live streaming with

7/31/2019 Live Streaming With

1/14

1

Live Streaming with

Receiver-based Peer-division MultiplexingHyunseok Chang Sugih Jamin Wenjie Wang

AbstractA number of commercial peer-to-peer systems forlive streaming have been introduced in recent years. The behaviorof these popular systems has been extensively studied in severalmeasurement papers. Due to the proprietary nature of thesecommercial systems, however, these studies have to rely on ablack-box approach, where packet traces are collected froma single or a limited number of measurement points, to infervarious properties of traffic on the control and data planes.Although such studies are useful to compare different systemsfrom end-users perspective, it is difficult to intuitively under-stand the observed properties without fully reverse-engineeringthe underlying systems. In this paper we describe the network

architecture of Zattoo, one of the largest production live stream-ing providers in Europe at the time of writing, and present alarge-scale measurement study of Zattoo using data collected bythe provider. To highlight, we found that even when the Zattoosystem was heavily loaded with as high as 20,000 concurrentusers on a single overlay, the median channel join delay remainedless than 2 to 5 seconds, and that, for a majority of users, thestreamed signal lags over-the-air broadcast signal by no morethan 3 seconds.

Index TermsPeer-to-peer system, live streaming, networkarchitecture

I. INTRODUCTION

THERE is an emerging market for IPTV. Numerous com-mercial systems now offer services over the Internet that

are similar to traditional over-the-air, cable, or satellite TV.

Live television, time-shifted programming, and content-on-

demand are all presently available over the Internet. Increased

broadband speed, growth of broadband subscription base, and

improved video compression technologies have contributed to

the emergence of these IPTV services.

We draw a distinction between three uses of peer-to-peer

(P2P) networks: delay tolerant file download of archival ma-

terial, delay sensitive progressive download (or streaming) of

archival material, and real-time live streaming. In the first

case, the completion of download is elastic, depending on

available bandwidth in the P2P network. The applicationbuffer receives data as it trickles in and informs the user

upon the completion of download. The user can then start

playing back the file for viewing in the case of a video

file. Bittorrent and variants are example of delay-tolerant file

download systems. In the second case, video playback starts as

H. Chang is with Alcatel-Lucent Bell Labs, Holmdel, NJ 07733 USA (e-mail: [email protected]).

S. Jamin is with EECS Department, University of Michigan, Ann Arbor,MI 48109 USA (e-mail: [email protected]).

W. Wang is with IBM Ressearch, CRL, Beijing 100193, China (e-mail:[email protected]).

This work was done when authors Chang and Wang were at Zattoo Inc.

soon as the application assesses it has sufficient data buffered

that, given the estimated download rate and the playback rate,

it will not deplete the buffer before the end of file. If this

assessment is wrong, the application would have to either

pause playback and rebuffer, or slow down playback. While

users would like playback to start as soon as possible, the

application has some degree of freedom in trading off playback

start time against estimated network capacity. Most video-on-

demand systems are examples of delay-sensitive progressive-

download application. The third case, real-time live streaming,

has the most stringent delay requirement. While progressivedownload may tolerate initial buffering of tens of seconds or

even minutes, live streaming generally cannot tolerate more

than a few seconds of buffering. Taking into account the

delay introduced by signal ingest and encoding, and network

transmission and propagation, the live streaming system can

introduce only a few seconds of buffering time end-to-end and

still be considered live [1].

The Zattoo peer-to-peer live streaming system was a free-

to-use network serving over 3 million registered users in eight

European countries at the time of study, with a maximum

of over 60,000 concurrent users on a single channel. The

system delivers live streams using a receiver-based, peer-

division multiplexing scheme as described in Section II. Toensure real-time performance when peer uplink capacity is

below requirement, Zattoo subsidizes the networks bandwidth

requirement, as described in Section III. After delving into

Zattoos architecture in detail, we study in Sections IV and V

large-scale measurements collected during the live broadcast

of the UEFA European Football Championship, one of the

most popular one-time events in Europe, in June, 2008 [2].

During the course of the month of June 2008, Zattoo served

more than 35 million sessions to more than one million distinct

users. Drawing from these measurements, we report on the

operational scalability of Zattoos live streaming system along

several key issues:

1) How does the system scale in terms of overlay size and

its effectiveness in utilizing peers uplink bandwidth?

2) How responsive is the system during channel switching,

for example, when compared to the 3-second channel

switch time of satellite TV?

3) How effective is the packet retransmission scheme in

allowing a peer to recover from transient congestion?

4) How effective is the receiver-based peer-division multi-

plexing scheme in delivering synchronized sub-streams?

5) How effective is the global bandwidth subsidy system

in provisioning for flash crowd scenarios?

IEEE/ACM TRANSACTIONS ON NETWORKING, VOL. 19, NO. 1, FEB 2011


2/14

2

6) Would a peer further away from the stream source

experience adversely long lag compared to a peer closer

to the stream source?

7) How effective is error-correcting code in isolating packet

losses on the overlay?

We also discuss in Section VI several challenges in in-

creasing the bandwidth contribution of Zattoo peers. Finally,

we describe related works in Section VII and conclude inSection VIII.

I I . SYSTEM ARCHITECTURE

The Zattoo system rebroadcasts live TV, captured from

satellites, onto the Internet. The system carries each TV

channel on a separate peer-to-peer delivery network and is not

limited in the number of TV channels it can carry. Although

a peer can freely switch from one TV channel to another, and

thereby departing and joining different peer-to-peer networks,

it can only join one peer-to-peer network at any one time.

We henceforth limit our description of the Zattoo delivery

network as it pertains to carrying one TV channel. Fig. 1shows a typical setup of a single TV channel carried on the

Zattoo network. TV signal captured from satellite is encoded

into H.264/AAC streams, encrypted, and sent onto the Zattoo

network. The encoding server may be physically separated

from the server delivering the encoded content onto the Zattoo

network. For ease of exposition, we will consider the two as

logically co-located on an Encoding Server. Users are required

to register themselves at the Zattoo website to download a free

copy of the Zattoo player application. To receive the signal

of a channel, the user first authenticates itself to the Zattoo

Authentication Server. Upon authentication, the user is granted

a ticket with limited lifetime. The user then presents this ticket,

along with the identity of the TV channel of interest, to theZattoo Rendezvous Server. If the ticket specifies that the user

is authorized to receive signal of the said TV channel, the

Rendezvous Server returns to the user a list of peers currently

joined to the P2P network carrying the channel, together with

a signed channel ticket. If the user is the first peer to join a

channel, the list of peers it receives contain only the Encoding

Server. The user joins the channel by contacting the peers

returned by the Rendezvous Server, presenting its channel

ticket, and obtaining the live stream of the channel from them

(see Section II-A for details).

Each live stream is sent out by the Encoding Server as

n logical sub-streams. The signal received from satellite is

encoded into a variable-bit rate stream. During periods ofsource quiescence, no data is generated. During source busy

periods, generated data is packetized into a packet stream,

with each packet limited to a maximum size. The Encoding

Server multiplexes this packet stream onto the Zattoo network

as n logical sub-streams. Thus the first packet generated isconsidered part of the first sub-stream, the second packet that

of the second sub-stream, the n-th packet that of the n-th sub-stream. The n+1-th packet cycles back to the first sub-stream,etc. such that the i-th sub-stream carries the mn+i-th packets,where m 0, 1 i n, and n a user-defined constant.We call a set of n packets with the same index multiplier m

Rendezvous Server

Authentication Server

Feedback Server

other admin servers

Demultiplexer

Encoding

Servers

Fig. 1. Zattoo delivery network architecture.

a segment. Thus m serves as the segment index, while iserves as the packet index within a segment. Each segment

is of size n packets. Being the packet index, i also serves asthe sub-stream index. The number mn + i is carried in each

packet as its sequence number.Zattoo uses the Reed-Solomon (RS) error correcting code

(ECC) for forward error correction. The RS code is a sys-

tematic code: of the n packets sent per segment, k < npackets carry the live stream data while the remainder carries

the redundant data [3, Section 7.3]. Due to the variable-bit

rate nature of the data stream, the time period covered by a

segment is variable, and a packet may be of size less than the

maximum packet size. A packet smaller than the maximum

packet size is zero-padded to the maximum packet size for

the purposes of computing the (shortened) RS code, but is

transmitted in its original size. Once a peer has received kpackets per segment, it can reconstruct the remaining n kpackets. We do not differentiate between streaming data and

redundant data in our discussion in the remainder of this paper.

When a new peer requests to join an existing peer, it

specifies the sub-stream(s) it would like to receive from the

existing peer. These sub-streams do not have to be consecutive.

Contingent upon availability of bandwidth at existing peers,

the receiving peer decides how to multiplex a stream onto

its set of neighboring peers, giving rise to our description of

the Zattoo live streaming protocol as a receiver-based, peer-

division multiplexing protocol. The details of peer-division

multiplexing is described in Section II-A while the details of

how a peer manages sub-stream forwarding and stream recon-

struction is described in Section II-B. Receiver-based peer-

division multiplexing has also been used by the latest version

of CoolStreaming peer-to-peer protocol though it differs from

Zattoo in its stream management (Section II-B) and adaptive

behavior (Section II-C) [4].

A. Peer-Division Multiplexing

To minimize per-packet processing time of a stream, the

Zattoo protocol sets up a virtual circuit with multiple fan outs

at each peer. When a peer joins a TV channel, it establishes

a peer-division multiplexing (PDM) scheme amongst a set of

neighboring peers, by building a virtual circuit to each of the



3/14

3

neighboring peers. Baring departure or performance degrada-

tion of a neighbor peer, the virtual circuits are maintained

until the joining peer switches to another TV channel. With the

virtual circuits set up, each packet is forwarded without further

per-packet handshaking between peers. We describe the PDM

boot strapping mechanism in this section and the adaptive

PDM mechanism to handle peer departure and performance

degradation in Section II-C.

The PDM establishment process consists of two phases:

the search phase and the join phase. In the search phase, the

new, joining peer determines its set of potential neighbors. In

the join phase, the joining peer requests peering relationships

with a subset of its potential neighbors. Upon acceptance of a

peering relationship request, the peers become neighbors and

a virtual circuit is formed between them.

Search phase. To obtain a list of potential neighbors, a

joining peer sends out a SEARCH message to a random subset

of the existing peers returned by the Rendezvous Server. The

SEARCH message contains the sub-stream indices for which

this joining peer is looking for peering relationships. The sub-

stream indices is usually represented as a bitmask of n bits,where n is the number of sub-streams defined for the TVchannel. In the beginning, the joining peer will be looking for

peering relationships for all sub-streams and have all the bits

in the bitmask turned on. In response to a SEARCH message,

an existing peer replies with the number of sub-streams it can

forward. From the returning SEARCH replies, the joining peer

constructs a set of potential neighbors that covers the full set of

sub-streams comprising the live stream of the TV channel. The

joining peer continues to wait for SEARCH replies until the

set of potential neighbors contains at least a minimum number

of peers, or until all SEARCH replies have been received.

With each SEARCH reply, the existing peer also returns a

random subset of its known peers. If a joining peer cannotform a set of potential neighbors that covers all of the sub-

streams of the TV channel, it initiates another SEARCH round,

sending SEARCH messages to peers newly learned from the

previous round. The joining peer gives up if it cannot obtain

the full stream after two SEARCH rounds. To help the joining

peer synchronize the sub-streams it receives from multiple

peers, each existing peer also indicates for each sub-stream

the latest sequence number it has received for that sub-stream,

and the existence of any quality problem. The joining peer can

then choose sub-streams with good quality that are closely

synchronized.

Join phase. Once the set of potential neighbors is estab-

lished, the joining peer sends JOIN requests to each potentialneighbor. The JOIN request lists the sub-streams for which

the joining peer would like to construct virtual circuit with the

potential neighbor. If a joining peer has l potential neighbors,each willing to forward it the full stream of a TV channel, it

would typically choose to have each forward only 1/l-th of thestream, to spread out the load amongst the peers and to speed

up error recovery, as described in Section II-C. In selecting

which of the potential neighbors to peer with, the joining

peer gives highest preference to topologically close-by peers,

even if these peers have less capacity or carry lower quality

sub-streams. The topological location of a peer is defined

Zattoo

Peer and

Player

Application

PDM

Fig. 2. Zattoo peer with IOB.

to be its subnet number, autonomous system (AS) number,

and country code, in that order of precedence. A joining

peer obtains its own topological location from the Zattoo

Authentication Server as part of its authentication process.

The list of peers returned by both the Rendezvous Server

and potential neighbors all come attached with topological

locations. A topology-aware overlay not only allows us to

be ISP-friendly, by minimizing inter-domain traffic and

thus save on transit bandwidth cost, but also helps reduce

the number of physical links and metro hops traversed inthe overlay network, potentially resulting in enhanced user-

perceived stream quality.

B. Stream Management

We represent a peer as a packet buffer, called the IOB,

fed by sub-streams incoming from the PDM constructed as

described in Section II-A.1 The IOB drains to (1) a local

media player if one is running, (2) a local file if recording

is supported, and (3) potentially other peers. Fig. 2 depicts

a Zattoo player application with virtual circuits established to

four peers. As packets from each sub-stream arrive at the peer,

they are stored in the IOB for reassembly to reconstruct thefull stream. Portions of the stream that have been reconstructed

are then played back to the user. In addition to providing

a reassembly area, the IOB also allows a peer to absorb

some variabilities in available network bandwidth and network

delay.

The IOB is referenced by an input pointer, a repair pointer,

and one or more output pointers. The input pointer points

to the slot in the IOB where the next incoming packet with

sequence number higher than the highest sequence number

1In the case of the Encoding Server, which we also consider a peer on theZattoo network, the buffer is fed by the encoding process.



4/14

4

received so far will be stored. The repair pointer always

points one slot beyond the last packet received in order and is

used to regulate packet retransmission and adaptive PDM as

described later. We assign an output pointer to each forwarding

destination. The output pointer of a destination indicates the

destinations current forwarding horizon on the IOB. In accor-

dance to the three types of possible forwarding destinations

listed above, we have three types of output pointers: player

pointer, file pointer, and peer pointer. One would typically

have at most one player pointer and one file pointer but

potentially multiple concurrent peer pointers, referencing an

IOB. The Zattoo player application does not currently support

recording.

Since we maintain the IOB as a circular buffer, if the

incoming packet rate is higher than the forwarding rate of a

particular destination, the input pointer will overrun the output

pointer of that destination. We could move the output pointer

to match the input pointer so that we consistently forward the

oldest packet in the IOB to the destination. Doing so, however,

requires checking the input pointer against all output pointers

on every packet arrival. Instead, we have implemented the IOBas a double buffer. With the double buffer, the positions of the

output pointers are checked against that of the input pointer

only when the input pointer moves from one sub-buffer to the

other. When the input pointer moves from sub-buffer a to sub-buffer b, all the output pointers still pointing to sub-buffer b aremoved to the start of sub-buffer a and sub-buffer b is flushed,ready to accept new packets. When a sub-buffer is flushed

while there are still output pointers referencing it, packets that

have not been forwarded to the destinations associate with

those pointers are lost to them, resulting in quality degradation.

To minimize packet lost due to sub-buffer flushing, we would

like to use large sub-buffers. However, the real-time delay

requirement of live streaming limits the usefulness of latearriving packets and effectively puts a cap on the maximum

size of the sub-buffers.

Different peers may request for different numbers of, possi-

bly non-consecutive, sub-streams. To accommodate the differ-

ent forwarding rates and regimes required by the destinations,

we associate a packet map and forwarding discipline with each

output pointer. Fig. 3 shows the packet map associated with an

output peer pointer where the peer has requested sub-streams

1, 4, 9, and 14. Every time a peer pointer is repositioned

to the beginning of a sub-buffer of the IOB, all the packet

slots of the requested sub-streams are marked NEEDed and

all the slots of the sub-streams not requested by the peer are

marked SKIP. When a NEEDed packet arrives and is storedin the IOB, its state in the packet map is changed to READY.

As the peer pointer moves along its associated packet map,

READY packets are forwarded to the peer and their states

changed to SENT. A slot marked NEEDed but not READY,

such as slot n + 4 in Fig. 3, indicates that the packet is lostor will arrive out-of-order and is bypassed. When an out-of-

order packet arrives, its slot is changed to READY and the

peer pointer is reset to point to this slot. Once the out-of-order

packet has been sent to the peer, the peer pointer will move

forward, bypassing all SKIP, NEED, and SENT slots until it

reaches the next READY slot, where it can resume sending.

SKIP NEED

SENT READY

1 n

segment 0

4 9 14

segment m-1

Fig. 3. Packet map associated with a peer pointer.

Peer0

Peer1

Player

File

inputpointer

...

...

PacketMap

PacketMap

PacketMap

segment 0

segmentm-1

segment size: n

sub-buffer 0

sub-buffer 1

received packet empty slot

segment 0

segmentm-1

IOB

PacketMap

repair

pointer

Fig. 4. IOB, input/output pointers and packet maps.

The player pointer behaves the same as a peer pointer except

that all packets in its packet map will always start out marked

NEEDed.

Fig. 4 shows an IOB consisting of a double buffer, with an

input pointer, a repair pointer, and an output file pointer, anoutput player pointer, and two output peer pointers referencing

the IOB. Each output pointer has a packet map associated

with it. For the scenario depicted in the figure, the player

pointer tracks the input pointer and has skipped over some

lost packets. Both peer pointers are lagging the input pointer,

indicating that the forwarding rates to the peers are bandwidth

limited. The file pointer is pointing at the first lost packet.

Archiving a live stream to file does not impose real-time delay

bound on packet arrivals. To achieve the best quality recording

possible, a recording peer always waits for retransmission of

lost packets that cannot be recovered by error correction.

In addition to achieving lossless recording, we use re-

transmission to let a peer recover from transient networkcongestion. A peer sends out a retransmission request when

the distance between the repair pointer and the input pointer

has reached a threshold of R packet slots, usually spanningmultiple segments. A retransmission request consists of an R-bit packet mask, with each bit representing a packet, and the

sequence number of the packet corresponding to the first bit.

Marked bits in the packet mask indicate that the corresponding

packets need to be retransmitted. When a packet loss is

detected, it could be caused by congestion on the virtual

circuits forming the current PDM or congestion on the path

beyond the neighboring peers. In either case, current neighbor



5/14

5

peers will not be good sources of retransmitted packets. Hence

we send our retransmission requests to r random peers that arenot neighbor peers. A peer receiving a retransmission request

will honor the request only if the requested packets are still

in its IOB and it has sufficient left-over capacity, after serving

its current peers, to transmit all the requested packets. Once a

retransmission request is accepted, the peer will retransmit all

the requested packets to completion.

C. Adaptive PDM

While we rely on packet retransmission to recover from

transient congestions, we have two channel capacity adjust-

ment mechanisms to handle longer-term bandwidth fluctua-

tions. The first mechanism allows a forwarding peer to adapt

the number of sub-streams it will forward given its current

available bandwidth, while the second allows the receiving

peer to switch provider at the sub-stream level.

Peers on the Zattoo network can redistribute a highly

variable number of sub-streams, reflecting the high variability

in uplink bandwidth of different access network technologies.

For a full-stream consisting of sixteen constant-bit rate sub-

streams, our prior study show that based on realistic peer

characteristics measured from the Zattoo network, half of the

peers can support less than half of a stream, 82% of peers can

support less than a full-stream, and the remainder can support

up to ten full streams (peers that can redistribute more than

a full stream is conventionally known as supernodes in the

literature) [5]. With variable-bit rate streams, the bandwidth

carried by each sub-stream is also variable. To increase peer

bandwidth usage, without undue degradation of service, we

instituted measurement-based admission control at each peer.

In addition to controlling resource commitment, another goal

of the measurement-based admission control module is tocontinually estimate the amount of available uplink bandwidth

at a peer.

The amount of available uplink bandwidth at a peer is

initially estimated by the peer sending a pair of probe packets

to Zattoos Bandwidth Estimation Server. Once a peer starts

forwarding sub-streams to other peers, it will receive from

those peers quality-of-service feedbacks that inform its update

of available uplink bandwidth estimate. A peer sends quality-

of-service feedback only if the quality of a sub-stream drops

below a certain threshold.2 Upon receiving quality feedback

from multiple peers, a peer first determines if the identified

sub-streams are arriving in low quality. If so, the low quality of

service may not be caused by limit on its own available uplinkbandwidth; in which case, it ignores the low quality feedbacks.

Otherwise, the peer decrements its estimate of available uplink

bandwidth. If the new estimate is below the bandwidth needed

to support existing number of virtual circuits, the peer closes

2Depending on a peers NAT and/or firewall configuration, Zattoo useseither UDP or TCP as the underlying transport protocol. The quality of a sub-stream is measured differently for UDP and TCP. A packet is considered lostunder UDP if it doesnt arrive within a fixed threshold. The quality measurefor UDP is computed as a function of both the packet lost rate and the bursterror rate (number of contiguous packet losses). The quality measure for TCPis defined to be how far behind a peer is, relative to other peers, in servingits sub-streams.

a virtual circuit. To reduce the instability introduced into the

network, a peer closes first the virtual circuit carrying the

smallest number of sub-streams.

A peer attempts to increase its available uplink bandwidth

estimate periodically: if it has fully utilized its current estimate

of available uplink bandwidth without triggering any bad

quality feedback from neighboring peers. A peer doubles the

estimated available uplink bandwidth if current estimate is

below a threshold, switching to linear increase above the

threshold, similar to how TCP maintains its congestion win-

dow size. A peer also increases its estimate of available uplink

bandwidth if a neighbor peer departs the network without any

bad quality feedback.

When the repair pointer lags behind the input pointer by Rpacket slots, in addition to initiating a retransmission request,

a peer also computes a loss rate over the R packets. If theloss rate is above a threshold, the peer considers the neighbor

slow and attempts to reconfigure its PDM. In reconfiguring

its PDM, a peer attempts to shift half of the sub-streams

currently forwarded by the slow neighbor to other existing

neighbors. At the same time, it searches for new peer(s) toforward these sub-streams. If new peer(s) are found, the load

will be shifted from existing neighbors to the new peer(s).

If sub-streams from the slow neighbor continues to suffer

after the reconfiguration of the PDM, the peer will drop the

neighbor completely and initiate another reconfiguration of the

PDM. When a peer loses a neighbor due to reduced available

uplink bandwidth at the neighbor or due to neighbor departure,

it also initiates a PDM reconfiguration. A peer may also

initiate a PDM reconfiguration to switch to a topologically

closer peer. Similar to the PDM establishment process, PDM

reconfiguration is accomplished by peers exchanging sub-

stream bitmasks in a request/response handshake, with each

bit of the bitmask representing a sub-stream. During and aftera PDM reconfiguration, slow neighbor detection is disabled

for a short period of time to allow for the system to stabilize.

III. GLOBAL BANDWIDTH SUBSIDY SYSTEM

Each peer on the Zattoo network is assumed to serve a

user through a media player, which means that each peer

must receive, and can potentially forward, all n sub-streamsof the TV channel the user is watching. The limited redistri-

bution capacity of peers on the Zattoo network means that

a typical client can contribute only a fraction of the sub-

streams that make up a channel. This shortage of bandwidth

leads to a global bandwidth deficit in the peer-to-peer net-

work. Whereas bittorrent-like delay-tolerant file downloads orthe delay-sensitive progressive download of video-on-demand

applications can mitigate such global bandwidth shortage by

increasing download time, a live streaming system such as

Zattoos must subsidize the bandwidth shortfall to provide

real-time delivery guarantee.

Zattoos Global Bandwidth Subsidy System (or simply, the

Subsidy System), consists of a global bandwidth monitoring

subsystem, a global bandwidth forecasting and provisioning

subsystem, and a pool of Repeater nodes. The monitoring

subsystem continuously monitors the global bandwidth re-

quirement of a channel. The forecasting and provisioning



6/14

6

subsytem projects global bandwidth requirement based on

measured history and allocates Repeater nodes to the chan-

nel as needed. The monitoring and provisioning of global

bandwidth is complicated by two highly varying parameters

over time, client population size and peak streaming rate, and

one varying parameter over space, available uplink bandwidth,

which is network-service provider dependent. Forecasting of

bandwidth requirement is a vast subject in itself. Zattoo

adopted a very simple mechanism, described in Section III-B

which has performed adequately in provisioning the network

for both daily demand fluctuations and flash crowds scenarios

(see Section IV-C).

When a bandwidth shortage is projected for a channel, the

Subsidy System assigns one or more Repeater nodes to the

channel. Repeater nodes function as bandwidth multiplier, to

amplify the amount of available bandwidth in the network.

Each Repeater node serves at most one channel at a time;

it joins and leaves a given channel at the behest of the

Subsidy System. Repeater nodes receive and serve all nsub-streams of the channel they join, run the same PDM

protocol, and are treated by actual peers like any other peerson the network; however, as bandwidth amplifiers, they are

usually provisioned to contribute more uplink bandwidth than

the download bandwidth they consume. The use of Repeater

nodes makes the Zattoo network a hybrid P2P and content

distribution network.

We next describe the bandwidth monitoring subsystem of

the Subsidy System, followed by design of the simple band-

width projection and Repeater node assignment subsystem.

A. Global Bandwidth Measurement

The capacity metric of a channel is the tuple (D, C),

where D is the aggregate download rates required by allusers on the channel, and C is the aggregate upload capacityof those users. Usually C < D and the difference betweenthe two is the global bandwidth deficit of the channel. Since

channel viewership changes over time as users join or leave

the channel, we need a scalable means to measure and update

the capacity metric. We rely on aggregating the capacity

metric up the peer division multiplexing tree. Each peer in the

overlay periodically aggregates the capacity metric reported

by all its downstream receiver peers, adds its own capacity

measure (D, C) to the aggregate, and forwards the resultingcapacity metric upstream to its forwarding peers. By the time

the capacity metric percolates up to the Encoding Server, it

contains the total download and upload rate aggregates of thewhole streaming overlay. The Encoding Server then simply

forwards the obtained (D, C) to the Subsidy Server.

B. Global Bandwidth Projection and Provisioning

For each channel, the Subsidy Server keeps a history of the

capacity metric (D, C) reports received from the channelsEncoding Server. The channel utilization ratio (U) is the ratioD over C. Based on recent movements of the ratio U, weclassify the capacity trend of each channel into the following

four categories.

Stable: the ratio U has remained within [S-, S+] forthe past Ts reporting periods.

Exploding: the ratio U increased by at least E betweenTe (e.g., Te = 2) reporting periods.

Increasing: the ratio U has steadily increased by I (IE) over the past Ti reporting periods.

Falling: the ratio U has decreased by F over the past Tfreporting periods.

Orthogonal to the capacity-trend based classification above,

each channel is further categorized in terms of its capacity

utilization ratio as follows.

Under-utilized: the ratio U is below the low thresholdL, e.g., U 0.5.

Near Capacity: the ratio U is almost 1.0, e.g., U 0.9.

If the capacity trend of a channel has been Exploding, one

or more Repeater nodes will be assigned to it immediately. If

the channels capacity trend has been Increasing, it will be

assigned a Repeater node with a smaller capacity. The goal

of the Subsidy System is to keep a channel below Near

Capacity. If a channels capacity trend is Stable and the

channel is Under-utilized, the Subsidy System attempts tofree Repeater nodes (if any) assigned to the channel. If a

channels capacity utilization is Falling, the Subsidy System

waits for the utilization to stabilize before reassigning any

Repeater nodes.

Each Repeater node periodically sends a keep-alive message

to the Subsidy System. The keep-live message tells the Sub-

sidy System which channel the Repeater node is serving, plus

its CPU and capacity utilization ratio. This allows the Subsidy

System to monitor the health of Repeater nodes and to increase

the stability of the overlay during Repeater reassignment.

When reassigning Repeater nodes from a channel, the Subsidy

System will start from the Repeater node with the lowest

utilization ratio. It will notify the selected Repeater node to

stop accepting new peers and then to leave the channel after

a specified time.

In addition to Repeater nodes, the Subsidy System may

recognize extra capacity from idle peers whose owners are

not actively watching a channel. However, our previous study

shows that a large number of idle peers are required to make

any discernible impact on the global bandwidth deficit of a

channel [5]. Our current Subsidy System therefore does not

solicit bandwidth contribution from idle peers.

IV. SERVER-SIDE MEASUREMENTS

In the Zattoo system, two separate centralized collectorservers collect usage statistics and error reports, which we

call the stats server and the user-feedback server re-

spectively. The stats server periodically collects aggregated

player statistics from individual peers, from which full session

logs are constructed and entered into a session database.

The session database gives a complete picture of all past

and present sessions served by the Zattoo system. A given

database entry contains statistics about a particular session,

which includes join time, leave time, uplink bytes, download

bytes, and channel name associated with the session. We

study the sessions generated on three major TV channels from



7/14

7

TABLE ISESSION DATABASE (6/1/20086/30/2008).

Channel # sessions # distinct users

ARD 2,102,638 298,601

Cuatro 1,845,843 268,522

SF2 1,425,285 157,639

TABLE IIFEEDBACK LOGS (6/20/20086/29/2008).

Channel # feedback logs # sessions

ARD 871 1,253

Cuatro 2,922 4,568

SF2 656 1,140

three different countries (Germany, Spain, and Switzerland),

from June 1st to June 30th, 2008. Throughout the paper, we

label those channels from Germany, Spain, and Switzerland as

ARD, Cuatro, and SF2, respectively. Euro 2008 games were

held during this period, and those three channels broadcast a

majority of the Euro 2008 games including the final match.

See Table I for information about the collected session datasets.

The user-feedback server, on the other hand, collects

users error logs submitted asynchronously by users. The user

feedback data here is different from peers quality feedback

used in PDM reconfiguration described in Section II-C. Zat-

too player maintains an encrypted log file which contains,

for debugging purposes, detailed behavior of client-side P2P

engine, as well as history of all the streaming sessions initiated

by a user since the player startup. When users encounter

any error while using the player, such as log-in error, join

failure, bad quality streaming etc., they can choose to report

the error by clicking a Submit Feedback button on the

player, which causes the Zattoo player to send the generatedlog file to the user-feedback server. Since a given feedback

log not only reports on a particular error, but also describes

normal sessions generated prior to the occurrence of the

error, we can study users viewing experience (e.g., channel

join delay) from the feedback logs. Table II describes the

feedback logs collected from June 20th to June 29th. A given

feedback log can contain multiple sessions (for the same or

different channels), depending on users viewing behavior.

The second column in the table represents the number of

feedback logs which contain at least one session generated

on the channel listed in the corresponding entry in the first

column. The numbers in the third column indicate how many

distinct sessions generated on said channel are present in thefeedback logs.

A. Overlay Size and Sharing Ratio

We first study how many concurrent users are supported by

the Zattoo system, and how much bandwidth is contributed by

them. For this purpose, we use the session database presented

in Table I. By using the join/leave timestamps of the collected

sessions, we calculate the number of concurrent users on a

given channel at time i. Then we calculate the average sharingratio of the given channel at the same time. The average

TABLE IIIAVERAGE SHARING RATIO.

ChannelAverage sharing ratioOff-peak Peak

ARD 0.335 0.313

Cuatro 0.242 0.224

SF2 0.277 0.222

sharing ratio is defined as total users uplink rate divided

by their download rate on the channel. A sharing ratio of

one means users contribute to other peers in the network as

much traffic as they download at the time. We calculate the

average sharing ratio from the total download/uplink bytes

of the collected sessions. We first obtain all the sessions

which are active across time i. We call a set of suchsessions Si. Then assuming uplink/download bytes of eachsession are spread uniformly throughout the entire session

duration, we approximate the average sharing ratio at time

i as

iSi

uplink bytes(i)/duration(i)

iSi

download bytes(i)/duration(i).

Fig. 5 shows the overlay size (i.e., number of concurrentusers) and average sharing ratio super-imposed across the

month of June, 2008. According to the figure, the overlay

size grew to more than 20,000 (e.g., 20,059 on ARD on 6/18

and 22,152 on Cuatro on 6/9). As opposed to the overlay

size, the average sharing ratio tends to stay flatter throughout

the month. Occasional spikes in the sharing ratio all occurred

during 2AM to 6AM (GMT) when the channel usage is very

low, and therefore may be considered statistically insignificant.

By segmenting a 24-hour day into two time periods, e.g.,

off-peak hours (0AM-6PM) and peak hours (6PM-0AM),

Table III shows the average sharing ratio in the two time

periods separately. Zattoos usage during peak hours typically

accounts for about 50% to 70% of the total usage of theday. According to the table, the average sharing ratio during

peak hours is slightly lower than, but not very much different

from during off-peak hours. Cuatro channel in Spain exhibits

relatively lower sharing ratio than the two other channels. One

bandwidth test site [6] reports that average uplink bandwidth in

Spain is about 205 kbps, which is much lower than in Germany

(582 kbps) and Switzerland (787 kbps). The lower sharing

ratio on the Spanish channel may reflect regional difference

in residential access network provisioning. The balance of the

required bandwidth is provided by Zattoos Encoding Server

and Repeater nodes.

B. Channel Switching Delay

When user clicks on a new channel button, it takes some

time (a.k.a. channel switching delay) for the user to be

able to start watching streamed video on Zattoo player. The

channel switching delay has two components. First, Zattoo

player needs to contact other available peers and retrieve all

required sub-streams from them. We call the delay incurred

during this stage join delay. Once all necessary sub-streams

have been negotiated successfully, the player then needs to

wait and buffer the minimum amount of streams (e.g., 3

seconds) before starting to show the video to the user. We call



8/14

8

0

5000

10000

15000

20000

25000

0 5 10 15 20 25 3000.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

OverlaySize

SharingRatio

Day in 2008/6

Overlay SizeSharing Ratio

(a) ARD

0

5000

10000

15000

20000

25000

0 5 10 15 20 25 3000.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

OverlaySize

SharingRatio

Day in 2008/6


(b) Cuatro

0

5000

10000

15000

20000

25000

0 5 10 15 20 25 3000.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

OverlaySize

SharingRatio

Day in 2008/6


(c) SF2

Fig. 5. Overlay size and sharing ratio.

0

0.2

0.4

0.6

0.8

1

0 2 4 6 8 10 12 14 16 18 20 22 24

CDF

Arrival Hour

Feedback LogsSession Database

(a) ARD

0

0.2

0.4

0.6

0.8

1

0 2 4 6 8 10 12 14 16 18 20 22 24

CDF

Arrival Hour


(b) Cuatro

0

0.2

0.4

0.6

0.8

1

0 2 4 6 8 10 12 14 16 18 20 22 24

CDF

Arrival Hour


(c) SF2

Fig. 6. CDF of user arrival time.

the resulting wait time buffering delay. The total channel

switching delay experienced by users is thus the sum of join

delay and buffering delay. PPLive reports channel switching

delay around 20 to 30 seconds, but can be as high as 2 minutes,

of which join delay typically accounts for 10 to 15 seconds [7].

We measure the join delay experienced by Zattoo users from

the feedback logs described in Table II. Debugging informa-

tion contained in the feedback logs tells us when user clickedon a particular channel, and when the player has successfully

joined the P2P overlay and starting to buffer content. One

concern in relying on user-submitted feedback logs to infer

join delay is the potential sampling bias associated with them.

Users typically submit feedback logs when they encounter

some kind of errors, and that brings up the question of whether

the captured sessions are representative samples to study.

We attempt to address this concern by comparing the data

from feedback logs against those from the session database.

The latter captures the complete picture of users channel

watching behavior, and therefore can serve as a reference. In

our analysis, we compare the user arrival time distribution

obtained from the two data sets. For fair comparison, we useda subset of the session database which was generated during

the same period when the feedback logs were collected (i.e.,

from June 20th to 29th).

Fig. 6 plots the CDF distribution of user arrivals per hour

obtained from feedback logs and session database separately.

The steep slope of the distributions during hour 18-20 (6-

8PM) indicates the high frequency of user arrivals during

those hours. On ARD and Cuatro, the user arrival distributions

inferred from feedback logs are almost identical to those from

session database. On the other hand, on SF2, the distribution

obtained from feedback logs tends to grow slowly during early

TABLE IVMEDIAN CHANNEL JOIN DELAY.

ChannelMedian join delay Maximum overlay size

Off-peak Peak Off-peak Peak

ARD 2.29 sec 1.96 sec 2,313 19,223

Cuatro 3.67 sec 4.48 sec 2,357 8,073

SF2 2.49 sec 2.67 sec 1,126 11,360

hours, which indicates that feedback submission rate duringoff-peak hours on SF2 was relatively lower than normal. Later

during peak hours, however, feedback submission rate picks

up as expected, closely matching the actual user arrival rate.

Based on this observation, we argue that feedback logs can

serve as representative samples of daily user activities.

Fig. 7 shows the CDF distributions of channel join delay

for ARD, Cuatro and SF2 channels. We show the distributions

for off-peak hours (0AM-6PM) and peak hours (6PM-0AM)

separately. Median channel join delay is also presented in a

similar fashion in Table IV. According to the CDF distribu-

tions, 80% of users experience less than 4 to 8 seconds of

join delay, and 50% of users even less than 2 seconds of join

delay. Also, Table IV shows that even a 10-fold increase onthe number of concurrent users during peak hours does not

unduly lengthen the channel join delay (up to 22% increase

in median join delay).

C. Repeater Node Assignment

As illustrated in Fig. 5, the live coverage of Euro Cup games

brought huge flash crowds to the Zattoo system. With typical

users contributing only about 2530% of the average streaming

bitrate, Zattoo must subsidize the balance. As described in

Section III, the Zattoos Subsidy System assigns Repeater



9/14

9

00.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30

CDF

Channel Join Delay (Sec)

Off-Peak HoursPeak Hours

(a) ARD

00.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30

CDF



(b) Cuatro

00.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30

CDF



(c) SF2

Fig. 7. CDF of channel join delay.

0

5000

10000

15000

20000

25000

18 19 20 21 22 23 240

20

40

60

80

100

120

140

160

180

200

OverlaySize

NumberofRepeatersAssigned

Hour of Day

Overlay SizeNumber of Repeaters

(a) ARD

0

5000

10000

15000

20000

25000

12 14 16 18 20 22 240

20

40

60

80

100

120

140

160

180

200

OverlaySize


Hour of Day


(b) Cuatro

0

5000

10000

15000

20000

25000

18 19 20 21 22 23 240

20

40

60

80

100

120

140

160

180

200

OverlaySize


Hour of Day


(c) SF2

Fig. 8. Overlay size and channel provisioning.

nodes to channels that require more bandwidth than its own

globally available aggregate upload bandwidth.

Fig. 8 shows the Subsidy System assigning more Repeater

nodes to a channel as flash crowds arrived during each Euro

Cup game and then gradually reclaiming them as the flash

crowd departed. For each of the channel reported, we choose

a particular date with the biggest flash crowd on the channel.

The dip in overlay sizes on ARD and Cuatro channels occurred

during the half-time break of the games. The Subsidy Server

was less aggressive in assigning Repeater nodes to the ARD

channel, as compared to the other two channels, because the

Repeater nodes in the vicinity of the ARD server have higher

capacity than those near the other two.

V. CLIENT-SIDE MEASUREMENTS

To further study the P2P overlay beyond details obtain-

able from aggregated session-level statistics, we run several

modified Zattoo clients which periodically retrieve the internal

states of other participating peers in the network by exchang-

ing SEARCH/JOIN messages with them. After a given probesession is over, the monitoring client archives a log file where

we can analyze control/data traffic exchanged and detailed

protocol behavior. We run the experiment during Zattoos live

coverage of Euro 2008 (June 7th to 29th). The monitoring

clients tuned to game channels from one of Zattoos data

centers located in Switzerland while the games were broadcast

live. The data sets presented in this paper were collected

during the coverage of the championship final on two separate

channels: ARD in Germany and Cuatro in Spain. Soccer teams

from Germany and Spain participated in the championship

final.

As described in Section II-A, Zattoos peer discovery is

guided by peers topology information. To minimize potential

sampling bias caused by our use of single vantage point for

monitoring, we assigned empty AS number and country

code to our monitoring clients, so that their probing is not

geared towards those peers located in the same AS and

country.

A. Sub-Stream Synchrony

To ensure good viewing quality, peer should not only

obtain all necessary sub-streams (discounting redundant sub-

streams), but also have those sub-streams delivered temporally

synchronized with each other for proper online decoding. Re-

ceiving out-of-sync sub-streams typically results in pixelated

screen on the player. As described in Sections 1 and II-C,

Zattoos protocol favors sub-streams that are relatively in-

sync when constructing the PDM, and continually monitors

the sub-streams progression over time, replacing those sub-

streams that have fallen behind and reconfiguring the PDM

when necessary. In this section we measure the effectivenessof Zattoos Adaptive PDM in selecting sub-streams that are

largely in-sync.

To quantify such inter-sub stream synchrony, we measure

the difference in the latest (i.e., maximum) packet sequence

numbers belonging to different incoming sub-streams. When a

remote peer responds to a SEARCH query message, it includes

in its SEARCH reply the latest sequence numbers that it has

received for all sub-streams. If some sub-streams happen to be

lossy or stalled at that time, the peer marks such sub-streams

in its SEARCH replies. Thus, we can inspect SEARCH replies

from existing peers to study their inter-sub stream synchrony.



10/14

10

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

CDF

Number of Bad Sub-streams

Euro 2008 final on ARDEuro 2008 final on Cuatro

(a) CDF for number of bad sub-streams

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 100 200 300 400 500 600 700 800 900 1000

CDF

Sub-stream Synchrony (# of Packets)

Euro 2008 final on ARDEuro 2008 final on Cuatro

(b) CDF for sub-stream synchrony

Fig. 9. Sub-stream synchrony.

In our experiment, we collected SEARCH replies from

4,420 and 6,530 distinct peers from ARD and Cuatro channels

respectively, during the 2-hour coverage of the final game.

From the collected SEARCH replies, we check how many

sub-streams (out of 16) are bad (e.g., lossy or missing) foreach peer. Fig. 9(a) shows the CDF distribution of the number

of bad sub-streams. According to the figure, about 99% (ARD)

and 96% (Cuatro) of peers have 3 or less bad sub-streams.

Current Zattoo deployment dedicates k = 3 sub-streams(out of n = 16) for loss recovery purposes. That is, given asegment of 16 consecutive packets, if peer has received at least

13 packets, it can reconstruct the remaining 3 packets from

the RS error correcting code (see Section 1). Thus if peer can

receive any 13 sub-streams out of 16 reliably, it can decode

the full stream properly. The result in Fig. 9 (a) suggests that

the number of bad sub-streams is low enough as to not cause

quality issues in the Zattoo network.

After discounting bad sub-streams, we then look at the

synchrony of the remaining good sub-streams in each peer.

Fig. 9(b) shows the CDF distribution of the sub-stream syn-

chrony in the two channels. Sub-stream synchrony of a given

peer is defined as the difference between maximum and mini-

mum packet sequence numbers among all sub-streams, which

is obtained from the peers SEARCH reply. For example, if

some peer has sub-stream synchrony measured at 100, it means

that the peer has one sub-stream that is ahead of another sub-

stream by 100 packets. If all the packets are received in order,

the sub-stream synchrony of a peer measures at most n 1.If we received multiple SEARCH replies from the same peer,

we average the sub-stream synchrony across all the replies.

Given the 500 kbps average channel data rate, 60 consecutive

packets roughly correspond to 1-second worth of streaming.

Thus, the figure shows that on Cuatro channel, 20% of peers

have their sub-streams completely in-sync, while more than

90% have their sub-streams lagging each other by at most

5 seconds; on ARD channel, 30% are in-sync, and more than

90% are within 1.5 seconds. The buffer space of Zattoo player

has been dimensioned sufficiently to accommodate such low

degree of out-of-sync sub-streams.

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

-350 -300 -250 -200 -150 -100 -50 0 50 100

CDF

Relative Peer Synchrony (# of Packets)

Euro 2008 final on CuatroEuro 2008 final on ARD

Fig. 10. Peer synchrony.

B. Peer Synchrony

While sub-stream synchrony tells us stream quality differentpeers may experience, peer synchrony tells us how varied in

time peers viewing points are. With small scale P2P networks,

all participating peers are likely to watch live streaming

roughly synchronized in time. However, as the size of the

P2P overlay grows, the viewing point of edge nodes may be

delayed significantly compared to those closer to the Encoding

Server. In the experiment, we define the viewing point of a

peer as the median of the latest sequence numbers across

its sub-streams. Then we choose one peer (e.g., a Repeater

node directly connected to the Encoding Server) as a reference

point, and compare other peers viewing point against the

reference viewing point.

Fig. 10 shows the CDFs of relative peer synchrony. Therelative peer synchrony of peer X is obtained by the viewingpoint of X subtracted by the reference viewing point. So peersynchrony at -60 means that a given peers viewing point

is delayed by 60 packets (roughly 1 second for a 500 kbps

stream) compared to the reference viewing point. A positive

viewing point means that a given peers stream gets ahead

of the reference point, which could happen for peers which

receive streams directly from the Encoding Server. The figure

shows that about 1% of peers on ARD and 4% of peers on

Cuatro experienced more than 3 seconds (i.e., 180 packets)

delay in streaming compared to the reference viewing point.



11/14

11

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

-350 -300 -250 -200 -150 -100 -50 0 50 100

CDF

Peer Synchrony

Peer hops from ES: 6Peer hops from ES: 5Peer hops from ES: 4Peer hops from ES: 3Peer hops from ES: 2

(a) ARD

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

-350 -300 -250 -200 -150 -100 -50 0 50 100

CDF

Peer Synchrony

Peer hops from ES: 6Peer hops from ES: 5Peer hops from ES: 4Peer hops from ES: 3Peer hops from ES: 2

(b) Cuatro

Fig. 11. Peer synchrony vs. peer-hop from the Encoding Server.

To better understand to what extent peers position in the

overlay affects peer synchrony, we plot in Fig. 11 the CDFs

of peer synchrony at different depths, i.e., distances from the

Encoding Server. We look at how much delay is introduced

for sub-streams traversing i peers from the Encoding Server(i.e., depth i). For this purpose, we associate per-sub streamviewing point information from SEARCH reply with per-sub

stream overlay depth information from JOIN reply, where

SEARCH/JOIN replies were sent from the same peer, close in

time (e.g., less than 1 second apart). If the length of peer-hop

from the Encoding Server has an adverse effect on playback

delay and viewing point, we expect peers further away from

the Encoding Server to be further offset from the reference

viewing point, resulting in CDFs that are shifted to the upper

left corner for peers at further distances from the Encoding

Server. The figure shows an absence of such a leftward shift

in the CDFs. The median delay at depth 6 does not grow

by more than 0.5 seconds compared to the median delay at

depth 2. The figure shows that having more peers further away

from the Encoding Servers increases the number of peers that

are 0.5 seconds behind the reference viewing point, without

increasing the offset in viewing point itself. On the Zattoo

network, once the virtual circuits comprising a PDM have

been set up, each packet are streamed with minimal delay

through each peer. Hence each peer-hop from the Encoding

Server introduces delay only in tens of milliseconds range,

attesting to the suitability of the network architecture to carry

live media streaming on large-scale P2P networks.

C. Effectiveness of ECC in Isolating Loss

Now we investigate the effects of overlay sizes on the per-

formance scalability of the Zattoo system. Here we focus on

client-side quality (e.g., loss rate). As described in Section 1,

Zattoo-broadcast media streams are RS encoded, which allows

peers to reconstruct a full stream once they obtain at least k ofn sub-streams. Since the ECC-encoded stream reconstructionoccurs hop by hop in the overlay, it can mask sporadic packet

losses, and thus prevent packet losses from being propagated

throughout the overlay at large.

To see if such packet loss containment actually occurs in

the production system, we run the following experiments. We

let our monitoring client join a game channel, stay tuned for

15 seconds, and then leave. We wait for 10 seconds after the

15-second session is over. We repeat this process throughout

the 2-hour game coverage and collect the logs. A given log file

from each session contains a complete list of packet sequence

numbers received from the connected peers during the session,

from which we can detect upstream packet losses. We then

associate individual packet loss with the peer-path from the

Encoding Server. This gives us a rough sense of whether

packets traversing longer hops would experience higher loss

rate. In our analysis, we discount packets delivered for the first

3 seconds of a session to allow the PDM to stabilize.

Fig. 12 shows how the average packet loss rate changes

across different overlay depths (in all cases < 1%). On bothchannels, the packet loss rate does not grow with overlaydepth. This result confirms our expectation that ECC help

localize packet losses on the overlay.

V I . PEE R-TO-P EE R SHARING RATIO

The premise of a given P2P systems success in providing

scalable stream distribution is sufficient bandwidth sharing

from participating users [5]. Section IV-A shows that the

average sharing ratio of Zattoo users ranges from 0.2 to 0.35,

which translates into bandwidth uplinks ranging from 100 kbps

to 175 kbps. This is far lower than the numbers reportedas typical uplink bandwidth in countries where Zattoo is

available [6]. Aside from the possibility that users bandwidth

may be shared with other applications, we find factors such

as users behavior, support for variable-bit rate encoding,

and heterogeneous NAT environments contributing to sub-

optimal sharing performance. Designers of P2P streaming

systems must pay attention to these factors to achieve good

bandwidth sharing. However, one must also keep in mind that

improving bandwidth sharing should not be at the expense

of compromised user viewing experience, e.g., due to more

frequent uplink bandwidth saturation.



12/14

12

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0 1 2 3 4 5 6 7

PacketLossRate(%)

Peer Hops from Encoding Server

(a) ARD

0

0.1

0.2

0.3

0.4

0.5

0.6

0 1 2 3 4 5 6 7

PacketLossRate(%)

Peer Hops from Encoding Server

(b) Cuatro

Fig. 12. Packet loss rate vs. peer hops from Encoding Server.

0

10

20

30

40

50

60

0 10 20 30 40 50 60

AverageReceiver-PeerChurns

Session Length (Min)

Fig. 13. Receiver-peer churn frequency.

User churns: It is known that frequent user churns ac-

companied by short sessions, as well as flash crowd behaviorpose a unique challenge in live streaming systems [7], [8].

User churns can occur in both upstream (e.g., forwarding

peers) and downstream (e.g., receiver peers) connectivity on

the overlay. While frequent churns of forwarding peers can

cause quality problems, frequent changes of receiver peers can

lead to under-utilization of a forwarding peers uplink capacity.

To estimate receiver peers churn behavior, we visualize

in Fig. 13 the relationship between session length and the

frequency of receiver-peer churns experienced during the ses-

sions. We studied receiver-peer churn behavior for the Cuatro

channel by using Zattoos feedback logs described in Table II.

The y-value at session length x minutes denotes the average

number of receiver-peer churns experienced for sessions withlength [x, x + 1). According to the figure, the receiver-peerchurn frequency tends to grow linearly with session length,

and peers experience approximately one receiver-peer churn

every minute.

Variable-bit rate streams: Variable-bit rate (VBR) en-

coding is commonly used in production streaming systems,

including Zattoo, due to its better quality-to-bandwidth ra-

tio compared to constant-bit rate encoding. However, VBR

streams may put additional strain on peers bandwidth con-

tribution and quality optimization. As described in Sec-

tion II-C, the presence of VBR streams require peers to

perform measurement-based admission control when allocat-

ing resources to set up a virtual circuit. To avoid degrada-

tion of service due to overloading of the uplink bandwidth,

the measurement-based admission control module must be

conservative both in its reaction to increases in availablebandwidth and its allocation of available bandwidth to newly

joining peers. This conservativeness necessarily lead to under-

utilization of resources.

NAT reachability: Asymmetric reachability imposed by

prevalent NAT boxes adds to the difficulty in achieving full

utilization of users uplink capacity. Zattoo delivery system

supports 6 different NAT configurations: open host, full cone,

IP-restricted, port-restricted, symmetric, and UDP-disabled,

listed in increasing degree of restrictiveness. Not every pair-

wise communication can occur among different NAT types.

For example, peers behind a port-restricted NAT box cannot

communicate with those of symmetric NAT type (we docu-

mented a NAT reachability matrix in an earlier work [5]).

To examine how the varied NAT reachability comes into

play as far as sharing performance is concerned, we performed

the following two controlled experiments. In both experiments,

we run four Zattoo clients, each with a distinct NAT type,

tuned to the same channel concurrently for 30 minutes. In

the first experiment, we fixed the maximum allowable uplink

capacity (denoted max cp) of those clients to the same con-

stant value.3 In the second experiment, we let the max cp of

those clients self-adjust over time (which is the default setting

of Zattoo player). In both experiments, we monitored how the

uplink bandwidth utilization ratio (i.e., ratio between actively

used uplink bandwidth and maximum allowable bandwidthmax cp) changes over time. The experiments were repeated

three times for reliability.

Fig. 14 shows the capacity utilization ratio from these two

experiments for four different NAT types. Fig. 14(a) visualizes

how fast the capacity utilization ratio converges to 1.0 over

time after the client joins the network. Fig. 14(b) plots the

average utilization ratio (i.e., [

tcurrent capacity(t)dt] /

[

tmax cp(t)dt]). In both constant and adjustable cases, the

3The experiments were performed from one of Zattoos data centers inEurope, and there was sufficient uplink bandwidth available to support 4 *max cp.



13/14

13

results clearly exemplify the varied sharing performance across

different NAT types. Especially, the symmetric NAT type,

which is the most restrictive among the four, shows inferior

sharing ratio compared to the rest. Unfortunately, however, the

symmetric NAT type is the second most popular NAT type

for Zattoo population [5], and therefore can seriously affect

Zattoos overall sharing performance. There is a relatively

small number of peers running IP-restricted NAT-type, hence

its sharing performance has not been fine-tuned at Zattoo.

Reliability of NAT detection: To allow communications

among clients behind a NAT gateway (i.e., NATed clients),

each client in Zattoo performs a NAT detection procedure

upon player startup to identify its NAT configuration and

advertises it to the rest of the Zattoo network. Zattoos NAT

detection procedure implements a UDP-based standard STUN

protocol which involves communicating with external STUN

servers to discover the presence/type of a NAT gateway [9].

The communication occurs in UDP with no reliable transport

guarantee. Lost or delayed STUN UDP packets may lead

to inaccurate NAT detection, preventing NATed clients from

contacting each other, and therefore adversely affect theirsharing performance.

To understand the reliability and accuracy of STUN-based

NAT detection procedure, we utilize Zattoos session database

(Table I) which stores among other things NAT information,

public IP address, and private IP address for each session. We

assume that all the sessions associated with the same pub-

lic/private IP address pair would be generated under the same

NAT configuration. If sessions with the same public/private

IP address pair report inconsistent NAT detection results, we

consider those sessions as having experienced failed NAT

detection, and apply the majority rule to determine the correct

result for that configuration.

Fig. 15 plots the daily trend of NAT detection failure ratederived from ARD, Cuatro and SF2 during the month of June,

2008. The NAT detection failure rate at hour x is the numberof bogus NAT detection results divided by the total number of

NAT detections occurring on that hour. The total number of

NAT detections indicates how busy Zattoo system was through

out. The NAT detection failure rate grows from around 1.5%

at 2-3AM to a peak of almost 6% at 8PM. This means that at

the busiest times, the NAT type of about 6% of clients are not

determinable, leading to them not contributing any bandwidth

to the P2P network.

VII. RELATED WORKS

Aside from Zattoo, several commercial peer-to-peer systemsintended for live streaming have been introduced since 2005,

notably PPLive, PPStream, SopCast, TVAnts, and UUSee from

China, and Joost, Livestation, Octoshape, and RawFlow from

the EU. A large number of measurement studies have been

done on one or the other of these systems [7], [10][16].

Many research prototypes and improvements to existing P2P

systems have also been proposed and evaluated [4], [17][24].

Our study is unique in that we are able to collect network

core data from a large production system with over 3 million

registered users with intimate knowledge of the underlying

network architecture and protocol.

0

1

2

3

4

5

6

7

0 2 4 6 8 10 12 14 16 18 20 22 240.0E+

5.0E+4

1.0E+5

1.5E+5

2.0E+5

2.5E+5

3.0E+5

3.5E+5

NATDetectionFailureRate(%)

Num

berofNATDetections

Hour of Day

NAT failure rate# of NAT detections

Fig. 15. NAT detection failure rate.

P2P systems are usually classified as either tree-based push

or mesh-based swarming [25]. In tree-based push schemes,

peers organize themselves into multiple distribution trees [26]

[28]. In mesh-based swarming, peers form a randomly con-

nected mesh, and content is swarmed via dynamically con-

structed directed acyclic paths [21], [29], [30]. Zattoo is not

only a hybrid between these two, similar to the latest version

of Coolstreaming [4], its dependence on Repeater nodes also

makes it a hybrid of P2P network and a content-distribution

network (CDN), similar to PPlive [7].

VIII. CONCLUSION

We have presented a receiver-based, peer-division multi-

plexing engine to deliver live streaming content on a peer-to-

peer network. The same engine can be used to transparently

build a hybrid P2P/CDN delivery network by adding Repeater

nodes to the network. By analyzing large amount of usage data

collected on the network during one of the largest viewing

event in Europe, we have shown that the resulting network

can scale to a large number of users and can take good

advantage of available uplink bandwidth at peers. We have also

shown that error-correcting code and packet retransmission

can help improve network stability by isolating packet losses

and preventing transient congestion from resulting in PDM

reconfigurations. We have further shown that the PDM and

adaptive PDM schemes presented have small enough overhead

to make our system competitive to digital satellite TV in terms

of channel switch time, stream synchronization, and signal lag.

ACKNOWLEDGMENT

The authors would like to thank Adam Goodman for devel-oping the Authentication Server, Alvaro Saurin for the error

correcting code, Zhiheng Wang for an early version of the

stream buffer management, Eric Wucherer for the Rendezvous

Server, Zach Steindler for the Subsidy System, and the rest of

Zattoo development team for fruitful collaboration.

REFERENCES

[1] R. auf der Maur, Die Weiterverbreitung von TV- und Radioprogrammenuber IP-basierte Netze, in Entertainment Law (f. d. Schweiz), 1st ed.Stampfli Verlag, 2006.

[2] UEFA, Euro2008, http://www1.uefa.com/.



14/14

14

0

0.2

0.4

0.6

0.8

1

0 5 10 15 20 25 30

CurrentCapacityUtilizationRatio

Time Since Joining (Min)

Full coneIP-restricted

Port-restrictedSymmetric

(a) Constant maximum capacity

0

0.2

0.4

0.6

0.8

1

AverageCap

acityUtilizationRatio

Full-cone IP-restricted Port-restricted Symmetric

(b) Adjustable maximum capacity

Fig. 14. Uplink capacity utilization ratio.

[3] S. Lin and D. J. Costello, Jr., Error Control Coding, 2nd ed. PearsonPrentice-Hall, 2004.

[4] S. Xie, B. Li, G. Y. Keung, and X. Zhang, CoolStreaming: Design,Theory, and Practice, IEEE Trans. on Multimedia, vol. 9, no. 8,December 2007.

[5] K. Shami et al., Impacts of Peer Characteristics on P2PTV NetworksScalability, in Proc. IEEE INFOCOM Mini Conference, April 2009.

[6] Bandwidth-test.net, Bandwidth test statistics across different countries,http://www.bandwidth-test.net/stats/country/.

[7] X. Hei, C. Liang, J. Liang, Y. Liu, and K. W. Ross, Insights intoPPLive: A Measurement Study of a Large-Scale P2P IPTV System, inProc. IPTV Workshop, International World Wide Web Conference, 2006.

[8] B. Li et al., An Empirical Study of Flash Crowd Dynamics ina P2P-Based Live Video Streaming System, in Proc. IEEE GlobalTelecommunications Conference, 2008.

[9] J. R. et al, STUN - Simple Traversal of User Datagram Protocol (UDP)Through Network Address Translators (NATs), 1993, RFC 3489.

[10] A. Ali, A. Mathur, and H. Zhang, Measurement of Commercial Peer-to-Peer Live Video Streaming, in Proc. Workshop on Recent Advancesin Peer-to-Peer Streaming, August 2006.

[11] L. Vu et al., Measurement and Modeling of a Large-scale Overlay

for Multimedia Streaming, in Proc. Intl Conf. on HeterogeneousNetworking for Quality, Reliability, Security and Robustness, 2007.

[12] T. Silverston and O. Fourmaux, Measuring P2P IPTV Systems, inProc. ACM NOSSDAV, November 2008.

[13] C. Wu, B. Li, and S. Zhao, Magellan: Charting Large-Scale Peer-to-Peer Live Streaming Topologies, in Proc. ICDCS07, 2007, p. 62.

[14] M. Cha et al., On Next-Generation Telco-Managed P2P TV Architec-tures, in Proc. Intl Workshop on Peer-to-Peer Systems, 2008.

[15] D. Ciullo et al., Understanding P2P-TV Systems Through Real Mea-surements, in Proc. IEEE GLOBECOM, November 2008.

[16] E. Alessandria et al., P2P-TV Systems under Adverse Network Con-ditions: a Measurement Study, in Proc. IEEE INFOCOM, April 2009.

[17] S. Banerjee et al., Construction of an Efficient Overlay MulticastInfrastructure for Real-time Applications, in Proc. IEEE INFOCOM,2003.

[18] D. Tran, K. Hua, and T. Do, ZIGZAG: An Efficient Peer-to-PeerScheme for Media Streaming, in Proc. IEEE INFOCOM, 2003.

[19] D. Kostic et al., Bullet: High Bandwidth Data Dissemination Usingan Overlay Mesh, in Proc. ACM SOSP, Bolton Landing, NY, USA,October 2003.

[20] R. Rejaie and S. Stafford, A Framework for Architecting Peer-to-PeerReceiver-driven Overlays, in Proc. ACM NOSSDAV, 2004.

[21] X. Liao et al., AnySee: Peer-to-Peer Live Streaming, in Proc. IEEEINFOCOM, April 2006.

[22] F. Pianese et al., PULSE: An Adaptive, Incentive-Based, UnstructuredP2P live Streaming System, IEEE Trans. on Multimedia, vol. 9, no. 6,2007.

[23] J. Douceur, J. Lorch, and T. Moscibroda, Maximizing Total Uploadin Latency-Sensitive P2P Applications, in Proc. ACM SPAA, 2007, pp.270279.

[24] Y.-W. Sung, M. Bishop, and S. Rao, Enabling Contribution Awarenessin an Overlay Broadcasting System, in Proc. ACM SIGCOMM, 2006.

[25] N. Magharei, R. Rejaie, and Y. Guo, Mesh or Multiple-Tree: AComparative Study of Live P2P Streaming Approaches, in Proc. IEEE

INFOCOM, May 2007.[26] V. N. Padmanabhan, H. J. Wang, and P. A. Chou, Resilient Peer-to-Peer

Streaming, in Proc. IEEE ICNP, November 2003.

[27] M. Castro et al., SplitStream: High-Bandwidth Multicast in CooperativeEnvironments, in Proc. ACM SOSP, October 2003.

[28] J. Liang and K. Nahrstedt, DagStream: Locality Aware and FailureResilient Peer-to-Peer Streaming, in Proc. SPIE Multimedia Computingand Networking, January 2006.

[29] N. Magharei and R. Rejaie, PRIME: Peer-to-Peer Receiver-drIvenMEsh-based Streaming, in Proc. IEEE INFOCOM, May 2007.

[30] M. Hefeeda et al., PROMISE: Peer-to-Peer Media Streaming UsingCollectCast, in Proc. ACM Multimedia, November 2003.

Hyunseok Chang received the B.S. degree in elec-trical engineering from Seoul National University,Seoul, Korea, in 1998, and the M.S.E. and Ph.D.degrees in computer science and engineering fromthe University of Michigan, Ann Arbor, in 2001and 2006, respectively. He is currently a Member

of Technical Staff with the Network Protocols andSystems Department, Al- catel-Lucent Bell Labs,Holmdel, NJ. Prior to that, he was a Software Ar-chitect at Zattoo, Inc., Ann Arbor, MI. His researchinterests include content distribution network, mo-

bile applications, and network measurements.

Sugih Jamin received the Ph.D. degree in computerscience from the University of Southern California,Los Angeles, in 1996. He is an Associate Professorwith the Department of Electrical Engineering andComputer Science, University of Michigan, AnnArbor. He cofounded a peer-to-peer live streamingcompany, Zattoo, Inc., Ann Arbor, MI, in 2005.Dr. Jamin received the National Science Founda-tion (NSF) CAREER Award in 1998, the Presiden-tial Early Career Award for Scientists and Engineers

(PECASE) in 1999, and the Alfred P. Sloan ResearchFellowship in 2001.

Wenjie Wang received the Ph.D. degree in com-puter science and engineering from the Universityof Michigan, Ann Arbor, in 2006. He is a ResearchStaff Member with IBM Research China, Beijing,China. He co-founded a peer-to-peer live streamingcompany, Zattoo Inc., Ann Arbor, MI, in 2005, basedon his Ph.D. dissertation, and was CTO. He alsoworked in Microsoft Research China and OracleChina as a visiting student and consultant intern.


live streaming with

Documents