Towards the Quality of Service for VoIP Traffic in IEEE 802.11 Wireless Networks - PowerPoint PPT Presentation

1 / 59
About This Presentation
Title:

Towards the Quality of Service for VoIP Traffic in IEEE 802.11 Wireless Networks

Description:

DD-EE-FF. 1.1.1.2. Example 2: Malicious user detection. Handoff. 22. Passive DAD. AUC. DHCP server ... DD-EE-FF. 1.1.1.2. AA-BB-CC. 1.1.1.1. FORCE RENEW. IP: ... – PowerPoint PPT presentation

Number of Views:64
Avg rating:3.0/5.0
Slides: 60
Provided by: sangh1
Category:

less

Transcript and Presenter's Notes

Title: Towards the Quality of Service for VoIP Traffic in IEEE 802.11 Wireless Networks


1
Towards the Quality of Service for VoIP Traffic
in IEEE 802.11 Wireless Networks
  • Sangho Shin
  • PhD candidate
  • Computer Science
  • Columbia University

2
VoIP over WLANs
3
Problems on VoIP in WLANs
  • User mobility Handoff

4
Problems on VoIP in WLANs
  • User mobility Handoff
  • Capacity

5
Problems on VoIP in WLANs
  • User mobility Handoff
  • Capacity
  • Call admission

6
QoS problems on VoIP in WLANs
QoS
7
Outline
  • Layer 2 handoff
  • Layer 3 handoff
  • pDAD
  • Measurement
  • APC
  • DPCF

Handoff
Capacity
QoS
Call Admission Control
  • QP-CAT

8
Handoff
Handoff
  • Layer 2 handoff
  • Handoff between two APs
  • Layer 3 handoff
  • Handoff between two subnets

9
Handoff
Selective Scanning Caching
A layer 2 handoff algorithm to minimize the
scanning time
Sangho Shin, Andrea G. Forte, Anshuman Singh
Rawat, and Henning Schulzrinne. Reducing MAC
layer handoff latency in IEEE 802.11 wireless
LANs. ACM MobiWac 2004
10
Layer 2 Handoff
Handoff
  • Standard Layer 2 handoff procedure

500ms
2ms
2ms
11
Fast L2 Handoff
Handoff
  • Selective Scanning
  • Scan the channels that APs are most likely
    installed on
  • Previously scanned APs channels
  • Non-overlapping channels
  • Do not scan the current channel

Channel mask
6
1,11
11
Channel mask
11
1
12
Fast L2 Handoff
Handoff
  • Selective Scanning
  • Scan the channels that APs are most likely
    installed on
  • Previously scanned channels
  • Non-overlapping channels
  • Do not scan the current channel
  • Caching
  • Locality
  • Store the scanned AP data to a cache
  • Perform handoff without scanning

Channel mask
13
Fast L2 Handoff
Handoff
AP2
  • Caching
  • Locality
  • Store the scanned AP data to a cache
  • Perform handoff without scanning

AP4
6
Cache
11
AP1
AP3
11
1
14
Fast L2 Handoff
Handoff
  • Implementation
  • HostAP driver Prism2 chipset
  • Requires changes only in the client wireless
    driver
  • Experimental results

Experiments in 802.11b WLANs
15
Layer 3 Handoff
Handoff
  • L3 handoff
  • Occurs when the subnet changes due to L2 handoff
  • Requires a new IP address
  • Problem of L3 handoff
  • Detection of subnet change
  • Long acquisition of a new IP address

16
Seamless L3 handoff
Handoff
  • Goal
  • Do not modify any standard or infrastructure
  • Fast subnet change detection
  • Subnet has each DHCP server or relay agent
  • Send a bogus DHCP request in the new subnet
  • Temp_IP
  • Scan unused IP address actively
  • Send APR requests to a range of IP addresses
  • Reduced the total L3 handoff to 180ms

Andrea Forte, Sangho Shin, and Henning
Schulzrinne. Improving Layer 3 Handoff Delay in
IEEE 802.11 Wireless Networks. IEEE WICON, Aug
2006.
17
Handoff
pDAD Passive Duplicate Address Detection
A real time DAD mechanism in the DHCP server
Sangho Shin, Andrea Forte, and Henning
Schulzrinne. Passive Duplicate Address Detection
for Dynamic Host Configuration Protocol (DHCP).
IEEE GLOBECOM, 2006.
18
Passive DAD
Handoff
  • Server side solution for seamless L3 handoff
  • Eliminate the DAD procedure in the DHCP server
    when assigning new IP addresses

Monitor the network
Collect IP addresses
Update IP list
Respond quickly to the request
160.123.234.31 160.123.231.32 160.123.235.35 160.1
23.232.36 160.123.238.38
19
Passive DAD
Handoff
  • Architecture

Address Usage Collector (AUC)
DHCP server
Lease table
IP
MAC
IP
MAC
Expire
Router
20
Passive DAD
Handoff
  • Example 1 IP address collection

DHCP server
IP
MAC
Expire
AUC
Lease table
IP1.1.1.1 MACAA-BB-CC
IP
MAC
Web server
Router
ARP query
IP1.1.1.1
IP1.1.1.1
MACAA-BB-CC
21
Passive DAD
Handoff
  • Example 2 Malicious user detection

DHCP server
IP
MAC
Expire
AUC
Lease table
IP1.1.1.2 MACDD-EE-FF
IP
MAC
Web server
Bad IP table
IP
MAC
Router
ARP query
IP1.1.1.1
MACAA-BB-CC
IP1.1.1.2
MACDD-EE-FF
22
Passive DAD
Handoff
  • Example 3 IP collision detection

DHCP server
IP
MAC
Expire
AUC
Lease table
IP1.1.1.1 MAC00-00-00
IP
MAC
Web server
Bad IP table
IP
MAC
Router
Block 00-00-00
Forward HTTP traffic
FORCE RENEW IP1.1.1.3
ARP query
IP1.1.1.1
MACAA-BB-CC
IP1.1.1.2
IP1.1.1.1
MACDD-EE-FF
MAC00-00-00
23
Outline
Handoff
Capacity
  • Layer 2 handoff
  • Layer 3 handoff
  • pDAD
  • Measurement
  • APC
  • DPCF

QoS
  • QP-CAT

24
VoIP Capacity
Capacity
  • Definition
  • The number of VoIP calls whose uplink and
    downlink delay are less than 60ms

Experimental result 64kb/s 20ms PI 802.11b 11Mb/s
25
VoIP Capacity
Capacity
  • Experimental measurement in the ORBIT test-bed
  • ORBIT test-bed (Rutgers Univ. NJ)
  • Open-Access Research Test-bed for Next-Generation
    Wireless Networks

Sangho Shin and Henning Schulzrinne. Experimental
measurement of the capacity for VoIP traffic in
IEEE 802.11 Wireless Networks. IEEE INFOCOM, 2007.
26
VoIP Capacity
Capacity
Experimental results in the ORBIT test-bed
Downlink delay
Downlink delay
Uplink delay
Uplink delay
CBR
VBR with 0.39 activity ratio
64kb/s VoIP traffic 20ms packetization
interval 11Mb/s data rate
27
VoIP Capacity
Capacity
  • Factors that affects the VoIP capacity
  • Preamble size
  • ACK data rate
  • 11Mb/s (QualNet) ? 16 calls
  • 2 Mb/s (MadWifi driver, NS-2) ? 15 calls
  • Offset among VoIP packets of other clients
  • Simulator ? Synchronized ? high collision rate
  • Experiments ? Randomized ? lower collision rate
  • ARF (Auto Rate Fallback)
  • Simulator? Fixed rate?15 calls
  • Experiments? ARF enabled by default?14 calls

28
VoIP Capacity
Capacity
  • Factors that affects the experimental results
  • Scanning
  • Scanning related frames delays VoIP packets
  • Simulator ?No scanning
  • Experiments ? Scan APs due to retransmissions
  • Retry limit
  • Long retry limit (4) ? short transmission time,
    high packet loss
  • Short retry limit (7) ? long transmission time,
    low packet loss
  • Network buffer size
  • Buffer size ? ? packet loss ? delay ?
  • Buffer size ? ? packet loss ? delay ?

29
Capacity
DPCF Dynamic Point Coordination Function
An improved polling based PCF MAC protocol
Takehiro Kawata, Sangho Shin, Andrea G. Forte,
and Henning Schuzrinne. Improving The Capacity
for VoIP Traffic in IEEE 802.11 Networks with
Dynamic PCF. IEEE WCNC 2005.
30
Dynamic PCF
Capacity
  • PCF (Point Coordination Function)
  • Polling based media access
  • No contention, no collision
  • Polling overhead
  • No data to transmit ? Unnecessary polls waste
    bandwidth
  • Big overhead, considering the small VoIP packet
    size

Polling overhead
31
Dynamic PCF
  • Dynamic Polling List
  • Keeps the talking nodes only
  • More Data bit
  • Set the More Data bit, then the AP polls the node
    again
  • Synchronization
  • Synchronize the polls with data

32
Simulation results
Capacity
  • VoIP capacity
  • Increased from 30 calls to 37 calls
  • Polls decreased by 50, Null Functions by 90
  • 760 frames / second 7.29 VBR Calls

33
Capacity
APC Adaptive Priority Control
A new packet scheduling algorithm at the AP in DCF
Sangho Shin and Henning Schulzrinne. Balancing
uplink and downlink delay of VoIP traffic in IEEE
802.11 Wireless Networks using Adaptive Priority
Control (APC). ACM QShine 2005.
34
APC
Capacity
  • Big gap between uplink and downlink delay
  • ?Unfair resource distribution between uplink and
    downlink in DCF

Solution ? High priority to AP
How? How much?
35
APC
Capacity
  • How?
  • Contention Free Transmission (CFT)
  • Transmit P packets contention free (w/o backoff)
  • How much? (Optimal P)
  • P ? QAP/QC
  • QAP is the number of packets in the queue of the
    AP
  • QC is the average number of packets in the queue
    of all clients
  • Adapts to instant change of uplink and downlink
    traffic volume

P3
P4
D
D
D
D
D
D
D
U
U
backoff
36
APC
Capacity
  • Example

Downlink volume gt Uplink volume
QAP 12, QC2, P6
QAP 6, QC1, P6
37
APC
Capacity
Simulation results
802.11b 11Mb/s 64kb/s VBR traffic 20ms pkt
intvl 0.39 activity ratio
28 Calls ? 35 calls (25)
38
Outline
Handoff
Capacity
  • Layer 2 handoff
  • Layer 3 handoff
  • pDAD
  • Measurement
  • APC
  • DPCF

QoS
  • QP-CAT

39
CAC
QP-CAT Queue size Prediction using Computation
of Additional Transmissions
A novel call admission control algorithm
Sangho Shin and Henning Schulzrinne. Call
Admission Control in IEEE 802.11 Wireless
Networks using QP-CAT. IEEE INFOCOM 2008.
40
Admission Control using QP-CAT
CAC
  • QP-CAT
  • Metric Queue size of the AP
  • Strong correlation between the queue size of the
    AP and delay
  • Key idea predict the queue size increase of the
    AP due to new VoIP flows, by monitoring the
    current packet transmissions

Correlation between queue size of the AP and
delay (Experimental results with 64kb/s VoIP
calls)
41
QP-CAT
CAC
  • Basic flow of QP-CAT

42
QP-CAT
CAC
  • Computation of Additional Transmission
  • Virtual Collision
  • Deferrals of virtual packets

43
QP-CAT
CAC
Simulation results
16 calls 1 virtual call (predicted by QP-CAT)
17 calls 1 virtual call (predicted by QP-CAT)
17th call is admitted
16 calls (actual)
17 calls (actual)
44
QP-CAT
CAC
  • Experimental results (64kb/s 20ms PI)

11Mb/s
1 node - 2Mb/s
2 nodes - 2Mb/s
3 nodes - 2Mb/s
45
QP-CAT
CAC
  • QP-CATe
  • QP-CAT with 802.11e
  • Emulate the transmission during TXOP

TXOP
D
D
D
TCP
Tc
46
Conclusion
  • Reduced the layer 2 handoff time using Selective
    Scanning and Caching
  • Achieved the seamless layer 3 handoff using Temp
    IP and pDAD
  • Measured the VoIP capacity in wireless networks
    via experiments and identified the factors that
    affect the VoIP capacity
  • Improved the VoIP capacity using DPCF and APC
  • Can perform call admission control with fully
    utilizing the channel bandwidth, using QP-CAT

47
Other research
  • Implementation of SIP Servlet
  • Development of a SIP client in a PDA (SHARP
    Zaurus)
  • Soft Handoff using dual wireless cards
  • Measurement of usage of IEEE 802.11 wireless
    networks in an IETF meeting

48
  • Thank you!

49
VoIP Capacity in IEEE 802.11e
Experimental results using AC_VO and AC_VI
Experimental results with TCP traffic using AC_VO
50
Comparison b/w poll and VoIP frame
  • Poll size
  • 28B (MAC header CRC)
  • Total TX time PHY (128 us) MAC (26 us) 154
    us
  • Data
  • 28B 160B
  • Total TX time PHY (128 us) MAC (26 us) VoIP
    data (116 us) 270 us
  • A Poll 154/270 60 of a VoIP frame

51
Development of SIP VoIP Client
SIP
52
Development of SIP VoIP Client
Prototype
53
Development of SIP VoIP Client
54
Layer 2 Handoff
  • Handoff process

55
APC
Downlink Uplink
Downlink gt Uplink
QAP 12, QC2, P6
QAP 6, QC1, P6
QAP 8, QC2, P4
QAP 4, QC1, P4
Downlink lt Uplink
QAP 4, QC2, P2
QAP 2, QC1, P2
56
Seamless L3 handoff
  • Goal
  • Do not modify any standard or infrastructure
  • Fast subnet change detection
  • Idea Subnet has each DHCP server or relay agent
  • Broadcast a bogus DHCP request
  • The DHCP server responds with DHCP NACK
  • Check the IP address of the DHCP server
  • Extension of L2Cache
  • Stores the subnet ID
  • IP address of DHCP or relay agent
  • Temp IP
  • Idea Unused IPs every 5 IPs used
  • Scan potentially unused IP addresses in the new
    subnet
  • Transmit multiple ARP packets
  • Pick a non-responded IP address as a temporary IP
    address
  • Use it until a new IP address is assigned by the
    DHCP server

57
Seamless L3 handoff
  • Implementation
  • Linux, HostAP driver, SIP client
  • Experiments

Total handoff time
ms
CU WLAN
CS LAN
180
30

Experiments in 802.11b
58
Dynamic PCF
  • Problems of PCF in VBR VoIP traffic
  • Polling during silence periods
  • Synchronization problem
  • Multiple packetization intervals

Silence period
Talking period
Example 64kb/s 20ms PI with 0.39 AR Total waste
9 VBR calls
poll
Null
voice
Node 1 10ms, Node 2 20ms
59
Dynamic PCF
  • Dynamic Polling List
  • Keeps the talking nodes only
  • More Data bit
  • Set the More Data bit, then APs polls the node
    again
  • Synchronization
Write a Comment
User Comments (0)
About PowerShow.com