May, 2014 – Slide 1 Communication & Network Lab
Resource Allocation in Heterogeneous
Networks
Master’s Thesis Presentation
May 23rd 2014
By Trung Kien Vu
Advisor: Prof. Sungoh Kwon
Committee: Prof. Chong-Koo An, (chair)
Prof. Sungoh Kwon,
Prof. Sunghwan Kim.
May, 2014 – Slide 2 Communication & Network Lab
Research Interests:
- Interference management and resource allocation: Heterogeneous
and Small Cell Networks.
 Trung Kien Vu and Sungoh Kwon, “eICIC-based Interference Mitigation in Small Cell
Networks”. [In preparation].
- Routing protocols: Mobile Ad-hoc Networks and Wireless Sensor
Networks.
 Trung Kien Vu and Sungoh Kwon, “Mobility-Assisted On-Demand Routing Algorithm
for MANETs in the Presence of Location Errors”, The Scientific World Journal, vol.
2014, Article ID 790103, 11 pages, 2014.
May, 2014 – Slide 3 Communication & Network Lab
Outline:
1. Introduction
2. Problem and Contributions
3. System Model
4. Proposed Algorithms
5. Simulation Results
6. Conclusions
May, 2014 – Slide 4 Communication & Network Lab
Introduction:
 The demand for mobile data traffic
 Solution: Heterogeneous Networks (HetNets)
May, 2014 – Slide 5 Communication & Network Lab
Introduction:
 What is HetNet?
 Consisting of multiple types of access nodes such as macro cells
and smallcells (picocells and femtocells).
 Smallcells are deployed under the coverage of macrocell.
 Smallcells provide the indoor and outdoor wireless services by
extending the network coverage and increasing the network
capacity.
Node types Transmission Power Coverage
Macrocells 43-46 dBm Few Km
Picocells 23-30 dBm ≤ 300 m
Femtocells ≤ 23 dBm ≤ 50m
May, 2014 – Slide 6 Communication & Network Lab
Problem:
 HetNet consists of Macrocells and Femtocells.
 Macrocells and Femtocells share the same radio frequency
 It causes the cross-layer interference between Macrocells and
Femtocells
Macrocells
Picocells Femtocells
May, 2014 – Slide 7 Communication & Network Lab
Problem: Example
 Macro base station: MeNB
 Femto base station: HeNB
 Macro User: MUE
 Femto User: HUE
Figure 2: System
May, 2014 – Slide 8 Communication & Network Lab
Previous work:
 Enhanced Inter-Cell Interference Coordination eICIC is introduced
to address cross-layer interference between macro and femtocells.
 Almost Blank Subframe ABSF is one of eICIC techniques in which
the interfering cell will stop using some subframes in order to
reduce the ICI.
No Transmission Transmission
Figure 3: ABSF subframes
May, 2014 – Slide 9 Communication & Network Lab
Previous work:
 Previous work use fixed ABSF pattern and all HeNBs are globally
set to same ABSF based on number of users.
 There is no coordination mechanism between femtocell base
stations HeNBs.
 How many and which subframes should be muted ?
May, 2014 – Slide 10 Communication & Network Lab
Contributions:
 In this paper, our contributions include:
 Dynamically optimal ABSF Selection Algorithm for each
HeNB based on the Quality of Service of macro users.
 Interference HeNB Coalition Algorithm to reduce the mutual
interference.
May, 2014 – Slide 11 Communication & Network Lab
System Model: Objective
 Find the optimal muted rate 𝛼 𝑚 for each MUE m.
(defined as number of muted subframes (ABSF) over number of all subframes)
 That satisfies the Signal-to-Interference-and-Noise Ratio (SINR)
𝛾 𝑚 of MUE m.
Minimize 𝛼 𝑚
Subject to 𝛾 𝑚≽ 𝛾0 , m  ℳ
𝛾0 : SINR threshold
ℳ : Set of MUEs
May, 2014 – Slide 12 Communication & Network Lab
System Model:
 The SINR 𝛾 𝑚 at link ℒ 𝑚 between the MeNB and the MUE m is
calculated as
P(m) and G(M,m) : the transmission power of MeNB and path gain
between MeNB and MUE.
𝑃(𝐹𝑓 and 𝐺(𝐹𝑓, 𝑚 : the transmission power of HeNB and path
gain between HeNB and MUE. ℱ is a set of HeNB.
𝜎 𝑚 : the thermal noise at macro user m.
( , ) ( )
m
m
G M m P m



The received power from MeNB
Total interference and noise
( ) ( , )m f f f m
P F G F m 
 F
May, 2014 – Slide 13 Communication & Network Lab
System Model:
 ℒ : the set of links from the MeNB to their serving MUEs,
ℒ = (ℒ1, . . ; ℒℳ).
 The constraint 𝛾 𝑚≽ 𝛾0 can be transformed in matrix form as
 𝐅 𝑚, 𝑓 =
𝐺 𝐹 𝑓,𝑚 𝛾0
𝐺(𝑀,𝑚
, 𝐏ℱ = (P(1), … , P(𝐹𝑓 )) 𝑇
 𝐏ℳ = (P(1), … , P(ℳ)) 𝑇
,
 𝒃 = (b(1), … , b(ℳ)) 𝑇
such that 𝑏(𝑚 =
𝛾0 𝜎 𝑚
𝐺(𝑀,𝑚
 FP P bF M
May, 2014 – Slide 14 Communication & Network Lab
Proposed Algorithm – ABSF Selection:
 When the HeNBs stop transmission on some subframes, the (SINR)
𝛾 𝑚 at link ℒ 𝑚 can be rewritten as
 The received power from HeNbs is reduced in order to increase the
SINR of MUE m.
( , ) ( )
( ) ( , )(1 )
m
f f f m m
G M m P m
P F G F m

 

  F
Reduced interference rate
May, 2014 – Slide 15 Communication & Network Lab
Proposed Algorithm – ABSF Selection:
 Now, our objective can be transformed as
Minimize 𝛼 𝑚
Subject to 𝐀𝛼 𝑚 ≽ 𝐁
 where
 A unique solution to this problem is
 Now, we already get the optimal muted rate for MUE
,
.

  
A P
B P P b
F
F M
F
1
( ) .T T
m
 
 A AA B
May, 2014 – Slide 16 Communication & Network Lab
Proposed Algorithm – Interfering HeNB Coalition
 To group the mutual interfering HeNB to cooperate in
ABSF mode efficiently.
Figure 4: Coalition Example
May, 2014 – Slide 17 Communication & Network Lab
Proposed Algorithm – Interfering HeNB Coalition
 Including 2 mechanisms:
 Mechanism 1: to find the victim MUEs affected by
each HeNB.
 Mechanism 2: to group HeNBs having same Victim
MUE.
May, 2014 – Slide 18 Communication & Network Lab
Proposed Algorithm – Interfering HeNB Coalition
 Mechanism 1: to find the victim MUEs affected by
each HeNB.
 Detect the Victim MUEs
 Report interfering HeNB’s list to MeNB.
 Do set intersection algorithm by MeNB having the
same HeNBs.
 Send Victim MUE’s list to HeNB
May, 2014 – Slide 19 Communication & Network Lab
Proposed Algorithm – Interfering HeNB Coalition
 Mechanism 2: to group HeNBs having same Victim
MUE.
 Exchange the VMUE’s list to neighbor HeNBs
 Do set intersection algorithm having the same VMUE.
 Set the muted rate for HeNB.
 Active the ABSF mode.
May, 2014 – Slide 20 Communication & Network Lab
Simulation Results
 Simulation Parameters
Parameter Values
System bandwidth 10 MHz
Channel Model Urban Macro-Femto Scenario Model
MeNB Tx 46 dBm
Number of MUEs 20-100
HeNB Tx 23 dBm
Number of HeNBs 40-400
Number of MUE 40-400
Thermal Noise -174 dBm/Hz
Noise Figure 9 dB
Simulation Run Times 1000
May, 2014 – Slide 21 Communication & Network Lab
Simulation Results
 Algorithm Notations
Names Notice
Non ABSF Without eICIC
Optimal ABSF eICIC with optimal muted rate
Fixed ABSF - I eICIC with muted rate: 1/10
Fixed ABSF - II eICIC with muted rate: 2/10
Fixed ABSF – III eICIC with muted rate: 3/10
Each step Distance between HeNB and MUE is
gradually increased in order to reduce
interference
May, 2014 – Slide 22 Communication & Network Lab
Simulation Results
Figure 5: The required muted rate
0 5 10 15 20 25 30 35 40 45 50
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
Step
AvgMutedRateRequiredforallVMUEs
Optimal Muted Rate
May, 2014 – Slide 23 Communication & Network Lab
Simulation Results
Macro users Throughput Femto users Throughput
Figure 6: The user throughput –
performance balance between Macro and Femto users
0 5 10 15 20 25 30 35 40 45 50
1300
1350
1400
1450
1500
1550
1600
Step
AvgThroughputofFemtocellUsers[Kbps] Optimal ABSF
Non ABSF
Fixed ABSF - I
Fixed ABSF - II
Fixed ABSF - III
0 5 10 15 20 25 30 35 40 45 50
400
500
600
700
800
900
1000
Step
AvgThroughputofMacrocellUsers[Kbps]
Optimal ABSF
Non ABSF
Fixed ABSF - I
Fixed ABSF - II
Fixed ABSF - III
May, 2014 – Slide 24 Communication & Network Lab
Simulation Results
0 5 10 15 20 25 30 35 40 45 50
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
Step
OutageProbabilityofMacroUsers
Non ABSF
Fixed ABSF - I
Fixed ABSF - II
Fixed ABSF - III
Optimal ABSF
Figure 7: Outage Probability.
# 𝑜𝑓 𝑈𝑛𝑠𝑎𝑡𝑖𝑠𝑓𝑖𝑒𝑑 𝑈𝑠𝑒𝑟𝑠
# 𝑜𝑓 𝑡𝑜𝑡𝑎𝑙 𝑈𝑠𝑒𝑟𝑠
May, 2014 – Slide 25 Communication & Network Lab
Conclusions
 To address the cross-layer interference between macro and femto
cell layers by
 Propose a dynamically optimal ABSF eICIC framework based
on the quality of service of macro users.
 Group interfering HeNBs in ABSF mode to reduce mutual
interference among HeNBs
 The simulation results show that our algorithms outperform
previous work and bring good solutions for smallcell networks.
May, 2014 – Slide 26 Communication & Network Lab
Thank you for your time !

Resource Allocation in Heterogeneous Networks

  • 1.
    May, 2014 –Slide 1 Communication & Network Lab Resource Allocation in Heterogeneous Networks Master’s Thesis Presentation May 23rd 2014 By Trung Kien Vu Advisor: Prof. Sungoh Kwon Committee: Prof. Chong-Koo An, (chair) Prof. Sungoh Kwon, Prof. Sunghwan Kim.
  • 2.
    May, 2014 –Slide 2 Communication & Network Lab Research Interests: - Interference management and resource allocation: Heterogeneous and Small Cell Networks.  Trung Kien Vu and Sungoh Kwon, “eICIC-based Interference Mitigation in Small Cell Networks”. [In preparation]. - Routing protocols: Mobile Ad-hoc Networks and Wireless Sensor Networks.  Trung Kien Vu and Sungoh Kwon, “Mobility-Assisted On-Demand Routing Algorithm for MANETs in the Presence of Location Errors”, The Scientific World Journal, vol. 2014, Article ID 790103, 11 pages, 2014.
  • 3.
    May, 2014 –Slide 3 Communication & Network Lab Outline: 1. Introduction 2. Problem and Contributions 3. System Model 4. Proposed Algorithms 5. Simulation Results 6. Conclusions
  • 4.
    May, 2014 –Slide 4 Communication & Network Lab Introduction:  The demand for mobile data traffic  Solution: Heterogeneous Networks (HetNets)
  • 5.
    May, 2014 –Slide 5 Communication & Network Lab Introduction:  What is HetNet?  Consisting of multiple types of access nodes such as macro cells and smallcells (picocells and femtocells).  Smallcells are deployed under the coverage of macrocell.  Smallcells provide the indoor and outdoor wireless services by extending the network coverage and increasing the network capacity. Node types Transmission Power Coverage Macrocells 43-46 dBm Few Km Picocells 23-30 dBm ≤ 300 m Femtocells ≤ 23 dBm ≤ 50m
  • 6.
    May, 2014 –Slide 6 Communication & Network Lab Problem:  HetNet consists of Macrocells and Femtocells.  Macrocells and Femtocells share the same radio frequency  It causes the cross-layer interference between Macrocells and Femtocells Macrocells Picocells Femtocells
  • 7.
    May, 2014 –Slide 7 Communication & Network Lab Problem: Example  Macro base station: MeNB  Femto base station: HeNB  Macro User: MUE  Femto User: HUE Figure 2: System
  • 8.
    May, 2014 –Slide 8 Communication & Network Lab Previous work:  Enhanced Inter-Cell Interference Coordination eICIC is introduced to address cross-layer interference between macro and femtocells.  Almost Blank Subframe ABSF is one of eICIC techniques in which the interfering cell will stop using some subframes in order to reduce the ICI. No Transmission Transmission Figure 3: ABSF subframes
  • 9.
    May, 2014 –Slide 9 Communication & Network Lab Previous work:  Previous work use fixed ABSF pattern and all HeNBs are globally set to same ABSF based on number of users.  There is no coordination mechanism between femtocell base stations HeNBs.  How many and which subframes should be muted ?
  • 10.
    May, 2014 –Slide 10 Communication & Network Lab Contributions:  In this paper, our contributions include:  Dynamically optimal ABSF Selection Algorithm for each HeNB based on the Quality of Service of macro users.  Interference HeNB Coalition Algorithm to reduce the mutual interference.
  • 11.
    May, 2014 –Slide 11 Communication & Network Lab System Model: Objective  Find the optimal muted rate 𝛼 𝑚 for each MUE m. (defined as number of muted subframes (ABSF) over number of all subframes)  That satisfies the Signal-to-Interference-and-Noise Ratio (SINR) 𝛾 𝑚 of MUE m. Minimize 𝛼 𝑚 Subject to 𝛾 𝑚≽ 𝛾0 , m  ℳ 𝛾0 : SINR threshold ℳ : Set of MUEs
  • 12.
    May, 2014 –Slide 12 Communication & Network Lab System Model:  The SINR 𝛾 𝑚 at link ℒ 𝑚 between the MeNB and the MUE m is calculated as P(m) and G(M,m) : the transmission power of MeNB and path gain between MeNB and MUE. 𝑃(𝐹𝑓 and 𝐺(𝐹𝑓, 𝑚 : the transmission power of HeNB and path gain between HeNB and MUE. ℱ is a set of HeNB. 𝜎 𝑚 : the thermal noise at macro user m. ( , ) ( ) m m G M m P m    The received power from MeNB Total interference and noise ( ) ( , )m f f f m P F G F m   F
  • 13.
    May, 2014 –Slide 13 Communication & Network Lab System Model:  ℒ : the set of links from the MeNB to their serving MUEs, ℒ = (ℒ1, . . ; ℒℳ).  The constraint 𝛾 𝑚≽ 𝛾0 can be transformed in matrix form as  𝐅 𝑚, 𝑓 = 𝐺 𝐹 𝑓,𝑚 𝛾0 𝐺(𝑀,𝑚 , 𝐏ℱ = (P(1), … , P(𝐹𝑓 )) 𝑇  𝐏ℳ = (P(1), … , P(ℳ)) 𝑇 ,  𝒃 = (b(1), … , b(ℳ)) 𝑇 such that 𝑏(𝑚 = 𝛾0 𝜎 𝑚 𝐺(𝑀,𝑚  FP P bF M
  • 14.
    May, 2014 –Slide 14 Communication & Network Lab Proposed Algorithm – ABSF Selection:  When the HeNBs stop transmission on some subframes, the (SINR) 𝛾 𝑚 at link ℒ 𝑚 can be rewritten as  The received power from HeNbs is reduced in order to increase the SINR of MUE m. ( , ) ( ) ( ) ( , )(1 ) m f f f m m G M m P m P F G F m       F Reduced interference rate
  • 15.
    May, 2014 –Slide 15 Communication & Network Lab Proposed Algorithm – ABSF Selection:  Now, our objective can be transformed as Minimize 𝛼 𝑚 Subject to 𝐀𝛼 𝑚 ≽ 𝐁  where  A unique solution to this problem is  Now, we already get the optimal muted rate for MUE , .     A P B P P b F F M F 1 ( ) .T T m    A AA B
  • 16.
    May, 2014 –Slide 16 Communication & Network Lab Proposed Algorithm – Interfering HeNB Coalition  To group the mutual interfering HeNB to cooperate in ABSF mode efficiently. Figure 4: Coalition Example
  • 17.
    May, 2014 –Slide 17 Communication & Network Lab Proposed Algorithm – Interfering HeNB Coalition  Including 2 mechanisms:  Mechanism 1: to find the victim MUEs affected by each HeNB.  Mechanism 2: to group HeNBs having same Victim MUE.
  • 18.
    May, 2014 –Slide 18 Communication & Network Lab Proposed Algorithm – Interfering HeNB Coalition  Mechanism 1: to find the victim MUEs affected by each HeNB.  Detect the Victim MUEs  Report interfering HeNB’s list to MeNB.  Do set intersection algorithm by MeNB having the same HeNBs.  Send Victim MUE’s list to HeNB
  • 19.
    May, 2014 –Slide 19 Communication & Network Lab Proposed Algorithm – Interfering HeNB Coalition  Mechanism 2: to group HeNBs having same Victim MUE.  Exchange the VMUE’s list to neighbor HeNBs  Do set intersection algorithm having the same VMUE.  Set the muted rate for HeNB.  Active the ABSF mode.
  • 20.
    May, 2014 –Slide 20 Communication & Network Lab Simulation Results  Simulation Parameters Parameter Values System bandwidth 10 MHz Channel Model Urban Macro-Femto Scenario Model MeNB Tx 46 dBm Number of MUEs 20-100 HeNB Tx 23 dBm Number of HeNBs 40-400 Number of MUE 40-400 Thermal Noise -174 dBm/Hz Noise Figure 9 dB Simulation Run Times 1000
  • 21.
    May, 2014 –Slide 21 Communication & Network Lab Simulation Results  Algorithm Notations Names Notice Non ABSF Without eICIC Optimal ABSF eICIC with optimal muted rate Fixed ABSF - I eICIC with muted rate: 1/10 Fixed ABSF - II eICIC with muted rate: 2/10 Fixed ABSF – III eICIC with muted rate: 3/10 Each step Distance between HeNB and MUE is gradually increased in order to reduce interference
  • 22.
    May, 2014 –Slide 22 Communication & Network Lab Simulation Results Figure 5: The required muted rate 0 5 10 15 20 25 30 35 40 45 50 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 Step AvgMutedRateRequiredforallVMUEs Optimal Muted Rate
  • 23.
    May, 2014 –Slide 23 Communication & Network Lab Simulation Results Macro users Throughput Femto users Throughput Figure 6: The user throughput – performance balance between Macro and Femto users 0 5 10 15 20 25 30 35 40 45 50 1300 1350 1400 1450 1500 1550 1600 Step AvgThroughputofFemtocellUsers[Kbps] Optimal ABSF Non ABSF Fixed ABSF - I Fixed ABSF - II Fixed ABSF - III 0 5 10 15 20 25 30 35 40 45 50 400 500 600 700 800 900 1000 Step AvgThroughputofMacrocellUsers[Kbps] Optimal ABSF Non ABSF Fixed ABSF - I Fixed ABSF - II Fixed ABSF - III
  • 24.
    May, 2014 –Slide 24 Communication & Network Lab Simulation Results 0 5 10 15 20 25 30 35 40 45 50 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 Step OutageProbabilityofMacroUsers Non ABSF Fixed ABSF - I Fixed ABSF - II Fixed ABSF - III Optimal ABSF Figure 7: Outage Probability. # 𝑜𝑓 𝑈𝑛𝑠𝑎𝑡𝑖𝑠𝑓𝑖𝑒𝑑 𝑈𝑠𝑒𝑟𝑠 # 𝑜𝑓 𝑡𝑜𝑡𝑎𝑙 𝑈𝑠𝑒𝑟𝑠
  • 25.
    May, 2014 –Slide 25 Communication & Network Lab Conclusions  To address the cross-layer interference between macro and femto cell layers by  Propose a dynamically optimal ABSF eICIC framework based on the quality of service of macro users.  Group interfering HeNBs in ABSF mode to reduce mutual interference among HeNBs  The simulation results show that our algorithms outperform previous work and bring good solutions for smallcell networks.
  • 26.
    May, 2014 –Slide 26 Communication & Network Lab Thank you for your time !