Transcript Voice Traffic Performance over Wireless LAN using the Point

Voice Traffic Performance over
Wireless LAN using the Point
Coordination Function
Wei Wei
Supervisor: Prof. Sven-Gustav Häggman
Instructor: Researcher Michael Hall
Helsinki University of Technology
Communications Laboratory
April, 2004
Contents
•
•
•
•
•
•
•
Background
Objectives
Introduction to WLAN
Simulation
Results
Conclusions
Future work
Why WLAN?
• Mobility
- It brings increased efficiency and productivity.
• Flexibility
- Fast and easy deployment.
- Can be set up where the wired networks are
imposible or difficult to reach.
Voice over WLAN (1)
• Nowadays, IEEE 802.11 WLAN standard
is being accepted widely and rapidly for
many different environments.
• Mainly, WLAN is used for Internet based
services like web browsing, email, and file
transfers.
Voice over WLAN (2)
• However, demand for supporting real-time
traffic applications such as voice over
WLAN has been increasing.
• To meet this need, IEEE 802.11 standard
defines an optional medium access
protocol, Point Coordination Function
(PCF).
Objectives
• To implement the basic PCF algorithm in a timedriven simulation program written in C language.
• To measure some metrics such as throughput,
delay, frame loss rate, etc.
• To evaluate the voice traffic performance in
WLAN using PCF to investigate if PCF is
capable of the real-time applications such as
voice service.
Network architecture (1)
Network architecture (2)
• Basically, WLAN network consists of four
components: Distribution System, Access Point,
Mobile Station, and wireless medium.
• Distribution System (DS):
- A backbone network that connects several
access points or Basic Service Sets.
- Wired or wireless, implemented independently.
- In general, Ethernet is used as the backbone
network technology.
Network architecture (3)
• Access Point (AP):
- Connected to the DS, wireless-to-wired
bridging function.
• Mobile Station (MS):
- In general, it’s referred to laptop computer.
• Wireless medium:
- Frequency Hopping, Direct Sequence Spread
Spectrum, Infra-red.
Network architecture (4)
• Basic Service Set (BSS):
- It consists of a group of stations that are
under control of DCF or PCF.
• Extended Service Set (ESS):
- It consists of several BSSs via DS.
- Provides larger network coverage area.
Network architecture (5)
• IEEE 802.11 defines two operation modes:
Ad-hoc mode and Infrastructure mode.
• Ad-hoc mode:
- A set of 802.11 wireless stations
communicate directly with each other,
without using access point.
- Also called Independent Basic Service
Set (IBSS).
Network architecture (6)
• Infrastructure mode:
- The network consists of at least one
access point and a set of mobile stations.
- AP bridges the wireless traffic to a wired
Ethernet or the Internet.
- AP can be compared with a base station
used in a celluar network.
IEEE 802.11 MAC layer
• IEEE 802.11 defines two medium access
methods: the mandatory Distributed
Coordination Function (DCF) for non-realtime applications, and the optional Point
Coordination Function (PCF) for real-time
applications.
DCF
• Basic access method of IEEE 802.11, using
Carrier Sense Multiple Access with
Collision Avoidance (CSMA/CA) to access
to the shared medium.
• Backoff before transmission, provide fair
access to the medium.
• No QoS guarantees, best effort.
PCF
• Optional access method, resides on top of
DCF.
• To support real-time applications.
• Centralized control.
• Polling based access mechanism.
Coexistence of DCF and PCF
Taken from IEEE 802.11 standard
Inter-Frame Space (IFS)
•
•
•
•
•
•
Basically 3 different IFSs.
Short IFS (SIFS)
PCF IFS (PIFS)
DCF IFS (DIFS)
SIFS < PIFS < DIFS
IFS determines priority:
- After a SIFS, only polled MS can send
- After a PIFS, only AP can send (PCF control)
- After a DIFS, every station can send according
to CSMA/CA (DCF)
PCF operation (1)
• The time on the medium is divided into two parts:
Contention-Free Period (CFP) controlled by PCF and
Contention Period (CP) controlled by DCF.
PCF operation (2)
• During a CFP, at least 2 maximum size
frames transmitted.
• During a CP, at least 1 maximum size
frame transmitted, including RTS/CTS and
ACK.
PCF operation (3)
PCF polling scheme (1)
• A poll list is created when the MSs
supporting real-time service negotiate with
Point Coordinator (PC) during the
association procedure.
• The MSs are put on the poll list in order.
• The poll list gives the highest privilege to
PCF supported MSs.
PCF polling scheme (2)
• The polling scheme is based on Round-Robin
scheduler recommended by IEEE 802.11
standard.
• Only the polled MS can transmit a frame.
• During one CFP, the MS can be polled once.
• If the CFP terminates before all MSs on the poll
list are polled, the poll list will resume at the next
MS in the following CFP.
• The CFP may terminate befor time, if all MSs on
the poll list have no data to send.
• Data frame, ACK, and poll combined to improve
efficiency.
Simulation scenario
• A single BSS in an infrastructure network configuration.
Simulation model assumptions
(1)
• Only use voice traffic during CFP, not consider
data traffic during CP.
• RTP/UDP/IP/MAC/PHY, this adds an overall
overhead of 78 bytes to every voice packet.
• G.711 PCM voice codec used, fixed traffic
interval 20ms or 40ms, 160bytes or 320bytes
payload, respectively.
• Buffer size = 1.
Simulation model assumptions
(2)
• Power saving mode is neglected.
• Foreshortened CFP is neglected.
• Fragmentation/Defragmentation is
neglected.
• Broadcast/Multicast frames not considered.
• Mobility, multipath interference, and
hidden-node problem are not considered.
• Basic rate used: 11 Mbps.
Functions included in simulation
(1)
• One access point and specific number of
VoIP stations
• Voice connections: bi-directional
deterministic stream of frames with
calculated duration and inter-frame
interval, PCM over RTP over UDP over IP
over LLC over MAC over PHY assumed
• SIFS and PIFS times
Functions included in simulation
(2)
• Acknowledgement, beacon, CF-poll, and
CF-end frames
• Piggybacking of Ack and CF-poll
information
• Random generation of erroneous frames
• Recording of simulation data
Simulation parameters
Channel rate
Channel frame
error rate (CFER)
Voice payload
Slot time
SIFS
PIFS
DIFS
11 Mbps
0.03
160/320 bytes
20 s
10 s
30 s
50 s
Metrics
•
•
•
•
•
Superframe size
Maximum number of VoIP MS
Throughput
Frame loss rate
Access delay
Results: superframe size
• Normalized throughput for different SF using
160-byte payload
Normalized throughput
160-byte payload
traffic interval=20ms, CFER=0.03
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
SF10
SF15
SF20
SF25
SF30
5
10
15
20
25
30
Number of VoIP MS
35
40
Results: superframe size
• Normalized throughput for different SF using
320-byte payload
Normalized throughput
320-byte payload
traffic interval = 40ms, CFER = 0.03
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
SF30
SF35
SF40
SF45
SF50
5
10
15
20 25
30 35
40
Number of VoIP MS
45
50
Results: max. number of VoIP
MS for 160-byte payload
Normalized throughput
160-byte payload
traffic interval = 20ms, CFER = 0.03
1
0.95
0.9
0.85
0.8
0.75
0.7
0.65
0.6
0.55
0.5
SF20
0
5
10
15 20 25 30 35
Number of VoIP MS
40
45
Results: max. number of VoIP
MS for 320-byte payload
Normalized throughput
320-byte payload
traffic interval = 40ms, CFER = 0.03
1
0.95
0.9
0.85
0.8
0.75
0.7
0.65
0.6
0.55
0.5
SF40
0
5 10 15 20 25 30 35 40 45 50 55
Number of VoIP MS
Results: capacity
Capacity (Mbps)
SF20 160-byte payload vs. SF40 320-byte payload
CFER = 0.03
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
SF20
SF40
0
5
10 15 20 25 30 35 40 45 50 55
Number of VoIP MS
Results: frame loss rate
SF20 160-byte payload vs. SF40 320-byte payload
Frame loss rate
0.6
0.5
0.4
SF20
SF40
0.3
0.2
0.1
0
0
5
10
15 20 25 30 35
Number of VoIP MS
40
45
Results: average access delay for
different SF using 160-byte payload
160-byte payload
traffic interval = 20ms, CFER = 0.03
Access delay (ms)
12
SF10
SF15
SF20
SF25
SF30
SF35
SF40
10
8
6
4
2
0
5
10
15
20
25
30
Number of VoIP MS
35
40
Results: average access delay for
different SF using 320-byte payload
320-byte payload
traffic interval = 40ms, CFER = 0.03
Access delay (ms)
25
20
SF30
SF35
SF40
SF45
SF50
15
10
5
0
5
10
15
20 25 30 35 40
Number of VoIP MS
45
50
Results: comparison of average access
delay btw. 160 and 320-byte payload
Access delay (ms)
SF20 160-byte payload vs. SF40 320-byte payload
CFER = 0.03
20
18
16
14
12
10
8
6
4
2
0
SF20
SF40
0
5
10 15 20 25 30 35 40 45 50 55
Number of VoIP MS
Cumulative percentage
of packets
Results: cumulative delay
distribution for 160-byte payload
Delay distribution for 160-byte payload
CFER = 0.03
1
0.8
SF10,
SF10,
SF20,
SF20,
0.6
0.4
0.2
0
0
10
20
30
Delay (ms)
40
50
V=10
V=20
V=10
V=20
Cumulative percentage
of packets
Results: cumulative delay
distribution for 320-byte payload
Delay distribution for 320-byte payload
CFER = 0.03
1
0.8
SF20,
SF20,
SF40,
SF40,
0.6
0.4
0.2
0
0
10
20
30
Delay (ms)
40
50
V=10
V=20
V=10
V=20
Conclusions
• The proper superframe size should be
approximately similar to the traffic interval,
which results in good performance.
• Longer payload provides higher normalized
throughput and lower frame loss rate, but longer
access delay.
• Maximum number of VoIP MS: for 160-byte
payload, 21; for 320-byte payload, 36.
• When the number of VoIP MS increases,
performance degrades dramatically. PCF
provides limited QoS.
Future works
• Perform an authentic evaluation in a WLAN
- Assumptions
- Realistic traffic model
• PCF problems
- unpredictable Beacon frame delay resulting in
shortened CFP
- unknown transmission time of polled stations
making it difficult for PC to predict and control
the polling scheldule for the remainder of CFP
• IEEE 802.11e introduced EDCF and HCF to
support QoS
Q&A
Thank you for your attention!