A study of virtualization control A study of virtualization control - - PowerPoint PPT Presentation
A study of virtualization control A study of virtualization control architectures for mobile networks architectures for mobile networks APII Workshop APII Workshop 2012. 10. 29 2012. 10. 29 Korea University Korea University Choong-Ho Cho
chcho@korea.ac.kr
chcho@korea.ac.kr Lead agency: Korea Univ. Joint agency: Chungbuk Univ. Lead agency: Korea Univ. Joint agency: Chungbuk Univ.
2
based on the VNET
3
▶ Background
Prospect: The trend of mobile data volumes and types in the world (Cisco, 2012.2) Mobile VoIP (0.3%) Mobile game (1.1%) Mobile file sharing (3.3%) Mobile M2M (4.7%) Mobile web/data (20%) Mobile video (70.5%) (Unit: Exa Byte/month)
Recently, mobile data traffic has been increasing sharply because there has been a sharp rise in demand for using smart phones and tablet PCs. Most mobile traffic (70% of data traffic, 50% of voice traffic) occurs in indoor environments. Market survey agencies (ABI and Ovum research ) and Cisco said,
Mobile data traffic has been increasing sharply and the femtocell is one of the best methods to solve thi s problem.
Accordingly, Major mobile telecommunication companies have set up investment plans for femtocell networks to serve indoor users with QoS.
(year)
4
Major Mobile Network Operators (MNOs) have recently become interested in Enterprise Femtocell Networks (EFNs) because EFNs are not only a solution for serving indoor users with heavy traffic but also a highly marketable area.
However, users in EFNs have more interference than Conventional Femtocell Networks (CFNs) since Femtocell Access Points (FAPs) are densely deployed. Thus, an efficient resource management scheme for EFNs is necessary.
MNOs and Mobile Virtual Network Operators (MVNOs) are looking for solutions to maintain the current infra and reduce CAPEX/OPEX for the core networks.
FUEs (MNO users and MVNO users) usually send data traffic to a Femtocell GateWay (FGW) through a FAP and the FGW forwards it to the destination through the MNO’s core network. Thus, a FGW sharing scheme for MNO and MVNO users is necessary.
▶ Needs
5
Objects : An architecture of Virtualized EFN for MVNO and a Resource
Management Schemes of a EFN using the VNET core.
Architecture design for mobile data transmission in the virtual network of KOREN Business building Department store Subway Airport
6
1. Mobile network sharing solution for the EFN service 2. Architecture design for EFNs based on the VNET Architecture Design for EFNs based on the VNET
7
Core network sharing solution
Femto Gateway / Access network sharing solution
▶ Mobile network sharing solution for the EFN service
Mobile network sharing solution for the EFN service
A B C A B C A B C SDN SDN SDN FUE
(Femtocell User Equipment)
FAP FGW
Research range
8
Gateway sharing solution of the EFN
Gateway sharing solution using the VNET of future Internet
▶ Femto Gateway sharing solution
Future Internet VNET
EFN GW
MVNO-B GW MVNO-A GW MVNO-C GW Future Internet VNET
EFN GW
GW Sharing solution
a) Each MVNO uses its own GW b) MVNOs share a GW
A sharing GW platform of the EFN for the VNET of future Internet
resource manager
▶ A Gateway platform for the EFN sharing solution
Proposed GW platform for sharing solution of the EFN
VNET Creator & Monitoring system
Integrated VNET resource manager
Resource controller Resource controller
VNET DB Operator FAP
Control plane
SDN
Control plane
FUE
9
10
▶ 공통 실험 환경 ▶ CFN Architecture
The CFN architecture is connected to 3G and 4G (EPS). : The CFN is independent, and thus it can be connected to other networks.
The CFN architecture (Femto Forum)
11
FAP-MS: FAP Management System FCS-MS: Femto Control Server Management System FGW-MS: Femto GateWay Management System HSS: Home Subscriber Server VNetRA: Virtual Network Resource Allocation FUE: Femto User Equipment FUEv: Femto User Equipment with Virtual network FAP: Femto Access Point RRM Radio Resource Management FCS: Femto Control Server FMS: Femto Management System
The data traffic from FAP is divided into MNO and MVNO groups in FGW In case of MVNO traffic, FCS and VNet GW transmit data through VNET on the future internet
▶ EFN architecture in VNET of future Internet
12
▶ Call admission with fixed VNET resource
※ NAS : Non Access stratum
VNET resource reservation Data transmission Authentication
13
▶ Call admission with dynamic VNET resource
※ NAS : Non Access stratum
VNET resource reservation Data transmission Authentication
14
1. Summary of resource management 2. System environment 3. Resource allocation using a graph coloring algorithm 4. Resource allocation using fractional frequency reuse scheme 5. VNET resource management for MVNO 6. Performance evaluation
15
▶ 공통 실험 환경 ▶ Summary of Resource management for EFNs Resource management schemes with limited capacity of MNO and MVNO resource in EFNs
Wireless resource allocation with capacity unlimited (not considering MVNO)
We have proposed two resource allocation schemes to improve system performance Graph coloring algorithm and Fractional Frequency Reuse (FFR)
Wireless resource allocation with limited MVNO VNET capacity (considering MVNO)
FAPs assign wireless resource to MVNO Femtocell User Equipments (FUEs) considering the amount of VNET resources capacity for each MVNO
16
▶ 공통 실험 환경 ▶ Considerations and solutions in EFNs
EFNs have more interference than CFNs since FAPs are densely deployed in EFNs. In EFNs, an FAP Control Server (FCS) manages wireless resource allocation for FAPs considering interference. In this research, we proposed two schemes using Graph coloring algorithm and Fractional Frequency Reuse (FFR) scheme to assign wireless channels for FAPs.
In the scheme with a graph coloring algorithm, the FCS assigns different channels for FAPs with an Interference Matrix. In the FFR scheme, the FCS assigns channels in regular groups using a graph coloring algorithm. Then, FAPs allocate channels of super group for FUEs which are in an inner zone.
We analyzed the proposed schemes with one FUE in each FAP
The FFR scheme was worse than the Graph coloring scheme because of the number of FUEs. Thus, we will evaluate two schemes with more than 1 FUE.
17
▶ Signal to Interference plus Noise Ratio
The SINR, , on subchannel k (1≤k≤K ) for FUE m [1≤m≤M ] served by FAP n [1≤n≤N ] is expressed by
, PT 는 Tx power, . d0 is a reference distance which is typically assumed to be 1 meter indoors. dnm is distance between FAP n and FUE m. s and fc are the speed of light and carrier frequency, respectively α is the path loss exponent, is the white noise power . A binary matrix decribe the subchannel allocation among the FAPs, where denotes that subchannel k is assigned to FAP n, otherwise .
,
2 | 1 N N n j j kj k jm k nm k nm
a R R
k nm
nm T k nm
d d P R
dB in 4 / log 20
10
d f s
c
2 N
N K kj
a
] [ A 1
kj
a
kj
a
18
▶ Throughput
Given a specific , the achievable transmit rate, , in bps/Hz between FAP n and FUE m is obtained by
and are the maximum and minimum spectral efficiencies, respectively. and Pe is target bit error rate.
FUE throughput
is total bandwidth of allocated RBs for FUE m served by FAP n. k nm
, for for for
min min max max min max
nm nm
max
min
) 1 ( log 2
k nm
) 5 ln( / 5 . 1
e
P
,
nm nm nm
W T
nm
W
19
▶ 공통 실험 환경 ▶ Graph Coloring
Based on a given network topology, the interference graph G=(V, E ) can be constructed. The vertex set V and edge set E denote fBSs and the severe interferences between fBSs, respectively. We use a greedy coloring algorithm and it creates a graph with a minimum number of colors, χ. Each color denotes a group of subchannels which are calculated by K/χ.
20
▶ 공통 실험 환경 ▶ FFR scheme
In order to increase the system throughput, we propose another resource allocation method called the FFR scheme with Graph coloring. We divide total bandwidth into two large groups called super and regular groups. The super group is assigned for FUEs in the inner zone which have good SINR values. The regular group is divided and allocated four FUEs in the outer zone using a Graph coloring algorithm.
Super group Regular group
21
▶ 공통 실험 환경 ▶ Resource allocation for MVNO FUE considering fixed MVNO VNET resources
Each MVNO has VNET resource (TMVNO) between FCS and VnetGW for enterprise femtocell networks. Then, the FCS manages FAPs to allocate wireless resource considering TMVNO . Each FAP allocates wireless resources using Fair scheduling algorithm to FUEs.
MVNO 22 13 12
T T T T
FAP 1 MNO Service MVNO Service FUE 1
(MNO)
VNET
VNetGW
FAP 2 FUE 1
(MNO)
FUE 2
(MVNO)
FUE 2
(MVNO)
FUE 3
(MVNO)
T12 T11 T13 T21 T22 TMVNO
FCS
An example of wireless resource allocation for MVNO FUEs.
22
▶ 공통 실험 환경 ▶ Experiment environment
Carrier Frequency : 2GHz Total bandwidth : 2MHz The number of channels (K): 20 The number of FAPs (N): 25 (5x5 grid model) The number of FUEs/FAP [M]: 1 FAP Tx power: 10dBm : -174dBm/Hz : 3 , : 10-3 : 4.4 bps , : 0 bps Traffic model: Fluid flow model, User traffic density: 1Mbps Building size: 50m x 50m We assume no interference from the macro base stations.
2 N
th
max
min
23
▶ 공통 실험 환경 ▶ Resource allocation
We compare three scheme such as Reuse 1, Graph coloring, and FFR in terms of SINR and throughput. Figures below shows resource allocation in our simulation. ( o: FAPs using channels in regular group, x: FUEs, : FAPs using channels in super group) The colors of FAPs mean different channels.
10 20 30 40 50 10 20 30 40 50 10 20 30 40 50 10 20 30 40 50
Graph coloring FFR
24
▶ 공통 실험 환경 ▶ SINR and user throughput
Graph coloring and FFR show better performance than Reuse 1 in terms of SINR and User throughput CDF. In user throughput, Graph coloring has better throughput than FFR since the number of FUE is 1. Accordingly, we now evaluate the proposed schemes with over 1 FUE.
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 User throughput (Mbps) CDF Reuse 1 graph FFR
5 10 15 20 25 30 35 40 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 SINR (dB) CDF Reuse 1 graph FFR
25
1. Summary of a FCS (Femto Control System) and a FMS (Femto Management System) 2. Architecture of framework for a FCS and a FMS 3. Implementation of framework for EFN on KOREN/SDN
26
A FCS(Femtocell Control Server) & a Femtocell Management Server (FMS) to control and monitor the EFNs
A Femtocell Control Unit (FCU) manages radio resource allocation in FCS. A Femto Virtual Network Unit(FVNU) creates Vnet connections and manages a Vnet resources.
Develop an web-based monitoring system on the VNET environment of future Internet ▶ FCS and FMS
27
FCSs manages their own FAPs and create VNETs on the future Internet. MNOs or MVNOs monitor the data traffic in the EFN and VNET using the FMS.
▶ The overall system configuration
28
Representation of the system as modular units that make up FCS and FMS Make progress in the development of prototype as modular units
▶ Module design for FCS and FMS
Femto GW Vnet Aggrigator
29
Obtain FUEs’ Transmission rates using the Femtocell RRM function Create the MVNO and Perform Interworking process of VNET
▶ The block diagram of FCS Emulator Module
30
Based on the data of the resource allocation algorithm for each FCS Based on the data according to each communication link
▶ The block diagram of an FMS monitoring module
Monitoring
31
Graph coloring algorithm
▶ Flowchart of each RRM algorithm module
Data transmission module Environmental Variable data, such buildings, FAP, FUE SINR of each FUE Resource allocation scheme using Graph coloring Transfer data Module sent to destination through transmission rate based on the measured SINR of each FUE
Start Create an interference matrix End Graph coloring Resource allocation
32
Implement the functionality of the traffic generator as a prototype emulator (added speed limit mode) We are now testing this emulator in out intranet In addition, we are testing data transmission through KOREN between two universities (Korea university in Seoul and Sejong)
▶ Development of a Prototype system for FCS Emulator
FCS traffic generator (server) FCS traffic generator (client)
33
▶ Development of an FCS Emulator for performance analysis
MNOs or MVNOs can monitor the entire experimental environment, FUE throughput, FUE SINR information. Warning signals if the EFN is in an abnormal situation.
34
▶ Development of an FMS monitoring system
Web-based monitoring system shows graphs and tables of the transmission rates of FUEs with various resource allocation schemes. We are analyzing the system performance w/wo using the VNET of the future Internet and show the results. Also, MNOs and MVNOs can check their FUEs’ transmission rates and utilization of fixed VNET resources. Warning signals if the EFN is in an abnormal situation
Each FUE’s Status information Realtime detection of FAP status
35
▶ Test Environment
We are now testing the EFN using KOREN. Data transmission through the VNET Data transmission through the KOREN (without the VNET)
The difference of testing environment according to KOREN and KOREN Virtual Network
36
▶ Resource management
Analyze the system performance of FFR in terms of SINR and throughput as the number of FUEs increases. FCS performance considering interference from macro cells Resource allocation for MVNO FUE with limited MVNO VNET resources
▶ Implementation
Analyze the proposed schemes with our FCS emulator using the VNET
Wireless resource allocation (not considering MVNO) Wireless resource allocation with limited MVNO VNET resources (considering MVNO) Develop a web-based femtocell traffic monitoring system