a channel self-organizing protocol for multi-cell plc networks · sharing transmission capacity (2)...
TRANSCRIPT
www.ifn.et.tu-dresden.de/TK/
A Channel Self-Organizing Protocol for Multi-Cell PLC Networks
Le Phu DoITG-FG-5.2, 07 Oct. 2010
www.ifn.et.tu-dresden.de/TK/
2Contents
• PLC Networks with Multi-Cell Structure
• Sharing Transmission Capacity between Multiple PLC Cells
• Realization of Channels
• Solution Proposal– Communication between Cells– Algorithm Description– Example Results
• Conclusions
Broadband-PLC Network
Low voltage power supply network
High or medium voltage
PLC access network
In house network
Telecommunication backbone network
BS
RP
CB
Transformer station
Base station
Repeater
Cable box
BS
RP
CB
data signal line
power supply line
3
www.ifn.et.tu-dresden.de/TK/
Multi-Cell PLC Access Network Examples
• LV: Low Voltage
• MV: Medium Voltage
• HV: High Voltage
MV/LVMV/LVMV/LVMV/LV
LV LV LV LV
MV/LV Subst.
MV/LV Subst.
MV/LV Subst.
MV/LV Subst.
Backbone
The different colors indicate traffic coming from different LV PLC cells. MV/LVMV/LVMV/LVMV/LV
LV LV LV LV
HV/MV
MV/LV Subst.
MV/LV Subst.
MV/LV Subst.
MV/LV Subst.
MV/LV
LVHV/MV
MV/LV Subst.
MV/LV
LV
MV/LV Subst.
MV/LVMV/LV
LV LV
MV/LV Subst.
MV/LV Subst.
Backbone
Source: OPERA
4
www.ifn.et.tu-dresden.de/TK/
Transmission Capacity and Coexistence
21 22 23
24 25
5
6 46 47 48
43 44 45
739 40 41 42
• Interference: In-Line and In-Space
• The Inter Phy Protocol (IPP)*: Idear: Detection of the neighbor cells by listening to the beacon signal from the neighbors
Sys.1
Sys.2
Sys.1
Time frame t
Control Beacon period
* Specified in the IEEE P1901
5
www.ifn.et.tu-dresden.de/TK/
www.ifn.et.tu-dresden.de/TK/
6Organization of Channels
F
t
Control Beacon period
CH1
CH2
CH3
CH4
…
… CHF
CHF-1CH1
CH2
…
…
time frame (TF)
• A Resource Unit (Channel-CH) is configured as– Time Slot and
Frequency Band
• Dynamic reservation/ allocation of channels to each cell based on its number of active users
• To be investigated – Channel Reuse: to improve possible transmission capacity
– Fairness: same reservation resource per active users, in different cells
www.ifn.et.tu-dresden.de/TK/
7Objectives
• Sharing available transmission resource by all cells
• Maximizing the channel reuse: maximizing the transmission capacity of the network
• Fairness between active users in the cells: Same P for all cells
• Calculation of P: Number of allocated channels per number of active users:
i Ci iP B N=
Ni: =1 if no active user, =number of active users otherwise
BCi: Number of allocated channels in cell Ci
www.ifn.et.tu-dresden.de/TK/
8Sharing Transmission Capacity
• Static Distribution: Each cell is allocated a fixed amount of channels– User changes status (active/idle/off) dynamically -> may be unfair
between users in different cells
CH1, CH2
CH3, CH4
CH1, CH2
CH1,
CH2
Sharing Transmission Capacity (2)
• Dynamic Distribution: Channels of each cell will be varied according to the resource requirement– Centralized (like in the GSM system)
• Global control of the total capacity, need a control center
• Require resources for transmission between the control center and the HEs
– Distributed• Require communication between neighbor
cells
• Sharing the channel status with neighbors– Cell informs its status to its neighbors (in
beacon period)
– Processing received information to allow “moving” channels between cells
CH1, CH2
CH3, CH4
CH1
CH5
9
www.ifn.et.tu-dresden.de/TK/
10Between Cells Communication Issues
• Common Slot and Collision– One slot is reserved for inter-cell communications (in beacon period)
• Format– Regular broadcast the “channel announce” message (CA-Msg)
• Random transmission protocol– No collision detection mechanism
• Effect of collision
ID #D List_Ch (fi)
CA CA
TTF
t
Control Beacon period
AC
Traffic demand: #Active Users
AC: 1 Byte: Access or InHomeID: 6 Byte: Address of the Master#D: 1 Byte: ‚ of active usersList_Ch(fi): 2 bit for a channel• 0: Channel is free• 1: occupied by this cell• 2: used by one neighbor cell• 3: used by more than one neighbor cells
(length of a CA-Msg is 18 Byte for 80 channels)
www.ifn.et.tu-dresden.de/TK/
11Communication Between Cells
• Broadcasting of the CA-Msgs– Can be carried out by border nodes
• Sending of the CA-Msgs: – Randomly select a Beacon TS– No ACK
CPEHE
HE
RP
CPE 1
RP
CPE 3CPE 2
CPE4
Cell 3
Cell 1Cell 4
Cell 2
Transmission in downlink
Transmission in beacon period
Transmission in uplink
HE: Head endRP: RepeaterCPE: Customer Premise
Equipment
Cell 1 Cell 2
Send status
make decision
on channels,
send status
Make decision on channels
CA-Msg
CA-Msg
Between Cells Communication Issues
Cj
11
send ( i )j
i
pH
=+
• Probability to select a slot (1 neighbor has Hi neighbors)
1 11( j )
i
sendj
Cj i
pH H
= ⋅+∑
• Probability to select a slot
Ci
Cin
Cim
Ci2
Ci2n
Ci2m
Ci3Ci3n
Ci3m
12
www.ifn.et.tu-dresden.de/TK/
www.ifn.et.tu-dresden.de/TK/
13Master Station: Decision on Channels
• Two objectives⇒Two procedures– Maximizing the number of
channels allocated in all cells⇒ Seize Channels
– Same P – fairness (in different cells)
⇒ Release Channels
High-P
Low-P
Equil
P>min{PNB} “release”
P<min{PNB}“seize”
Transmit CA-Msg
Receive CA-Msg
Remove interferences
PNB: P of neighbor
i j
i
C j C iC
i j
B N B NB
N N−
−⎢ ⎥= ⎢ ⎥
+⎢ ⎥⎣ ⎦
Cj: neighbor of Ci , which has lowest P
www.ifn.et.tu-dresden.de/TK/
14An Example with the Procedures
• Channel seize– Combine channel in use by
its neighbors– seize the “unused” channel
• Channel release– Compare channel/demand
with neighbor (Cj) which has lowest channel/demand
– Find number of channels has to be released: can be obtained by Cj
• Not used by neighbors (except this cell) of Cj
• Validity of channel information
– Each channel announce message has a limit validity – time out
CH1 CH2 Ch3 Ch4 Ch5 Ch6
C1
C3
Used
C2 2 3 2 1 3 3
CH1 CH2 Ch3 Ch4 Ch5 Ch6
C1
C2 2 3 2 1 3 3
Ne(C2)
C1 Rel.
C1
C1
C2 1 3 1 1 3 3
C3
C2 C3C1
Sys.
C2
C1
Sys.
www.ifn.et.tu-dresden.de/TK/
15Evaluation Metrics
Given: Set of channels: F={1,..,F} , Set of cells: C={1,..,S}, Cell Ci contains Ni active users, Same traffic demand from each active user
Channel status:• Channel usage vector in each Master (Head end-HE): ChCi=(ai,1, ai,2, ai,3, … ai,F)
where ai,f=1: Ci uses channel f, ai,f=0: otherwiseTo be evaluated:• Channel reuse factor:
• Number of allocated channels
• Fairness index: –– 1: perfect, greater is better
• Interferences
,1 1 1
S S F
All Ci i fi i f
B B a= = =
= =∑ ∑∑
AllCR
BfF
=
, , ,11 1
( · )S F S
All i f i j j fji f j i
I a i a=
= = ≠
= ∨∑∑
2
12
1
SCi
i iAll S
Ci
i i
BN
FBSN
=
=
⎛ ⎞⎜ ⎟⎝ ⎠=
⎛ ⎞⋅ ⎜ ⎟
⎝ ⎠
∑
∑
1 1AllFS ≤ ≤
Small Network for Analysis
• Two cells (Ci=2), are neighbors (ii,j=1)• F=80
• C2 with fixed number of users N2 = 9• C1 with increasing and decreasing
number of users over the time (every 100TF), between 1 and 9
•
Cell 1
Cell 21 2 0.5send sendp p= =
16
www.ifn.et.tu-dresden.de/TK/
Small Network for AnalysisB C
1
#time frame (TTF)N
1#time frame (TTF)
Cha
nnel
-reu
se, f
CR
Fairn
ess,
F ALL
17
www.ifn.et.tu-dresden.de/TK/
0
10
20
30
40
50
60
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 0
1
2
3
4
5
6
7
8
9
10
11
12#Channels
#Users
0
0.5
1
1.5
2
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 0
1
2
Small Network for AnalysisB C
1
#time frame (TTF)N
1#time frame (TTF)
Cha
nnel
-reu
se, f
CR
Fairn
ess,
F ALL
18
www.ifn.et.tu-dresden.de/TK/
0
10
20
30
40
50
60
1000 1200 1400 1600 1800 2000 0
1
2
3
4
5
6
7
8
9
10
11
12#Channels
#Users
0
0.5
1
1.5
2
1000 1200 1400 1600 1800 2000 0
1
2
Example Network
MV/LV
MV/LV
MV/LV
MV/LV
LV
HV/MV
MV/LV Subst.
MV/LV Subst.
MV/LV Subst.
MV/LV Subst.
MV/LV
HV/MV
MV/LV Subst.
MV/LV
MV/LV Subst.
Backbone
MV/LV
MV/LV Subst.
MV/LV
MV/LV Subst.
MV/LV
MV/LV Subst.
LV
LV
LV
LV
LV
LV
MV/LV
MV/LV Subst.
MV/LV
MV/LV Subst.
MV/LV
MV/LV Subst.
MV/LV
MV/LV Subst.LV
LV LV
LV LV
LV
Cell 1
Cell 2
Cell 3
Cell 4
Cell 5Cell
6Cell
7
Cell 8
Cell 9
Cell 10
• Ring of 10 cells• Interferences with two neighbors• 9 users/cell with on/off model: • On and Off phases Geo
– E{TON}= 100TF– E{TOFF}= 100TF
19
www.ifn.et.tu-dresden.de/TK/
Simulation Results
0
0.5
1
1.5
2
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 0
1
2
3
4
5
6Fairness
Reuse
0
10
20
30
40
50
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 0
1
2
3
4
5
6
7
8
9
10Num Channels
Num Users
Cell1
B C1
#time frame (TTF)
N1
#time frame (TTF)
Cha
nnel
-reu
se, f
CR
Fairn
ess,
F ALL
20
www.ifn.et.tu-dresden.de/TK/
Simulation ResultsA
ver.
Cha
nnel
/use
r
Ave
r. U
ser/C
ell
E.g. If a channel can be used to transmit 1Mbps → Average speed/user ≈8.2Mbps
0
10
20
30
40
50
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 0
1
2
3
4
5
6
7
8
9
10Avg. Channels Per User
Avg. Users Per Cell
#time frame (TTF)
21
www.ifn.et.tu-dresden.de/TK/
#cha
nnel
s
0
100
200
300
400
500
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000
InterferenceEffective
#time frame (TTF)
#effective channels = #allocated channels - #interfering channels
Conclusions
• Model for dynamic channel allocation for Multi-Cell PLC supports for changing of traffic demands (user’s status)
• Allow channel self-organizing between cells with the distributed negotiation strategy: No need addition central control unit
• Required the realization of channel announce messages between neighbor cells
• Can be applied to Access PLC, In-Home PLC
22
www.ifn.et.tu-dresden.de/TK/
www.ifn.et.tu-dresden.de/TK/
23
Thank You!
Ref: L. P. Do, R. Lehnert, "Distributed Dynamic Resource Allocation for Multi-Cell PLC Networks", 13th IEEE International Symposium on Power-Line Communications and Its Applications (IEEE-ISPLC 2009), Dresden, Germany. 29 Mar-01 Apr. 2009.