PPSP Y.Zhang
Internet Draft China Mobile
Intended status: Standard Track N.Zong
HuaweiTech
G.Camarillo
Ericsson
. J.Seng
PPlive
R.Yang
Yale University
Expires: November 2009 May 27, 2009
Problem Statement of P2P Streaming Protocol (PPSP)
draft-zhang-ppsp-problem-statement-02.txt
Status of this Memo
This Internet-Draft is submitted to IETF in full conformance with the
provisions of BCP 78 and BCP 79.
Internet-Drafts are working documents of the Internet Engineering
Task Force (IETF), its areas, and its working groups. Note that
other groups may also distribute working documents as Internet-Drafts.
Internet-Drafts are draft documents valid for a maximum of six months
and may be updated, replaced, or obsoleted by other documents at any
time. It is inappropriate to use Internet-Drafts as reference
material or to cite them other than as "work in progress."
The list of current Internet-Drafts can be accessed at
http://www.ietf.org/ietf/1id-abstracts.txt
The list of Internet-Draft Shadow Directories can be accessed at
http://www.ietf.org/shadow.html
This Internet-Draft will expire on November 27, 2009.
Copyright Notice
Copyright (c) 2009 IETF Trust and the persons identified as the
document authors. All rights reserved.
Zhang Expires November 27, 2009 [Page 1]
Internet-Draft Problem Statement of P2P Streaming Protocol May 2009
This document is subject to BCP 78 and the IETF Trust's Legal
Provisions Relating to IETF Documents in effect on the date of
publication of this document (http://trustee.ietf.org/license-info).
Please review these documents carefully, as they describe your
rights and restrictions with respect to this document.
Abstract
We propose to develop an open peer-to-peer (P2P) streaming protocol
named PPSP. This document describes problem related to PPSP and
outlines considerations that have to be taken in account when
arriving at equitable solutions.
Zhang Expires November 27, 2009 [Page 2]
Internet-Draft Problem Statement of P2P Streaming Protocol May 2009
Table of Contents
Zhang Expires November 27, 2009 [Page 3]
Internet-Draft Problem Statement of P2P Streaming Protocol May 2009
1. Introduction
Streaming traffic is among the fastest growing traffic on the
Internet. In a recent white paper, Cisco predicts that by 2012, 90%
of all Internet traffic will be video [Cisco].
There are two basic architectures for delivering streaming traffic on
the global Internet: the client-server paradigm and the peer to peer
(P2P) paradigm [P2PStreamingSurvey]. A particular advantage of the
P2P paradigm over the client-server paradigm is its scalability. As
an example, PPLive [PPLive], one of the largest P2P streaming vendors,
is able to distribute large-scale, live streaming programs such as
the CCTV Spring Festival Gala to more than 2 million users with only
a handful of servers. CNN[CNN] reported that P2P streaming by
Octoshape played a major role in its distribution of the historical
inauguration address of President Obama. It is well demonstrated in
practice that P2P streaming can deliver videos encoded at a rate of
about 400 Kbps, in the presence of rapid user joins/leaves, with
positive user experiences.
With the preceding technical advantages, P2P streaming is seeing
rapid deployment. Large P2P streaming applications such as PPLive
[PPLive], PPstream [PPstream] and UUSee [UUSee] each has a user base
exceeding 100 millions. P2P streaming traffic is becoming a major
type of Internet traffic in some Internet networks. For example,
according to the statistics of a major Chinese ISP, the traffic
generated by P2P streaming applications exceeded 50% of the total
backbone traffic during peak time in 2008. There were reports that
major video distributors such as Youtube [youtube] and tudou [tudou]
are conducting trials of using P2P streaming as a component of their
delivery infrastructures.
Given the increasing integration of P2P streaming into the global
content delivery infrastructure, the lacking of an open, standard P2P
streaming protocol becomes a major missing component in the Internet
protocol stack. Multiple, similar but proprietary P2P streaming
protocols result in repetitious development efforts and lock-in
effects. More importantly, it leads to substantial difficulties when
integrating P2P streaming as an integral component of a global
content delivery infrastructure. For example, proprietary P2P
streaming protocols do not integrate well with existing cache and
other edge infrastructures.
1.1. Research or Engineering
As [P2PStreamSurvey] identifies, there exist multiple proprietary P2P
streaming systems including PPLive, PPstream, UUsee, Pando, abacast,
Zhang Expires November 27, 2009 [Page 4]
Internet-Draft Problem Statement of P2P Streaming Protocol May 2009
and Coolstreaming. A natural question to ask is whether the
development of P2P streaming is mature and ready for standardization.
We admit that P2P streaming will continue to improve and evolve.
However, our investigation shows that existing P2P streaming systems
are largely converging, sharing similar architecture and signaling
protocols [draft-zhang-ppsp-protocol-comparison-measurement-00]. The
competition of P2P streaming vendors become increasingly on contents.
1.2. Objective and outline
Our objective is to develop a standard P2P streaming protocol that
can operate in both fixed and mobile Internet. The protocol will
serve as an enabling technology, building on the development
experiences of existing P2P streaming protocols. It will integrate
with IETF efforts on distributed resource location, traffic
localization, and streaming control mechanisms. It allows effective
integration with edge infrastructures such as cache and mobile edge
equipment.
This document provides a problem statement for designing PPSP. The
rest of the document is organized as follows. In Section 2, we
introduce terminologies. In Section 3, we identify key use cases
where a standard P2P streaming protocol can facilitate the
integration of P2P streaming into the network infrastructure, and
therefore benefit both the content distributors and the networks. In
Section 4, we describe the main issues to consider when designing
PPSP. In Section 5, we outline the main scope of work.
Zhang Expires November 27, 2009 [Page 5]
Internet-Draft Problem Statement of P2P Streaming Protocol May 2009
2. Definitions
The following terms have special meaning in the definition of thePeer
to Peer Streaming Protocol (PPSP) problem.
Tracker: A dedicated server who is in charge of maintaining peer list
and replying peer request for peer selection.
Peer: A peer refers to a participant in a P2P streaming system who
acts both as ''client'' and ''server''.
Chunk: A chunk is a basic unit of partitioned streaming, which is
used for the purpose of storage and advertising to neighbors what
parts of a movie a peer holds [Sigcomm:P2P streaming].
Bitmap: Bitmap is a table indicating which chunks a peer has.
Peering networks: Two directly connected Internet Service Providers.
Apart from infrastructure and operational costs, peering traffic is
usually free, within the contract of a peering a greement [draft-
marocco-alto-problem-statement-03].
3. Problem Statement Scope
We perceive a number of problems related to scalable and economical
streaming on the Internet. The major issues are the following:
1) The difficulty of an open PPSP due to the existence of many
proprietary and non-interoperable protocols in current p2p
streaming applications;
2) The difficulty of integrating current edge equipments such as
cache and content distribution network (CDN) nodes into P2P
streaming;
3) The difficulty of integrating related protocols into P2P streaming,
like RELOAD,ALTO,RTSP, which is beyond current P2P streaming usage;
4) The lack of a standard solution for scalable and economical
streaming signaling interaction suitable both for fixed internet
and mobile Internet; a related problem is that the difficulty of
deploying p2p streaming in mobile Internet;
The subsections below discuss these problems in more detail.
Zhang Expires November 27, 2009 [Page 6]
Internet-Draft Problem Statement of P2P Streaming Protocol May 2009
3.1. Proprietary protocols and an open PPSP
Currently there exist some P2P streaming systems like PPLive,
PPstream, UUsee, Pando[Pando] and
Coolstreaming[Coolstreaming].Although using similar system
architecture as well as signaling interaction processes[draft-zhang-
ppsp-protocol-comparison-measurement-00], due to their proprietary
protocols, it's hard to develop an open PPSP protocol without
checking all the typical P2P streaming systems, identifying the key
issues and considering all the requirements.
3.2. Integrating cache and CDN into P2P streaming
To make P2P streaming stable and traffic local enough, cache and
other edge infrastructure is a very promising means [draft-marocco-
alto-problem-statement-03].
However there are a few obstacles in deploying P2P caches [HTPT]:
firstly, P2P caching systems are likely to be very complicated.
Unlike web traffic standardized in using HTTP transport through few
dedicated ports like 80, there is no a standard P2P protocol and
every P2P protocol uses its own port. Therefore, P2P caching systems
are forced to take an ad hoc approach by enumerating and handling
every P2P protocol. Another drawback of this ad hoc approach is the
requirement of regular update of the P2P cache engines to handle
newly emerged popular P2P protocols. Secondly, extra, possibly huge,
investment is required for the equipment and facility purchase and
also the administrative cost.
If we can utilize the existing cache and other edge equipments like
CDN nodes, the cost can be heavily reduced. Meanwhile because it's
widely used, the performance of P2P streaming can increase much.
Therefore how to utilize the edge infrastructure is a big issue.
Current web cache or other widely deployed edge equipments like CDN
doesn't support P2P streaming yet[HTPT].
3.3. Integrating existing protocols into P2P streaming
There are several protocols related to p2p streaming having been or
being defined in IETF, including RELAOD,ALTO,RTSP,etc.., where
P2PSIP,ALTO and MMusic are most related WGs. Unfortunately there is
no one protocol above considered by current P2P streaming
applications.
We claim that PPSP is not a stand-alone protocol. It could and should
integrate with the existing related protocols to the largest extent.
Zhang Expires November 27, 2009 [Page 7]
Internet-Draft Problem Statement of P2P Streaming Protocol May 2009
We analyze several protocols PPSP may relate to in detail in section4.
3.4. Mobility and wireless issue
Mobility and wireless becomes a very important feather to support in
future internet [GENI],[FIND]. Some operators have also started
research projects on mobile and wireless Internet. For example, China
Mobile came out with its DSN (distributed services network) strategy
last year to build its mobile Internet [draft-zhang-ppsp-dsn-
introduction-00].
Along with the introduction of mobile and wireless characteristics
into Internet, mobile streaming will become more and more
popular[MobileTV].In Korea the mobilTV subscriber is 17 million
accounting one third of the mobile subscriber. In Italy there are 1
million mobileTV users. In the period of Beijing Olympic games, there
are more than 1 million usage of China mobile's mobileTV.
Most of current mobileTV are developed in client/server model. It's a
natural idea to increase its scalability and decrease the cost by
deploying P2P mobile streaming.
However there are a lot of differences in mobile/wireless Internet
compared with fixed Internet environment. This makes it hard to copy
current P2P streaming protocol in fixed Internet environment.
3.4.1. End to end communication is harder
Unlike fixed Internet, it's difficult to realize end-to-end
communication in mobile Internet. Mobile terminals cannot connect
with each other directly. The connection must be set up by
mobile/wireless access nodes. Therefore mobile terminals are hard to
become peers without the cooperation of mobile/wireless access
equipments. It's obviously a heavy burden for the mobile/wireless
access points.
Zhang Expires November 27, 2009 [Page 8]
Internet-Draft Problem Statement of P2P Streaming Protocol May 2009
+------------------------------------------------------------+
| |
| +-------+ |
| | Cache | |
| +-------+ |
| ^ |
| | |
| Peerlist V |
| +---------------+ +----+ +-----------+ |
| | Tracker |<---->|PEER|<---->| AP |Bottleneck|
| +---------------+ +----+ +-----------+ |
| ^ ^ ^ ^ |
| | | | Low | |
| | | |Bandwidth| |
| V V V V |
| +---+ End-end +---+ +------+ +------+ |
| |PC |<------> |PC | |Mobile|<--->|Mobile| |
| | | works | | |Phone | ?|Phone | |
| +---+ Well +---+ +------+ +------+ |
| |
| |
| |
+------------------------------------------------------------+
Figure 1 Mobile Internet communication
3.4.2. Limited Bandwidth resource
Even if end-to-end communication is ensured, there are still problems
for P2P mobile streaming.As shown in Figure1, the following bandwidth
is limited and the transmission cost is relatively high:
a) between mobile terminals and mobile access nodes;
b) between mobile access nodes;
c) between mobile network to fixed network.
It raises some requirements for PPSP:
1) The overhead of PPSP cannot be much. Because PPSP is a signaling
protocol, the overhead refers to the ratio of signaling traffic
with respect to overall streaming traffic. Different solutions
have different overhead. [Computer Networks-Traffic]analyzes four
P2P streaming overhead, namely PPLive,PPStream,SOPCast and
TVAnts.There is a tradeoff consideration in PPSP. We know that
frequent exchange in bitmap may cause a big overhead but it also
Zhang Expires November 27, 2009 [Page 9]
Internet-Draft Problem Statement of P2P Streaming Protocol May 2009
enables a small peer cache to keep the smooth play of the program.
On one hand, the overhead should be minimum to reduce the traffic
and increase the efficiency so the bitmap exchange frequency
should be small. On the other hand, a mobile terminal has usually
small cache size so the bitmap exchange frequency should be large.
How to solve the conflict and balance between the cache size and
bitmap exchange frequency is a problem PPSP may face with.
2) Cross-domain traffic must be reduced. For most mobile Internet
domains, their network scale is rather smaller compared with the
Tier1 and Tier2 peering networks. Currently peering networks are
free from the cross-traffic. However the low-tier ISPs have to pay
for dual-directional traffic fee even it's a traffic initiated
from higher tier ISPs. It makes mobile Internet provider worse in
case P2P streaming applications generate great traffic with a
low-cross-domain bandwidth.
3.4.3. Other difference
Mobility, low battery and capability of mobile terminals make us
considering more factors in peer selection to ensure smooth streaming.
A new protocol might therefore allow more information to report for
trackers to do peer selection.
Zhang Expires November 27, 2009 [Page 10]
Internet-Draft Problem Statement of P2P Streaming Protocol May 2009
4. Design Issues
This section introduces key issues when designing PPSP.
4.1. Architecture Choice
There are multiple proposed p2p streaming solutions. Some are used in
practice and while others are popular in theory. The basic idea of
P2P streaming is to partition streaming into chunks and peers are
used in relaying chunk transmission. The solutions can be categorized
by tree-based and mesh-based [Survey]. The former is ''pushing'' chunks
to the audience and the latter is enabling the audience to ''pulls''
the desired chunks from the peers who has.
In the tree-based p2p streaming, tree construction and maintenance
can be done in either a centralized or a distributed fashion [Survey].
In the mesh-based p2p streaming, no specific topology maintenance is
needed. The task is how to locate and retrieve real-time data from
multiple sources with different chunks for a streaming. In order to
realize this, each peer has to actively locate cache and exchange
chunks from other peers by the guidance of the tracker.
The biggest challenge in tree-based p2p streaming is two-fold: First,
the depth of the tree affects the maximum delay the system has;
Second, tree-based streaming still cannot recovery fast enough to
handle frequent peer churn. Therefore they are hardly used in
practice. On the contrary mesh-based p2p streaming is widely used to
overcome these drawbacks for real deployment, like pplive, ppstream,
coolstreaming, etc,..
4.2. Integration with existing protocols
PPSP will not be a stand-alone protocol. A major advantage of a
standard protocol is that we can explicit consider its interactions
with other protocols, and design for an integrated system.
In particular, we identify that PPSP will interact with the following
protocols: RELAOD, ALTO, and RTSP, where P2PSIP, ALTO and MMusic are
the most related WGs.
4.2.1. Integration with RELOAD
RELOAD defined by P2PSIP deals with resource location in end to end
commutation. The iterative and recursive routing process in RELOAD is
shown in Fig4, which is different from PPSP. That is, the data stored
in RELOAD is user profile data and the requester knows exactly what the
data is (e.g., the location of Alice@chinamobile.com). While in PPSP,
Zhang Expires November 27, 2009 [Page 11]
Internet-Draft Problem Statement of P2P Streaming Protocol May 2009
there are many peers storing different data pieces of ''CNN Live''. The
only thing a peer must know is the metadata of the movie. A gossip
protocol to communicate with other peers is needed to get the real
data quickly.
+------------------------------------------------+
| +---------------+ |
| | Peer | |
| +---------------+ |
| ^ | ^ | |
| 1,2 | |1' 3,4| |3' |
| | | | | |
| V V V V |
| +-------------+ 2' +------------+ |
| | Peer |----->| Peer | |
| +-------------+ +------------+ |
| |
+------------------------------------------------+
Figure 2 P2PSIP process
The biggest difference between P2PSIP and PPSP lies in their
different Search efficiency requirements. PPSP requires retrieving
real-time/para real-time data and therefore a tree/mesh topology is
built. DHT based topology supported in RELAOD may not be suitable for
peer organization.
Although P2PSIP doesn't fit for streaming peer organization, it can
be deployed in PPSP environment to some extent.
First, the topology organization of DHT defined in RELOAD can be used
to organize multiple trackers. Due to the large population, some
trackers may cooperatively serve a channel/file. Considering the
great amount of channels, there are many trackers in practice.
Because the search time of which channel/file the tracker stores
accounts small percentage in the whole searching procedure, DHT can
be used to query for peer list in case of thousand of channels or
million of files which are hard to use one tracker.
Second, there is a detailed solution for Firewall/NAT transversal in
in RELOAD. PPSP faces with the same problem, although not as serious
as RELAOD because there are lots of peer candidates with public
addresses. But we can definitely reuse RELAOD in case of NAT/Firewall
transversal which may occur in mobile environment.
Zhang Expires November 27, 2009 [Page 12]
Internet-Draft Problem Statement of P2P Streaming Protocol May 2009
4.2.2. Integration with ALTO
ALTO is to define a protocol realizing local traffic mostly for P2P
applications. The goal of an ALTO solution would be to help peers to
find the best sources and the best destinations for media flows they
receive and relay. In [draft-marocco-alto-problem-statement-03], the
authors mentioned p2p live streaming that can benefit from ALTO
service.
As we have seen in Section 1, P2P streaming traffic account much on
the Internet backbone. PPSP has to consider operator-friendly ways to
reduce the cross-backbone traffic in order to control the
transmission cost. ALTO is a good candidate to do so.
ALTO provides a 3rd party to provide peer locality information based
on the premise that the 3rd party has the knowledge of the network
topology. In addition there may have extra traffic localization means,
e.g., in [draft-zhang-alto-traceroute-00], peers can cluster the
nearby peers by simple and lightweight measurements. We can use this
mechanism in peer selection to further reduce the traffic.
4.2.3. Encapsulating RTSP
At a first sight, the function of PPSP protocol is similar to
traditional client/server streaming control protocols, RTSP. But in
fact RTSP don't involve the problems PPSP has.
In RTSP, the focus is to control the streaming, like PLAY, PAUSE;
however PPSP focuses on signaling peers for real-time resource
discovery, merge and synchronization instead of how to realize
streaming control.
On the other hand, considering the prevalence of RTSP, it can also be
reused in clients without installing PPSP-compatible software to use
P2P streaming service.
The basic idea in that is to set up a proxy node (e.g., using a cache)
which can act as a RTSP server who is actually a peer of PPSP.
4.2.4. Edge Device such as Cache Integration
As stated in Section 3, if the widely deployed web cache and CDN
nodes can be reused in p2p streaming, the video quality, local
traffic and some security problems may be better solved.[HTPT]
proposes a solution similar to what we propose in section 4.5 to
encapsulate and transport them using the HTTP protocol so that they
Zhang Expires November 27, 2009 [Page 13]
Internet-Draft Problem Statement of P2P Streaming Protocol May 2009
are cacheable.
Zhang Expires November 27, 2009 [Page 14]
Internet-Draft Problem Statement of P2P Streaming Protocol May 2009
5. Scope of Work
We propose to develop an open peer to peer streaming protocol, namely
PPSP. The basic task of PPSP is to define a protocol of locating and
transmitting real-time data efficiently from multiple sources with
different pieces in peer to peer environment.
PPSP focuses on how to negotiate with un-preassigned peers for needed
chunks. Therefore, PPSP is mainly a signaling protocol and the
protocol stack of PPSP is shown in Figure 3.
+-----------------------------+
| PPSP Application |
|-----------------------------|
| PPSP Signaling Protocol |
|-----------------------------|
| Transport Layer |
+-----------------------------+
Figure 3 PPSP Position in Protocol Stack
In order to locate real-time data efficiently from multiple sources
with different pieces, we have to solve the following problems:
1) To standardize the architecture for locating the data efficiently.
Tracker-based structure is widely used in current p2p streaming
practice. However some tracker-less structures like DHT peer
management solutions are also proposed. We need to decide which one
is better for p2p streaming;
2) To standardize the signaling interaction process.
In this part we actually want to standardize client registration
process (analogous to TCP 3-way handshake procedure), client
information exchange process (analogous to SIP session setup process)
and client report process.
The current tracker-based means is a two-step searching, i.e., peer
reporting to tracker coarse information about it has. Once the
tracker is asked, it informs peer of the coarse information; the
grain information is achieved by peers exchanging bitmap each other.
Some other means, e.g., peer reporting to tracker grain information
directly and tracker informs peer of the exact information or DHT
based searching, should be evaluated. In each proposal we also need
to define the message format in the interactions.
Zhang Expires November 27, 2009 [Page 15]
Internet-Draft Problem Statement of P2P Streaming Protocol May 2009
3) To discuss how to incorporate node information such as online time,
link status, node capability, battery information and some
application requirements parameters into the protocol to expand the
peer selection.
In this part, we actually want to standardize PPSP message headers
(analogous to IP header definition) and PPSP metadata format
(analogous to SIP header definition).
Zhang Expires November 27, 2009 [Page 16]
Internet-Draft Problem Statement of P2P Streaming Protocol May 2009
6. Security Considerations
PPSP has a similar assumption in peer privacy as P2PSIP, i.e., all
participants in the system are issued unique identities and
credentials through some mechanism not in the scope of this working
group, such as a centralized server. Hence the WG will not attempt a
solution to these issues for P2P streaming networks in general.
However PPSP has some unique privacy issue different from P2PSIP:
1) The content published by peers may not be checked by centralized
certificating server because of the high amount. Therefore P2P
streaming network faces with more serious malicious content
distribution danger than P2PSIP.
2) Content Pollution is a phenomenon P2PSIP may not meet with. But
these is a common problem faced by P2P streaming and file sharing.
It's recorded that there are about 50% of the content is polluted in
file sharing applications. Although it's not as serious as file
sharing content pollution, it's still a big issue P2P streaming
applications face with.
3) Because there is a tracker server who is critical to the P2P
streaming systems. It has more probability to launch attacks to the
tracker.
PPSP may include some security mechanisms to prevent malicious nodes
to pollute or launch attacks to the tracker. Security issues in PPSP
need to be further investigated.
7. Acknowledgments
Funding for the RFC Editor function is provided by the IETF
Administrative Support Activity (IASA).
8. References
8.1.Normative References
[Cisco] Approaching the Zettabyte Era by Cisco.
[PPLive] www.pplive.com
[PPStream] www.ppstream.com
[UUSee] www.uusee.com
[youtube] www.youtube.com
Zhang Expires November 27, 2009 [Page 17]
Internet-Draft Problem Statement of P2P Streaming Protocol May 2009
[tudou] www.tudou.com
[CNN] www.cnn.com
[Octoshape] www.octoshape.com
[ATT]http://mobile.sooyuu.com/Article/content/200905/217315094629281_
1.shtml
[Sigcomm:P2P streaming] Challenges, Design and Analysis of a Large-
scale P2P-VoD System,Yan Huang et al, Sigcomm08.
[draft-marocco-alto-problem-statement-03], Application-Layer Traffic
Optimization (ALTO) Problem Statement, E. Marocco et al, draft-
marocco-alto-problem-statement-03
[Pando]www.pando.com
[CoolStreaming] CoolStreaming/DONet: A Data-Driven Overlay Network
for Efficient Live Media Streaming, Xinyan Zhang et al,
[HPTP] HPTP: Relieving the Tension between ISPs and P2P, Guobin Shen
et al,
[draft-zhang-ppsp-protocol-comparison-measurement-
00]www.ietf.org/internet-drafts/draft-zhang-ppsp-protocol-comparison-
measurement-00.txt
[GENI] www.geni.net
[FIND]www.nets-find.net
[draft-zhang-ppsp-dsn-introduction-00]www.ietf.org/internet-
draft/draft-zhang-ppsp-dsn-introduction-00.txt
[MobileTV] MobileTV,Turning in or switching off, Arthur D. Little
[Computer Networks:Traffic] Traffic analysis of peer-to-peer IPTV
communities, Thomas Silverston et al, Computer Networks, 53 (2009)
470--484
[Survey]A survey on peer-to-peer video streaming systems Yong Liu et
al, Peer-to-Peer Netw Appl (2008) 1:18- -28,Springer.
[draft-zhang-alto-traceroute-00] www.ietf.org/internet-draft/draft-
zhang-alto-traceroute-00.txt
Zhang Expires November 27, 2009 [Page 18]
Internet-Draft Problem Statement of P2P Streaming Protocol May 2009
8.2.Informative References
Author's Addresses
Yunfei Zhang
China Mobile Communication Corporation
zhangyunfei@chinamobile.com
Ning Zong
Huawei Technologies Co., Ltd.
zongning@huawei.com
Gonzalo Camarillo
Ericsson
Gonzalo.Camarillo@ericsson.com
James Seng
PPLive
james.seng@pplive.com
Richard Yang
Yale University
yry@cs.yale.edu
Zhang Expires November 27, 2009 [Page 19]