Performance Evaluation of VoIP and TCP over wireless networks
Download
Report
Transcript Performance Evaluation of VoIP and TCP over wireless networks
Performance Evaluation of VoIP and
TCP over wireless networks
Presented By : Ala’ Khalifeh
Friday, March,17,2006
University of California-Irvine
Note:
1) This presentation is a brief description for the papers I have
read and mentioned in the reference page.
2)I have sent other presentation which I have prepared but did
not present.
The Big Picture
Industry Approach
The
infrastructure of the Next generation Internet is
expected to provide an interface to the wireless services
through well-defined standards and schemes.
The
Third Generation Partnership Project (3GPP/3GPP2)
has suggested the the IP Multimedia Subsystem (IMS)
standard .
IMS
defines a generic architecture for offering Voice over IP
(VoIP) and multimedia services over the wireless
environment and define the architecture of integrating the
wireless network with the Internet.
For operators, IMS takes the concept of layered
architecture one step further by defining a horizontal
architecture. It is well integrated with existing voice and
data networks, while adopting many of the key benefits
of the IT domain.
IMS Architecture
This is on the service providers side, on the the userend sides, many mobile companies are designing their
new mobile sets to make them capable of conducting
VoIP calls over wireless network and to support many
multimedia services on the mobile sets.
One important thing to consider is the performance of
TCP on carrying data over the wireless environment,on
one hand, and the performance of the real-time
services such as VoIP that are carried on top on UDP
on the other hand, and finally the co-existent between
the two .
The Roadmap of my research :
VoIP Technology
TCP Over
Wireless
Problems
TCP Behaviour
Wireless Tech.
VoIP over
Wireless
Problems
Suggested Solutions
The Network Architecture
VoIP Requirements
• Definition: Using the Internet to carry phone
conversations, known as Internet telephony or voice
over IP (VoIP).
• Quality of Service Requirements
– End-to-end delay, or called latency.(150 and 400 ms )
– Packet loss
• uses UDP as transport protocol.
• Packet loss Not to exceed 10-15 %)
– called jitter delay
TCP Operation
• TCP is the main transport protocol designed for the
wired environment
• Reliable transmission (Require Ack).
• Uses mechanisms for congestion control and detection
(Congestion Window).
• TCP uses an Algorithm called AIMD (Additive Increase,
Multiplicative Decrease) for changing its window size.
• The congestion window climbs exponentially fast
during slow start and hits the threshold .
• The congestion window then climbs linearly until loss
occurs, two scenarios might take place
– The transmitor does not receive an ACK within the
time-out period
– The transmitor receive duplicated ACK (3
duplicated ACK ).(ON BOARD ILLUSTRATION)
TCP Operation
• Which is more critical ?
– Time out or receiving three-Duplicated Ack?
Time out indicates
Real congestion
The Congestion
Window is set to 1
Three Duplicated
Ack (The network is
not congested)
The Congestion
Window is
Halved
TCP Congestion Window
Wireless Networks
• 802.11 Protocols Standards
– IEEE 802 Network Technology Family Tree 802.11
is a member of the IEEE 802 family, which is a
series of specifications for local area network (LAN)
technologies.
WiFi Network Architecture
Accessing The Network
CSMA/CD: carrier sensing, deferral as in CSMA
– collisions detected within short time
– colliding transmissions aborted, reducing channel
wastage
• collision detection:
– easy in wired LANs: measure signal strengths,
compare transmitted, received signals
– difficult in wireless LANs: receiver shut off while
transmitting
• human analogy: the polite conversationalist
• Why no collision detection in Wireless?
– difficult to receive (sense collisions) when
transmitting due to weak received signals (fading)
– can’t sense all collisions in any case: hidden
terminal, fading
• Goal: avoid collisions: CSMA/C(collision)A(voidance)
Accessing The Network
RTS/CTS
idea: allow sender to “reserve” channel rather than random
access of data frames: avoid collisions of long data frames
• sender first transmits small request-to-send (RTS) packets
to AP using CSMA
– RTSs may still collide with each other (but they’re short)
• AP broadcasts clear-to-send CTS in response to RTS
• CTS heard by all nodes
– sender transmits data frame
– other stations defer transmissions
Avoid data frame
collisions
completely using
small
reservation packets
TCP over Wireless media
• TCP considers the loss of packets as a signal of network
congestion and reduces its window
• Consequently This results in severe throughput
deterioration when packets are lost for other reasons than
congestion.
• Non congestion losses are mostly caused by transmission
errors in the wireless environment.
• The solutions proposed to this problem can be divided into
two main categories:
1)Hiding the lossy parts of the
Internet so that only congestion
losses are detected at the source
2)Enhancing TCP with some
mechanisms to help it to
distinguish between
different types of losses
1) Hiding non-congestion losses
• Two Proposed Solutions
Implementing a link-layer
mechanism such as automatic
repeat request (ARQ) protocol
+ Hide
wireless
losses from
TCP by
retransmitting
lost packets
- Possibility
of competing
retransmissions
between TCP
and link layer.
Forward Error Correction(FEC)
sending some redundant
information to rebuild the
corrupted part of the
packet
+ No
retransmission
for the
corrupted
packets
- redundant
bits are
waste of
bandwidth
if no errors
2) Enhancing the TCP
(The Snoop Protocol)
• These solutions try to improve the link quality by
retransmitting packets at the TCP level rather than at
the link level.
• Snoop Protocol:One of the most efficient improvement
to the TCP protocol.
• A TCP agent in the router at the input of the lossy link
keeps a copy of every data packet.
• It discards this copy when it sees the ACK of the
packet
• A packet is retransmitted locally when three duplicate
ACKs are received or when a local Timeout expires.
• This local Timeout is set of course to a value less than
that of the source.
2) Enhancing the TCP
(The Explicit Loss Notification (ELN))
• Here we inform the source explicitly of the occurrence
of a non-congestion loss via an Explicit Loss
Notification (ELN signal).
•
The source reacts by retransmitting the lost packet
without reducing its window size.
• The difficulty with such solution is that a packet
corrupted at the link level is discarded before reaching
TCP and then it is difficult to get this information.
2) Enhancing the TCP
(Congestion Detection and Avoidance)
• Here With some additional mechanisms in the
network or at the source the congestion is detected
and the throughput is reduced before the overflow of
network buffers
• Examples of this improvement can be found in the
Vegas version of TCP and the Explicit Congestion
Notification (ECN) proposals.
• If all the sources, receivers and routers are compliant
according to Vegas or ECN
• Congestion losses will considerably decrease
• The remaining losses could be considered as mostly
caused by problems other than congestion (
transmission losses) which are retransmitted without
window reduction.
VoIP over 802.11 (Vo802.11)
Performance Requirements
Two major technical problems that stand in the
way are
– 1) low VoIP capacity and high access delay in
WLAN.
– 2)unacceptable VoIP performance in the presence
of coexisting traffic from other applications
VoIP over 802.11 (Vo802.11)
• IEEE 802.11b, which can support data rates up to
11Mbps.
• A VoIP stream typically requires less than 10Kbps.
• Ideally, the number of simultaneous VoIP streams that
can be supported by an 802.11b WLAN is around
11M/10K = 1100, which corresponds to about 550 VoIP
sessions, each with two VoIP streams.
• However, it turns out that the current WLAN can only
support no more than a few VoIP sessions. For
example, if GSM 6.10 codec is used, the maximum
number of VoIP sessions that can be supported is 12.
VoIP over 802.11 (Vo802.11)
• This result is mainly due to the added packet-header
overheads as the short VoIP packets traverse the
various layers of the standard protocol stack, as well
as the inefficiency inherent in the WLAN MAC protocol,
as explained below.
• A typical VoIP packet at the IP layer consists of 40-byte
IP/UDP/RTP headers and a payload ranging from 10
to 30 bytes, depending on the codec used.
• So the efficiency at the IP layer for VoIP is already less
than 50%.
• At the 802.11 MAC/PHY layers, the drop of efficiency
is much worse.
• Consider a VoIP packet with 30-byte payload. The
transmission time for it at 11 Mbps is 30 * 8 / 11 = 22 μ
sec
VoIP over 802.11 (Vo802.11)
• The transmission time for the 40-byte IP/UDP/RTP
header is 40 * 8 / 11 = 29 μ sec .
•
However, the 802.11 MAC/PHY layers have additional
overhead of more than 800 μ sec , attributed to the
physical preamble, MAC header, MAC backoff time,
MAC acknowledgment, and inter transmission times of
packets and acknowledgments
• As a result, the overall efficiency drops to less than 3%
VoIP over 802.11 (Vo802.11)
• Even when the number of VoIP sessions is limited to
just half of the capacity in an 802.11b WLAN,
interference from just one TCP connection will cause
unacceptably large increases in the delay and packet
loss rate of VoIP traffic.
• A WLAN can support voice if you implement the
system with high performance and quality of service
(QoS) in mind.
• The 802.11e group is currently working on a QoS
upgrade 802.11e will prioritize traffic on the network,
making data give way to voice packets.
Supporting QoS : 802.11e
• Legacy 801.11 MAC protocol operates on “ Distributed
Coordination Function (DCF).”Listen before talk
Principle “.
• To reduce the probability of collisions, the DCF applies
a collision avoidance (CA) mechanism.
• 1) After detecting the medium as idle for a minimum
duration called DCF interframe space (DIFS)
• 2) an additional random time called backoff time. A
station initiates its transmission only if the medium
remains idle for this additional random time.
• Stations select the number of slots at random out of an
interval between (0 and contention window (CW)).
Supporting QoS : 802.11e
• This results in no mechanism to differentiate between
stations and their traffic, and therefore no QoS support
in the DCF.
Supporting QoS : 802.11e
• 802.11e introduces the enhanced dis- tributed channel
access (EDCA),
• The QoS support in EDCA is provided by the
introduction of access categories (ACs) and multiple
independent backoff entities.
• The ACs are labelled according to their target
application, i.e., AC_VO (voice), AC_VI (video), AC_BE
(best effort), and AC_BK (background).
• Here also arbitration interframe space (AIFS[AC]) is
used instead of DIFS, which is used by legacy
stations.
• The smaller the AIFS[AC], the higher the medium
access priority.
Supporting QoS : 802.11e
• The minimum size of the contention window,
CWmin[AC], is another parameter dependent on the
AC .
• The smaller the CWmin[AC], the higher the priority in
medium access.
References
• Anthony C.H. Ng,,David Malone,Douglas J. Leith, “Experimental
Evaluation of TCP Performance and Fairness in an 802.11e Testbed”,SIGCOMM’05 Workshops, August 22–26, 2005.
• Wei wang,soung chang liew,victor o. k. li, “solutions to
performance problems in voip over 802.11 wireless lan 1 , IEEE
transactions on vehicular technology, vol. 54, no. 1, january 2005.
• Mirko Franceschinis, Marco Mellia, Michela Meo, and Maurizio
Munafo. Measuring TCP over WiFi: A real case. Proceedings of the First
Workshop on Wireless Network Measurements, Trentino, Italy, April
2005.
• Sunghyun choi, stefan mangold, ” analysis of IEEE 802.11e for
QoS support in wireless LANs “, IEEE wireless communications
december 2003.
• Chadi Barakat, Eitan Altman, and Walid Dabbous, “On TCP
Performance in a Heterogeneous Network: A Survey”, IEEE
Communications Magazine January 2000
• George Xylomenos and George C. Polyzos, TCP Performance
Issues over Wireless Links, IEEE Communications Magazine.
April 2001
References
• Computer Networking: A Top Down Approach Featuring the
Internet, 3rd edition.Jim Kurose, Keith Ross Addison-Wesley, July
2004.
• 802.11® Wireless Networks: The Definitive Guide By Matthew
Gast,Publisher : O'Reilly
• Wireless LANs, Second Edition ,Copyright 2002 by Sams
Publishing.
• Cisco Building a Wireless LAN ,Eric Ouellet ,Robert Padjen
,Arthur P fund Ron Fuller Technical Editor ,Tim Blankenship
Technical Editor
• Wi-Fi Protected Access Data Encryption and Integrity Advanced
Encryption Standard, Technical report, by Published: July 29,
2005, Microsoft
References
• S. Vangala and Miguel A. Labrador. Performance of TCP over
Wireless Networks with the Snoop Protocol. In Proceedings of
IEEE LCN, pages 600--601, Tampa, FL, November 2002.
• http://www.wi-fiplanet.com/tutorials/article.php/2171721