Transcript netlab.csie.ntut.edu.tw

The Design of the Power Saving
Mechanisms in IEEE 802.16e
Networks (Defense)
Student: Lei Yan (嚴雷)
Advisor: Dr. Ho-Ting Wu (吳和庭)
Date: 2009/07/23
Institute of Computer Science and Information
Engineering,
National Taipei University of Technology
1
Outlines
 Backgrounds
and motivation
 Spec-defined power saving mechanism
 Proposed power saving mechanism
 Performance evaluation
 Conclusion and future works
2
Brief of IEEE 802.16
 Also
called “WiMAX” (Worldwide
Interoperability for Microwave Access)
 The “wireless” last mile technology
 Low cost and free to geographical limits
 Includes PHY/MAC solutions
 Higher transmission rate than 3.5G
 Ideal transmission rate and coverage: 75 Mbps
and 50 km
 Now been merchandised in Taiwan
3
Layering related to connections



Convergence sublayer
 Classification
Common Part sublayer
 Establishment
 Change
 Deletion
 Call admission control (CAC)
 Bandwidth allocation (BWA)
 Packet scheduling
Security sublayer
 Encryption/Decryption for data
MAC Common Part Sublayer
(CPS)
4
Frame structure of WiMAX
Frame n-1
Frame n
Frame n+1
DL Subframe
Preamble
FCH
DL-Burst
#1
DLFP
IE IE IE IE
UL Subframe
DL-Burst
#5
DL-Burst
#2-4
DL-MAP
UL-MAP
MAC
PDU
Pad
BW
request
UL-Burst
#1
R
UL-Burst
T
#2
G
MAC messages, Mac PDUs
IE IE IE
IE IE IE IE IE
Preamble FCH DL-Bursts#1-4 DL-Burst #5 DL-Burst #6
Preamble MAC
(optional) PDU
T
Initial
T ranging
G
DL-Burst
#6
Ranging
Opportunity
Preamble MAC
(optional) PDU
BW
request
UL-Burst
#1
MAC Midamble MAC
PDU (optional) PDU
UL-Burst
#2
MAC
PDU
Pad
5
每個DL-Burst內的資料，有可能是要送給不同SS
每個UL-Burst內的資料，都必須來自同一台SS
Outlines
 Backgrounds
and motivation
 Spec-defined power saving mechanism
 Proposed power saving mechanism
 Performance evaluation
 Conclusion and future works
6
Renaming for QoS metrics: Power saving class (PSC)
Service
type
Service type
for power
saving
UGS
Usage
UGS
VoIP
ertPS
VoIP (silence
suppressi
on)
ERTVR
PSC
Real-time
Non-Real
time
NRTVR
(nrtPS)
PSC I
BE
RTVR
rtPS
MPEG
PSC II
NRTVR
nrtPS
UGS
FTP
RTVR (rtPS)
BE
BE
HTTP
PSC III
ERTVR
(ertPS)
7
States of a MS with power
saving function
MS operation
ON-state (RF
device ON)
Sleep mode (RF
device OFF)
OFF-state (RF
device OFF)
Listening interval
(RF device ON)
Normal mode
time
Normal mode
Sleep mode
(a)
Normal mode
time
TS
TL
TS
TL
TS
TS
TL
TS
TL
TS
TL
TS
TL
TS
TL
TL
TS
...
TL
(b)
TS
TL
...
(c)
TS
(d)
8
TS: The sleep duration
TL: The listening duration
Outlines
 Backgrounds
and motivation
 Spec-defined power saving mechanism
 Proposed power saving mechanism
 Performance evaluation
 Conclusion and future works
9
What our research work has done?
 Includes
UGS/ertPS, rtPS, nrtPS, and BE for both
UL/DL traffics
 Integrated consideration to CENTRALLY
arrange calls to prolong sleep durations
 Complete solution (CAC + BWA + packet
scheduling)
 Reduction of ON-OFF alternations in MS
 Q: Why packet scheduling for power saving?
 A: Since BS can control all states for MS
PRECISELY with time
10
t
req
ues
idth
dw
es
qu
re
t
Ban
h
dt
wi
nd
Ba
BWA
adoption
1st OFDMA frame
U1
R1
Structure of our research
work
BS
N1
B1
...
UM RM NM BM
2nd OFDMA frame
Gr
U1
R1
N1
B1
...
AMPth OFDMA frame
an
t
UM RM
...
fo
r BM
NM b
ot
hU
U1
L
an
d
R1
N1
B1
...
UM RM NM BM
DL
Ui: UGS/ertPS packets for MS i
Ri: rtPS packets for MS i
Ni: nrtPS packets for MS i
Bi: BE packets for MS i
U-cycle for 1st packet
Packet scheduling
algorithm adoption
for power saving
U1 from
1st
OFDMA
frame
U1 from
2nd
OFDMA
frame
...
U-cycle for 2nd packet
DUGS: The delay bound for UGS
A complete cycle duration C for the power saving function
11
…
Assumptions for our research work
Frame overflow is
allowed (to schedule
across the bound
between frames)
 Merge the listening
ON-state (RF
duration into ONdevice ON)
state duration

Traffic 1
Fram e duration
Traffic 2
Traffic 3
…
MS operation
Sleep mode (RF
device OFF)
OFF-state (RF
device OFF)
Listening interval
(RF device ON)
Fram e duration
12
BWA and scheduling
BR
BWA->grant
BS
OFF state for call
A
ON state call A
OFF state time for call A
ON state for call A
time
t1
MS
Scheduling cycle (determined by scheduling algoirithm)
t2
BS
BS
GRANT
BR i
i
Ti cycle
1
ti 1
BR i+1
Ti estimated
GRANT
i+1
ti
13
MS
MS
Judgment in the proposed BWA algorithm
Gi 1  Bi  Bi 1
estimated
estimated
Ci  Ti
 ARi 1  Ti

(5)
cycle
Ti 1
Bi  Ci  Bi 1  Ci 1

 Bi  Ci iff . T estimated  T cycle  BWCAC
Gk  
(6)
i
i 1
 BW  T estimated , otherwise
 CAC i

(5) determines the credit. (6) is the stabilized version of (7)
14
Deployed packet scheduling algorithm
OFF-state for all MSs U1 R1 N1 B1 U2 R2 N2 B2
ON-state duration
for MS 1
…
UM RM NM BM OFF-state for all MSs
time
OFF-state duration for MS 1
Round-Robin fashion for the scheduling
 Pros: In the algorithm itself, all MSs have intact sleep
duration (intuitive design)
 Cons: No guarantee for UGS/ertPS delay
 Special assumption: all delay bounds are set and equal to
DUGS (100 ms)

15
Outlines
 Backgrounds
and motivation
 Spec-defined power saving mechanism
 Proposed power saving mechanism
 Performance evaluation
 Conclusion and future works
16
(1) Call request for new calls
System structure
of our research
work
(2) Admit/reject calls with CAC algorithm
(3) Bandwidth request for admitted calls
BS
DL
DL
(4) Determine the grant size/timing via BWA algorithm and packet scheduling
algorithm, respectively
UL
UL
(5) Call termination
(5) Call termination
MS
MS
17
BS
MS
Parameters for simulation
Number of BS
1
Number of MS
20~200
System capacity (Mbps)
50
Frame duration (ms)
5
Capacity per frame (byte)
31250
Amplification ratio
5, 10, 15, and 20
Simulation duration (sec)
500
Call duration (sec)
Uniform(60,200)
18
Parameters for simulation (cont.)
Service type
UGS
ertPS
rtPS
nrtPS
BE
Delay bound
(ms)
100
100
200
300
400
Transmission
rate
required in
call
request
(kbps)
64
64
Uniform(48,
60)
Uniform(48,
60)
Uniform(48,
60)
Bandwidth
request
size per
frame
(byte)
40
40
Uniform(30,
50)
Uniform(30,
50)
Uniform(30,
50)
0.44
0.44
0.03
0.02
0.20
Arrival rate
ON/OFF ratio
for ertPS
0.4/0.6
19
The traditional WiMAX scheme we compare to
No BWA and
scheduling (no
packet
dropping
rate/delay)
 We amplified
the negative
credit to
reduce packet
dropping/delay

Items
Proposed
Traditional
CAC algorithm in
use
Traditional
Traditional
Aggregation for
grant size
YES
NO
Adaptively adjust
the grant size
YES
NO
Frame overflow
Allowed
Not allowed
RF device is
allowed to be
turned off
YES
NO
20
Three scenarios for our comparison
main
Simulation results
WiMAX:
Management for BS
Proposed WiMAX
scheme
Proposed
BWA/Packet
scheduling algorithm
BS:
CAC, MS generation,
and management For
calls
Traditional
WiMAX scheme
MS:
Call generation
Traditional WiMAX
scheme
Proposed WiMAX
scheme with BWA
Proposed WiMAX
scheme without BWA
Call:
Parameter generation
21
Start
processing a
call
Gen. a call
Flow chart of the program
Perform CAC
for each call
NO
Is CAC
admitted
YES
Reject call
Is the first
BR
NO
YES
1. Run sch. alm. for call
2. Sch. time for
BR/BWA
Is call finished
shceduling
YES
NO
YES
Is call
expired
Are all calls in MSi
finished scheduling
NO
NO
Terminate call
YES
Sch. the rest at
next frame
YES
Is the call sch.
exceeding one frame
duration
MS i sleeps
NO
NO
Is MS being
awaken
Is scheduling
exceeding delay
bound
YES
NO
YES
Drop the rest
data
22
Increase of sleeping ratio

Heavier loading causes longer C but reduce number of cycle (number of
granting for MSs) => avg. proportion in ON-state is decreased
Aggregated frames
Aggregated frames
data
...
Aggregated frames
data
Cycle C
data
Cycle C
Simulation duration
Cycle C
Sleeping ratio for AMP=5
Sleeping ratio
0.999
0.998
Sleeping ratio with BWA
0.997
0.996
Sleeping ratio without
BWA
0.995
0.994
0.993
20
40
60
80
100 120
MS
140 160 180
200
23
Avg. pkt. dropping rate
Avg. pkt. dropping rate for AMP=15
0.25
0.2
UGS with BWA (Pro)
0.15
UGS without BWA (Pro)
0.1
UGS (T ra)
0.05
0
20
40
60
80
100
120
140
160
180
200
MS
Scenario 1: AMP=15
Avg. pkt. dropping rate
Avg. pkt. dropping rate for AMP=15
0.25
0.2
ertPS with BWA (Pro)
0.15
ertPS without BWA (Pro)
0.1
ertPS (T ra)
0.05
0
20
40
60
80
100
120
MS
140
160
180
200
24
Avg. pkt. dropping rate
Avg. pkt . dropping rat e for AMP =15
0.25
0.2
rt P S wit h BWA (P ro)
0.15
rt P S wit hout BWA (P ro)
0.1
rt P S (T ra)
0.05
0
20
40
60
80
100
120
140
160
180
200
MS
Avg. pkt. dropping rate
Avg. pkt . dropping rat e for AMP =15
0.25
0.2
nrt P S wit h BWA (P ro)
0.15
nrt P S wit hout BWA (P ro)
0.1
nrt P S (T ra)
0.05
0
20
40
60
80
100
120
140
160
180
200
MS
Avg. pkt. dropping rate
Avg. pkt . dropping rat e for AMP =15
0.25
0.2
BE wit h BWA (P ro)
0.15
BE wit out BWA (P ro)
0.1
BE (T ra)
0.05
0
20
40
60
80
100
120
MS
140
160
180
200
25
Avg. pkt. dropping rate
Avg. pkt. dropping rate for AM P=20
0.6
0.5
0.4
UGS with BWA (Pro)
0.3
UGS without BWA (Pro)
0.2
UGS (Tra)
0.1
0
20
40
60
80
100
120
140
160
180
200
MS
Scenario 2: AMP=20
Avg. pkt. dropping rate
Avg. pkt. dropping rate for AM P=20
0.6
0.5
0.4
ertPS with BWA (Pro)
0.3
ertPS without BWA (Pro)
0.2
ertPS (Tra)
0.1
0
20
40
60
80
100
120
MS
140
160
180
200
26
Avg. pkt. dropping rate
Avg. p kt. drop p ing rate for AM P=20
0.6
0.5
0.4
rtPS with BWA (Pro)
0.3
rtPS without BWA (Pro)
0.2
rtPS (Tra)
0.1
0
20
40
60
80
100
120
140
160
180
200
MS
Avg. p kt. drop p ing rat e for AM P=20
Avg. pkt. dropping rate
0.6
0.5
0.4
nrtPS wit h BWA (Pro)
0.3
nrtPS wit hout BWA (Pro)
0.2
nrtPS (Tra)
0.1
0
20
40
60
80
100
120
140
160
180
200
MS
Avg. p kt. drop p ing rate for AM P=20
Avg. pkt. dropping rate
0.6
0.5
0.4
BE with BWA (Pro)
0.3
BE without BWA (Pro)
0.2
BE (Tra)
0.1
0
20
40
60
80
100
120
MS
140
160
180
200
27
Avg. packet delay (ms)
Avg. packet delay for AMP=20
70
60
50
40
UGS with BWA (Pro)
UGS without BWA (Pro)
30
20
10
0
UGS (Tra)
20
40
60
80
100
120
140
160
180
200
MS
Avg. pkt. delay (ms)
Avg. pkt. delay for AMP=20
70
60
50
40
ertPS with BWA (Pro)
ertPS without BWA (Pro)
30
20
10
0
ertPS (Tra)
20
40
60
80
100 120 140 160 180 200
MS
28
Avg. pkt. delay (ms)
Avg. pkt. delay for AMP=20
70
60
50
40
rtPS with BWA (Pro)
rtPS without BWA (Pro)
30
20
10
0
rtPS (Tra)
20
40
60
80
100
120
140
160
180
200
MS
Avg. pkt. delay (ms)
Avg. pkt. delay for AMP=20
70
60
50
40
nrtPS with BWA (Pro)
nrtPS without BWA (Pro)
30
20
10
0
nrtPS (Tra)
20
40
60
80
100
120
140
160
180
200
MS
Avg. pkt .delay
Avg. pkt .delay for AMP-20
70
60
50
40
BE with BWA (Pro)
BE without BWA (Pro)
30
20
10
0
BE (Tra)
20
40
60
80
100
120
MS
140
160
180
200
29
Avg. pkt. dropping rate
Avg. pkt. dropping rate for MS=120
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
UGS with BWA (Pro)
UGS without BWA (Pro)
5
10
15
20
AMP
Scenario 3: MS=120
Avg. pkt. dropping rate
Avg. pkt. dropping rate for MS=120
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
ertPS with BWA (Pro)
ertPS without BWA (Pro)
5
10
15
AMP
20
30
Avg. pkt.dropping rate
Avg. pkt.dropping rate for MS=120
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
rtPS with BWA (Pro)
rtPS without BWA (Pro)
5
10
15
20
AMP
Avg. pkt. dropping rate
Avg. pkt. dropping rate for MS=120
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
nrtPS with BWA (Pro)
nrtPS without BWA (Pro)
5
10
15
20
AMP
Avg. pkt. dropping rate
Avg. pkt. dropping rate for MS=120
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
BE with BWA (Pro)
BE without BWA (Pro)
5
10
15
AMP
20
31
Avg. pkt. dropping rate
Avg. pkt. dropping rate for MS=200
0.6
0.5
0.4
UGS with BWA (Pro)
0.3
UGS without BWA (Pro)
0.2
0.1
0
5
10
15
20
AMP
Scenario 4: MS=200
Avg. pkt. dropping rate
Avg. pkt. dropping rate for MS=200
0.6
0.5
0.4
ertPS with BWA (Pro)
0.3
ertPS without BWA (Pro)
0.2
0.1
0
5
10
15
AMP
20
32
Avg. pkt. dropping rate
Avg. pkt. dropping rate for MS=200
0.6
0.5
0.4
rtPS with BWA (Pro)
0.3
rtPS without BWA (Pro)
0.2
0.1
0
5
10
15
20
AMP
Avg. pkt. dropping rate
Avg. pkt. dropping rate for MS=200
0.6
0.5
0.4
nrtPS with BWA (Pro)
0.3
nrtPS without BWA (Pro)
0.2
0.1
0
5
10
15
20
AMP
Avg. pkt. dropping rate
Avg. pkt. dropping rate for MS=200
0.6
0.5
0.4
BE with BWA (Pro)
0.3
BE without BWA (Pro)
0.2
0.1
0
5
10
15
AMP
20
33
Avg. pkt. delay (ms)
Avg. pkt. delay for MS=200
70
65
60
55
50
45
40
35
30
25
20
UGS with BWA (Pro)
UGS without BWA (Pro)
5
10
15
20
AMP
Avg. pkt. delay (ms)
Avg. pkt. delay for MS=200
70
65
60
55
50
45
40
35
30
25
20
ertPS with BWA (Pro)
ertPS without BWA (Pro)
5
10
15
AMP
20
34
Avg. pkt. delay (ms)
Avg. pkt. delay for MS=200
70
65
60
55
50
45
40
35
30
25
20
rtPS with BWA (Pro)
rtPS without BWA (Pro)
5
10
15
20
AMP
Avg. pkt. delay (ms)
Avg. pkt. delay for MS=200
70
65
60
55
50
45
40
35
30
25
20
nrtPS with BWA (Pro)
nrtPS without BWA (Pro)
5
10
15
20
AMP
Avg. pkt. delay (ms)
Avg. pkt. delay for MS=200
70
65
60
55
50
45
40
35
30
25
20
BE with BWA (Pro)
BE without BWA (Pro)
5
10
15
AMP
20
35
Outlines
 Backgrounds
and motivation
 Spec-defined power saving mechanism
 Proposed power saving mechanism
 Performance evaluation
 Conclusion and future works
36
Conclusion
Although the power saving based WiMAX scheme
introduces packet dropping and delay than the traditional
one, the power consumption is decreased
 Our proposed WiMAX scheme prolongs sleep duration
and decreases ON-OFF alternation for MSs
 Proposed BWA + scheduling is better than NO BWA +
scheduling in both packet dropping rate and delay
 The current judgment for parameters can make MSs
sleep at most 15 frames with tolerable packet dropping
rate

37
Future works
 Further
judgment for parameters to make the
packet dropping rate tolerable when AMP=20
 Use the LMS algorithm as substitute for BWA
 Implement the other two packet scheduling
algorithm for alternative
38
Thank you!!
 Your
suggestion is my treasure!!
39