daigai

Well-Known Member
Link tải luận văn miễn phí cho ae Kết Nối

Abstract—The explosive growth of wireless multimedia services
skyrockets the demand for system capacity of the cellular communication network. The heterogeneous network (Hetnet) has
been widely accepted as an effective wireless networking method
to improve system capacity. However, multi-tier networking
architecture and the great number of small cells coexisting with
macrocell make the interference management and spectral resources management more complex. The corresponding problems
thus draw great attentions in 3GPP standardization process. In
this paper, we introduce the cloud radio access network (CloudRAN), which is seen as a low cost delivery method of cellular
networks, into spectral resources management and interference
management of Hetnet. To achieve effective spectral resources
management and interference management in this architecture,
the following contributions are further made: 1) the spectral
resources allocation optimization problem of Hetnet is constructed, comprehensively considering the mutual interference,
spectral efficiency and quality of service (QoS); 2) to solve the
optimization problem, we propose the probability weighted based
spectral resources allocation algorithm; 3) to analyse the influence
of the proposed algorithm to the infrastructure network (e.g.,
macrocell), the closed-form outage probability of macrocell users
is derived. Finally, the comparison results show that the proposed
algorithm has higher frequency reuse efficiency and lower outage
probability of macrocell users.
Index Terms—Hetnet; Cloud-RAN; interference mitigation;
spectral resources allocation
I. INTRODUCTION
The heterogeneous network (Hetnet), which consists of
small cells (such as micro-, pico- and femto- cells) overlapped
with the macrocell, is considered as an effective solution to
handle explosive wireless traffic, especially the successively
increasing traffic demands in indoor environments [1]. In
essence, creation of small cells within the macrocell shortens
the communication links, especially for indoor users, which in
turn improves network frequency reuse efficiency spatially in
the Hetnet. As the spectral resources are in sparse for cellular
networks, such a frequency reuse efficiency improvement can
directly translate into capacity enhancement of the system.
Therefore, Hetnet gains significant interests in academic and
industrial areas (e.g., the 3GPP standardization [2]).
However, new forms of interference, e.g., mutual interference between small cell access nodes with macrocell users,
arise in Hetnet architecture paradigm, which mandates more
effective spectral resources allocation strategies to mitigate
these interference and improve frequency reuse efficiency.
Therefore, many studies have been done in the prior art.
Spectral resources allocation and interference mitigation are
mainly studied in two-tier networks (e.g., femtocell and macrocell coexistence scenario, relay and macrocell coexistence
scenario, etc.) [3]–[7]. For example, in the scenario of twotier femtocell networks, the authors in [3] suggested to assign
different spectral resources to macrocell and femtocell networks. Considering the scarcity of radio resources, orthogonal
division of spectral resources is obviously non-preferable.
In [4], the author proposed a cognitive-based interference
management solution for LTE-Advanced (LTE-A) femtocells
by sharing measured pathloss information among neighbors
and selecting component carriers according to the estimated
mutual interference. However, this method does not consider
the cross-tier interference between macrocell and small cells.
An interference avoiding algorithm which based on backhaul or air interface to acquire macrocell users’ uplink and
downlink scheduling information is provided in [5]. But this
algorithm will greatly increase the system overhead. In [6], the
authors proposed a dynamic frequency planning algorithm. It
regarded all small cell users as “macrocell users” and shared
spectral resources with the real macrocell user. Obviously,
when there are many small cell users, it will severely reduce
the capacity of the macrocell. However, as the two-tier network
scenario shifting to multi-tier Hetnet scenario, the complexity
of existing resources allocation and interference management
would increase exponentially, which is not well considered in
the prior art.
Cloud-RAN is a prevailing architecture for future cellular
network delivery [8], where massively radio remote units of small cells connect to base band processing and resources pool,
c.f., Figure 1. It facilitates coordinated resources allocation
as well as ubiquitous network coverage in a HetNet setting.
This unique networking paradigm harnesses the advantages of
The First IEEE ICCC International Workshop on Internet of Things (IOT 2013)
978-1-4799-1403-6/13/$31.00 ©2013 IEEE 88distributed networking architecture as well as the benefits of
having a centralized decision making capability.
Therefore, we combine Hetnet and Cloud-RAN, and propose a probability weighted based spectral resources allocation
algorithm in Hetnet under Cloud-RAN architecture. Concretely, the following contributions are made: 1) the spectral
resources allocation optimization problem of Hetnet is constructed, comprehensively considering the mutual interference,
spectral efficiency and quality of service (QoS); 2) to solve the
optimization problem, we propose the probability weighted
based spectral resources allocation algorithm; 3) to analyse
the influence of the proposed algorithm to the infrastructure
network (e.g., macrocell), the closed-form outage probability
of macrocell users is derived.
The remainder of the paper is organized as follows. The
system model is presented in Section II. Section III presents
problem formulation and the proposed algorithm. The outage probability of macrocell users are analyzed in Section
IV. Section V provides simulation settings and performance
evaluations. Finally, Section VI concludes the paper.
II. SYSTEM MODEL
Consider a Hetnet scenario under Cloud-RAN architecture,
which consists of a reference macrocell, small cells and
Cloud-RAN center, as depicted in Figure 1. Compared with
traditional RAN, Cloud-RAN reduces energy consumption and
improves the frequency reuse efficiency. It’s a natural evolution
of the distributed base transceiver station, which is composed
of the baseband unit (BBU) and remote radio unit (RRU).
When Cloud-RAN is applied to Hetnet scenario, there are
three main parts: RRU plus antennas are located at the remote
sites, the Cloud-RAN center composed of spectral resources
management unit and allocation unit, the high bandwidth lowlatency optical transport network connects the Cloud-RAN
center and the remote sites. In this scenario, the set of users
in the macrocell is defined as KM. Every small cell consists
of a small cell base station (SBS) and a small cell user (SUE).
The set of small cells is defined as KL.
Fig. 1. Hetnet under Cloud-RAN Architecture
III. PROBLEM FORMULATION AND PROPOSED ALGORITHM
A. Spectral resources partition
When users locate in the coverage of their non-anchored
base stations (BSs), i.e. macrocell users (MUEs) located near
active SBSs, or SUEs located near macrocell base station (MBS), they may experience significant cross-tier interferences.
To avoid this, the total spectral resources of the system can
be partitioned into two parts, according to [11] [12]. One
part is the shared spectral resources and the other is the
dedicated one. All the SBSs are classified as “noninterfering
ones” denoted by set KS and “interfering ones” denoted by set
KP (KL = KS ∪ KP ). Therefore, the cross-tier interference
can be avoided by allocating shared spectral resources to set
KS and MBS while dedicated spectral resources is exclusively
allocated to set KP.
Figure 2 shows the partition of spectral resources. B denotes
total spectral resources of the system, where B = BS + BP.
BS denotes the bandwidth of shared spectral resources and BP
means dedicated ones. It is assumed that vS and vP are the ratios of the shared and dedicated spectral resources accounting
for the proportion of the total system bandwidth, respectively.
Thus, we have BS = vSB and BP = vPB = (1−vS)B. Phần mềm is
the transmission power of the MBS. Pi means the transmission
power of the SBS i (i ∈ KL).
Fig. 2. Partition map of spectral resources in Hetnet under Cloud-RAN
Architecture
B. Optimization problem formulation based on spectral resources partition
The spectral resources partitioning problem is formulated
as a maximization problem of spectral efficiency which is
described as:
η =
1 B
(
|KM|
m
=1
log2Rm +
|KS|
i
=1
log2RS,i +
|KP |
j
=1
log2RP,j), (1)
| · | denotes the cardinality of the set. Rm represents the link
rate from MBS to MUE m. RS,i represents the link rate from
SBS i (i ∈ KS) to its user. RP,j represents the link rate from
SBS j (j ∈ KP) to its user. These link rates can be calculated
as
Rm = BSlog2 (1 + γm) , (2)
RS,i = BSlog2 (1 + γS,i) , (3)
RP,j = BPlog2 (1 + γP,j) , (4)
The First IEEE ICCC International Workshop on Internet of Things (IOT 2013)
89
Ket-noi.com kho tai lieu mien phi Ket-noi.com kho tai lieu mien phiwhere γm is the received signal to interference ratio (SIR)
of MUE. γS,i is the received SIR of SUE served by SBS i
(i ∈ KS). γP,j is the received SIR of SUE served by SBS j
(j ∈ KP ).
The optimal spectral resources allocation ratio can be obtained by solving
max
vS
η
s.t. 0 ≤ vS ≤ 1, γm > γmq , γS,i > γq, γP,j > γq ,
(5)
where γmq denotes the target SIR requirement of MUE. γq
denotes the target SIR requirement of SUE.
In order to find the optimal vS, firstly, the optimal solution
of (5) is obtained without considering the constrains. According to KKT conditions [10], the optimal ratio of the shared
spectral resources, i.e., vS∗ can be derived as
ν∗
s =
1
|KP |
|KM |+|KS| + 1
. (6)
From (6), it can be observed that for a fixed number
of MUEs as the cardinality of set KP decreases and the
cardinality of set KS increases, the optimal ratio of the
shared spectral resources increases, or say, the frequency reuse
efficiency increases.
Secondly, set KP and set KS are determined considering the
SIR requirement of MUEs and SUEs. The probability weighted based spectral resources allocation algorithm is proposed
in the next subsection.
C. Probability Weighted Based Spectral resources allocation
algorithm
The core idea of the proposed algorithm is to minimize the
cardinality of set KP under the condition of satisfying the SIR
requirements of SUEs and MUEs.
In the solution, Cloud-RAN center determines whether
small cells can use shared spectral resources or dedicated
spectral resources in a probability manner. The main procedure
of the algorithm is presented as follows:
1 SUE l (l ∈ KL) receives the pilot signal from MBS,
and then forward to Cloud-RAN center through fiber. Spectral
resources management unit in Cloud-RAN center judges if
SUE l can achieves the SIR requirement γq. If γl > γq, SBS
l is allocated to set F1, otherwise F2. γl is the received SIR
of SUE l.
2 SBS l (l ∈ F1) receives pilot signal from MUE, and
then forward to Cloud-RAN center through fiber. According
to the symmetric channel property, spectral resources management unit in Cloud-RAN center calculates the biggest crosstier interference from SBS l to MUE, denoted as Il .
3 Spectral resources management unit in Cloud-RAN
center measures the number of SBSs in F1, denoted as nF , i.e.
nF = |F1|. Then, calculates the probability of using shared
spectral resources of SBS l (l ∈ F1) as
pS(l) = min 1, nSFmqIl , (7)
where Sq
m denotes the limited interference under a certain SIR
requirement of MUE.
4 Spectral resources management unit calculates the cardinality of KS as |KS| = [l ∈F1 ps(l)]. Sort SBSs in
descending order according to their probability values pS(l).
Then, the top |KS| small cells are allocated to set KS, which
share spectral resources with macrocell. The small cells in set
KL − KS are allocated to set KP , which use the dedicated
spectral resources.
The algorithm mentioned above is on the condition that
there is one single MUE in the macrocell network. When there
are |KM| MUEs, the following steps should be taken.
1’ MUE m (m=1,2,...,|KM|) executes 1 , 2 , 3 , 4 as
mentioned above.
2’ Cloud-RAN center establishes KSm (m=1,2,...,|KM|)
to record the results. KS is determined by KS =
k|k ∈ KS1 ∩ KS2 ∩ ... ∩ KS|KM |, k ∈ KL.
3’ The rest SBSs are allocated to set KP .
IV. OUTAGE PROBABILITY ANALYSIS
The outage occurs when the total interference experienced
by MUE Sm is larger than the limited interference Smq required
by MUE. The outage probability, i.e., the failure probability
of transmission due to the cross-tier interference, can be
calculated as
po(pS, γmq ) = Pr(Sm ≥ Smq ). (8)
It is assumed that the distribution of the interference experienced by MUE is independent and identical. According
to the central limit, the distribution of the total interference
experienced by MUE Sm is approximately normal with the
mean μ and variance σ2, i.e., N(μ, σ2). Using Q-function,
the outage probability of MUE can be further derived as
po(pS, γmq ) = Q(Smq − μ
σ
). (9)
As mentioned above, if cross-tier interference to MUE
caused by SBS is smaller than SnmqF , corresponding SBS is
allocated to use shared spectral resources with the probability
pS(l) = 1. Otherwise, the SBS is assigned to use shared
spectral resources with the probability pS(l) = nSFmqI
i
as
defined in (7). Hence, the cross-tier interference experienced
by MUE can be calculated as
Sm(pS, γmq ) =
l ∈F1
pS(l)Il =
u∈ϕ
pS(u)Iu +
v∈φ
pS(v)Iv,
(10)
where ϕ denotes the set of SBSs which use shared spectral
resources with probability pS(u) = 1 and φ denotes the set
of SBSs which use shared spectral resources with probability
pS(v) = nSFmqIv . Iu and Iv denote the interferences generated
by SBS u (u ∈ ϕ) and SBS v (v ∈ φ), respectively.
Denoting nϕ = |ϕ| and nφ = |φ| , i.e., nF = nϕ + nφ.
After replacing pS(u) and pS(v) in (10) by pS(u) = 1 and
pS(v) = nSFmqIv , Sm(pS, γmq ) =
u∈ϕ
Iu
+ nφ SnmqF . It is assumed
that the distribution of Iu is independent and approximately
The First IEEE ICCC International Workshop on Internet of Things (IOT 2013)
90normal with the mean value ξ and variance value ψ2. Then,
the distribution of Sm is given by
Sm(pS, γmq ) → N(nϕξ + nφ Smq
nF
, nϕψ2). (11)
Therefore, the outage probability of MUE is obtained by
po(pS, γmq ) = Q(
Sq
m − nϕξ − nφ Snmq
F
√nϕψ ). (12)
V. PERFORMANCE EVALUTION
System simulation parameters obtained from [9] are listed in
Table 1. In the simulation, the transmission range of macrocell
is Rm. 10 MUEs are randomly placed within the macrocell
coverage area. Small cells are randomly placed and each
small cell has the transmission range Rs. Every small cell
contained one active SUE. The performance of the probability weighted based spectral resources allocation algorithm
(PWBA) is compared with the Joint Frequency Bandwidth
Dynamic Division, Clustering and Power Control Algorithm
(JFCPA) [11], the Hybrid Spectrum Usage Algorithm (HSUA)
[12] and the Overall Fairness and Efficient Algorithm (OFEA)
[13].
TABLE I
SIMULATION PARAMETERS
Simulation Parameter Value
transmission range of MBS and SBS (Rm and Rs) 1000m, 50m
Total transmit power of MBS and SBS 43 dBm, 20 dBm
Number of small cells (KL) 1000
Active SUEs per small cell 1
Active MUEs (KM) 10
Penetration Loss 10 dB
Shadowing standard deviation of macrocell/small cell 8 dB, 10 dB
Auto-correlated distance 50 m
System bandwidth 10 MHz
In JFCPA, frequency bands are divided according to the
density and the location of small cells. In HSUA, small
cells are distinguished to be inner and outer small cells
according to the received pilot power of MBS, i.e., Reference
Signal Received Power (RSRP). Inner small cells operate in
partitioned spectrum usage, and the outer small cells operate
in shared spectrum usage. In OFEA, macro-interfering small
cells are forced to use the dedicated sub-channels. The noninterfering small cells, on the other hand, will use the shared
sub-channels which are also used by the macrocell.
Figure 3 shows the average number of SBSs E[|KS|] which
use the shared spectral resources when the SIR requirement
of MUE changes from 0 dB to 10 dB. The largest E[|KS|]
comes from the proposed algorithm, which is much bigger
than that of other three algorithms. Because in the proposed
algorithm, only small cells that generate severe interference to
the reference MUE are assigned to use the dedicated spectral
resources, more small cells are able to share spectral resources
with MBS, or say, for the cardinality of set KS is bigger. It
indicates that the frequency reuse efficiency of the proposed
algorithm is much higher than other three algorithms.
Figure 4 shows the cumulative distribution function (CDF)
of |KS| as γmq = 0 dB. The distribution of |KS| in the
proposed algorithm is nearly the same as that in the JFCPA,
but |KS| of the proposed algorithm is bigger than other three
algorithms, which can also proves that the proposed algorithm
obtains higher frequency reuse efficiency.
Figure 5 shows the outage probability of the reference MUE
in the four algorithms when the SIR requirement of MUE
changes from 0 dB to 10 dB. Compared with other algorithms,
there is no outage in the proposed algorithm. According to
(10) small cells which share spectral resources with macrocell
are divided into two groups. ϕ for the set of SBSs which
use shared spectral resources with probability pS(u) = 1, and
φ for the set of SBSs which use shared spectral resources
with probability pS(v) = nSFmqIv . The total interference that
MUE experienced can be written as Sm(pS, γmq ) =
u∈ϕ
Iu
+
n
φ
Sq
m
n
F
< n
ϕ
Sq
m
nF
+ nφ SnmqF = Smq , so there is no outage in
the proposed algorithm. Therefore, the proposed algorithm can
effectively reduces the probability of transmission failure of
MUE.
Figure 6 shows the comparison results of the spectral
resources partitioning ratio in the four algorithms when the
SIR requirement of MUE changes from 0 dB to 10 dB. It
can be obtained that the proposed algorithm can effectively
improves the frequency reuse efficiency. This is due to the fact
that in the proposed algorithm, small cells use shared spectral
resources in a probability manner which maximize the number
of small cells that sharing spectral resources with macrocell
while satisfying the SIR requirements of MUEs and SUEs.
Fig. 3. Average number of |KS| v.s. SIR requirement of MUE
VI. CONCLUSION
In this paper, the spectral resources allocation optimization
problem of Hetnet under Cloud-RAN architecture is constructed. To solve the optimization problem, the probability weightThe First IEEE ICCC International Workshop on Internet of Things (IOT 2013)
91
Ket-noi.com kho tai lieu mien phi Ket-noi.com kho tai lieu mien phiFig. 4. CDF of |Ks| when the SIR requirement of MUE is 0 dB.
Fig. 5. Outage probability of MUE v.s. SIR requirement of MUE
Fig. 6. Partition ratio of spectral resources v.s. SIR requirement of MUE
ed based spectral resources allocation algorithm is proposed,
in which both SIR requirements of MUEs and SUEs are taken
into account. The closed-form outage probability of MUEs is
derived mathematically. Simulation results confirmed that the
proposed algorithm lowers outage probability of MUEs and
improves frequency reuse efficiency.
ACKNOWLEDGMENT
This work was supported by the National Science and Technology Special Project (No. 2012ZX03003010-004) and National Science Foundation under grant numbers No. 1145596
and No. 0830493.
REFERENCES
[1] A. Barbieri, A. Damnjanovic and T. Ji et al., “LTE Femtocells: System
Design and Performance Analysis,” in Proc. IEEE VTC Spring, May
2011, pp. 1-5.
[2] 3GPP TS 22.220, “Service requirements for Home Node B (HNB) and
Home eNode B (HeNB),” v9.5.0.
[3] V. Chandrasekhar and J. Andrews, “Spectrum allocation in tiered cellular
networks,” IEEE Trans. Communications, Oct. 2009, vol. 57, no. 10,
pp. 3059-3068.
[4] L. Zhang, L. Yang, and T. Yang, “Cognitive Interference Management for
LTE-A Femtocells with Distributed Carrier Selection,” in Proc. IEEE
VTC Fall, Sep. 2010, pp. 1-5.
[5] M. Sahin, I. Guvenc and M. Jeong et al., “Handling CCI and ICI in
OFDMA femtocell networks through frequency scheduling,” IEEE Trans.
Consumer Electronics, Nov. 2009, vol. 55, no. 4, pp. 1936-1944.
[6] D. Lopez-Perez , G. de la Roche and A. Valcarce et al., “Interference
avoidance and dynamic frequency planning for wimax femtocells networks,” in Proc. IEEE Singapore International Conference, Communication Systems, Nov. 2008, pp. 1579-1584.
[7] Yi. W, D. Zhang and H. Jiang et al., “A Novel Spectrum Arrangement
Scheme For Femtocell Deployment In LTE Macrocells,” in Proc. IEEE
20th International Symposium on Personal, Indoor and Mobile Radio
Communications, Sep. 2009, pp. 6-11.
[8] Z. Ghebretensae, K. Laraqui and S.Dahlfort et al., “Transmission solutions and architectures for heterogeneous networks built as C-RANs,” in
Proc. 2012 7th International ICST Conference on Communications and
Networking in China, Aug. 2012, pp.748-752.
[9] 3GPP TR 25.967, “Home node B radio frequency (RF) requirements
(FDD),” v9.0.0.
[10] D. Bertsekas, “Convex Optimization Theory,” 1nd ed. Belmont,
Massachusetts: Athena Scientific.
[11] H. Li, X. Xu and D. Hu et al., “Graph Method Based Clustering
Strategy for Femtocell Interference Management and Spectrum Efficiency
Improvement,” Journal of Communications and Networks, Dec. 2011,
vol. 13, no. 6, pp. 664-677.
[12] Y. Bai, J. Zhou and L. Chen, “Hybrid Spectrum Usage for Overlaying
LTE Macrocell and Femtocell,” in Proc. IEEE Global Telecommunications Conference, Dec. 2009, pp. 1-6.
[13] J. Lu , W. Zheng, and T. Su et al., “Interference mitigation spectrum
allocation for energy efficient OFDMA femtocell networks,” in Proc.
International Conference on ICCSNT, Dec. 2012, pp. 252-256.
Link Download bản DOC
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:

 

Các chủ đề có liên quan khác

Top