of number of transferred UEs
state the carrier with maximum coverage in less-loaded cell
would be. DACC-LBS can adjust the other CCs coverage to
share the load of maximum coverage carrier thus obtaining
signicant gain in number of successful transferred UEs by
120%.
V. CONCLUSION
A dynamic adjusting carrier coverage-based load balancing
scheme (DACC-LBS) is proposed in this paper for LTE-
Advanced system with CA. From simulation results, DACC-
LBS has shown signicant gains in load balancing factor
and system throughput, against TH-LBS. The results analysis
demonstrates that it is more efcient and feasible in the
scenario where the maximum coverage carrier in less-loaded
cell is overloaded. However the transmit power adjust will
bring about the increscent of same frequency interference
in neighbor cells. It need further study to investigate the
negative impact on the system performance and suppress the
interference.
REFERENCES
[1] Mikio Iwamura, Kamran Etemad, Mo-Han Fong, Ravi Nory and Robert
Love, Carrier Aggregation Framework in 3GPP LTE-AdvancedIEEE
Comunication Magazine, Vol. 48, Issue. 8, pp. 60-67, August 2010.
[2] Guangxiang Yuan, Xiang Zhang, Wenbo Wang and Yang Yang, Carrier
Aggregation for LTE-Advanced Mobile Communcation Systems2010
IEEE Vehicular Technology Conference, pp. 1-6, May 2010.
[3] 3GPP, TS 36.300 v9.3.0, Evolved Universal Terrestrial Radio Access
(E-UTRA); Stage 2 Description , Sept. 2009.
[4] 3GPP, TS 36.423 v10.0.0, 3rd Generation Partnership Project; Tech-
nical Specication Group Radio Access Network; Evolved Universal
Terrestrial Radio Access (E-UTRA); X2 application protocal (X2AP)
, Dec. 2010.
[5] 3GPP, TS 36.902 v9.3.1, 3rd Generation Partnership Project; Technical
Specication Group Radio Access Network; Evolved Universal Ter-
restrial Radio Access (E-UTRA); Self-conguring and self-optimizing
network (SON) use cases and solutions , Mar. 2011.
[6] 3GPP TSG , TR 36.814 v9.0.0, 3rd Generation Partnership Project;
Technical Specication Group Radio Access Network; Evolved Uni-
versal Terrestrial Radio Access (E-UTRA); Further advancements for
E-UTRA physical layer aspects , Apr. 2010.
[7] Lei Zhang, Fei Liu, Lin Huang and Wenbo Wang, Trafc Load Balance
Methods in the LTE-Advanced System with Carrier Aggregation2010
International Conference on Communications, Circuits and Systems
(ICCAS) , pp. 63-67, Sep. 2010.