Communications and Network, 2013, 5, 144-149
http://dx.doi.org/10.4236/cn.2013.53B2028 Published Online September 2013 (http://www.scirp.org/journal/cn)
A Novel Cooperative Multicast Scheme
Based on Fountain Code*
Qinghe Du1,2, Pinyi Ren1, Jincheng Lu1, Zhigang Chen1
1School of Electronic and Information Engineering, Xi’an Jiaotong University, China
2National Mobile Communications Research Laboratory, Southeast University, China
Received May, 2013
ABSTRACT
Multicast is an efficient way to support emerging multimedia services over wireless network. Fountain codes are used
in multicast systems to enable a robust transmission without CSI feedback and ARQ. We propose a cooperative multi-
cast scheme based on fountain code to improve the performance of multicast. The users are coordinated with each other
to decode the message at different time slots within the data transmission of a multicast session. Specifically, we take
the local channel state information (CSI) and the local residual energy information (REI) into consideration, and apply a
relay-selection and power-allocation strategy in our cooperative multicast scheme to prolong the network lifetime, while
keeping the transmission delay as low as possible. The simulation results show that the proposed scheme can achieve a
good tradeoff between transmission delay and network lifetime.
Keywords: Cooperative Multicast; Fountain Code; Residual Energy; Lifetime
1. Introduction
Recently, many important applications, such as video
conferencing, mobile TV, software updating, etc, require
the support of multicast. Multicast communication,
which can simultaneously deliver data stream to many
subscribers over a common channel, has become one of
the main features of the next generation of mobile com-
munication, such as IPTV over WiMax [1] and MBMS
in LTE [2]. In multicast systems, data transmission only
involves the downlink. A large number of multicast sub-
scribers are usually located in different places and ex-
perience different channels fading, so their channel quali-
ties may be profoundly different. To ensure fairness, the
multicast transmission rate is determined by the user with
the worst channel quality, which limits the efficiency of
multicast transmission.
Another challenge the multicast system faces is how to
transmit reliably when the channel conditions of the mul-
ticast group users are different and dynamically changing.
Conventional fixed-rate forward error correction (FEC)
code can recover some loss of packets, but it has to be
designed to carry a rate below the capacity of the channel,
which means the channel conditions should be known at
the transmitter. In multicast system, the instantaneous
channel state feedbacks for large number of users will be
a huge overhead. Furthermore, for any fixed-rate FEC
coding, the error or loss is still unavoidable during trans-
missions in time-varying wireless fading channels. When
packet loss occurs, unicast transmissions can use auto-
matic repeat request (ARQ) techniques [3] to handle it,
with decreased throughput induced. However, ARQ tech-
niques are not suitable to multicast system, since differ-
ent users may lose different packets, the source has to
repeat a large number of packets for different users.
Fountain codes, or rateless codes, can adapt its rate to
the channel realization and require no knowledge of
channel state information at the transmitter [4-5]. The
transmitter can generate as many encoding symbols as
needed, depending on the instantaneous quality of the
channel of the receivers. The receivers keep accumulat-
ing mutual information from the source and relays, until
they can successfully decoding the message. As the
fountain code enables a reliable multicast transmission
without CSI feedback and ARQ, it has been introduced
into the standards of MBMS and DVB [6,7]. The users
with different channel qualities have different speeds to
accumulate the mutual information. A straightforward
approach to improve the performance of multicast com-
*This paper is supported by the National Natural Science Foundation o
f
China under Grant No. 61102078, the National Science and Technol-
ogy Major Project under Grant No. 2010ZX03003-004-01, the Spe-
cialized Research Fund for the Doctoral Program of Higher Education
under Grant No.20110201120014, the Open Research Fund of Na-
tional Mobile Communications Research Laboratory, Southeast Uni-
versity (No.2011D10), and the Fundamental Research Funds for the
Central University.
C
opyright © 2013 SciRes. CN

Q. H. DU ET AL. 145
munication is let the users decode message first help the
users decode message after. In this paper, we introduce a
cooperative scheme based on fountain code to improve
the performance of multicast communication, which is
different from previous cooperative schemes in point-to-
point scenarios [8,9].
Due to user cooperation and fountain coding, we can
improve the efficiency and reliability of multicast trans-
mission. Most user terminals are battery-powered, and
forwarding message for other users will consume a lot of
power. Letting all decoded users forward message with
maximum transmission power can obviously bring great
performance gain for multicast transmission, but the
shortcoming of this approach is that it is not an energy
efficient way and will shorten the network lifetime,
which is measured by the first node running out of its
energy in the network [10]. We take the residual energy
and local channel state information of users into consid-
eration, and apply a relay selection and transmission
power allocation algorithm in our cooperative multicast
scheme. Simulation results show that the proposed
scheme can achieve a good tradeoff between transmis-
sion delay and network lifetime.
The rest of this paper is organized as follows. The
system model is introduced in Section 2. Section 3 de-
scribes the proposed cooperative multicast scheme. Si-
mulations results are presented and discussed in Section
4, illustrating the performance of the proposed scheme.
Finally, conclusions are drawn in Section 5.
2. System Model
2.1. Network Model
Consider a wireless multicast network with a circular cell
of radius as shown in Figure 1. The BS is located at the
center of the cell and multicasts to M users. The M users
are randomly located in the cell with uniformly distribu-
tion. The location of user can be deter-
mined by (,
(1 )ii
M
),
ii
d

where 0i dR

 and 02
i

  .
In the paper, we consider the large-scale path loss at-
tenuation, small-scale fading and additive white Gaussian
noise (AWGN) in the wireless channel, the link between
any two nodes are independent. Path-loss attenuation is
determined by the geographical environment and dis-
tance between the receiver and the transmitter, so the
path loss between user can be modeled as ,ij
,,
()
ij ij
PL rr


 (1)
where

is the path loss parameter, ,ij is the trans-
mission distance between user , given by
r
,ij
222
,
()2 cos()
ijiji jij
rdddd

  (2)
Small-scale fading is caused by multiple versions of a
transmitted signal with different delays and independent
Figure 1. Network model.
of the path loss, we commonly use the Rayleigh flat fad-
ing to describe it, which can be modeled as a zero-mean
circularly symmetric complex Gaussian random variable
with unit variance. All users in the network are in the
half duplex mode; they cannot transmit and receive on
the same frequency band at the same time.
2.2. Fountain Code
Fountain code are applied in BS, a source file is split into
K packets with equally size, then the encoder randomly
chooses a certain number of source packets and com-
bines them to an encoded packet by an exclusive-OR
operation, then the encoded packet is broadcast to the
users through wireless channel. This procedure can be
repeated continuously until the BS has received the ACK
signals of all multicast group users. Due to a cyclic re-
dundancy check (CRC), the receiver knows the encoded
packet is either correctly or erroneous received. If errors
occur, this packet will be dropped by the receiver. A
fountain code decoder tries to recover the source packets
from the received encoded packets after accumulating a
sufficient number of correct encoded packets, usually this
sufficient number is slightly higher than K, an additional
overhead of about 5 − 10% of K is typically required [11].
As users with different conditions have different packet
loss rate, they need different transmission time to receive
a sufficient number of packets and decode the message.
2.3. Cooperative Multicast Scheme
Make different users collaborate to accomplish the data
transmission is suitable for multicast scenarios based on
two reasons. Firstly, users are the part of multicast group,
so the security and incentive are no need to concern. Se-
condly, the information-theoretic results of [12] suggest
that cooperation can lead to significant diversity en-
hancement and reduce the packet losses. In our scheme, a
multicast session consists of two kinds of transmission

Q. H. DU ET AL.
146
modes, (i) point-to-multipoint (i.e. direct multicast), (ii)
multipoint-to-multipoint (i.e. cooperative multicast). At
the beginning of the multicast session, BS broadcast the
encoded packets to multicast group users with a fixed
transmission rate, and every user keeps accumulating
mutual information from BS. It is a point-to-multipoint
transmission mode as conventional multicast. For user
located at
i
(,),
ii
d

the received signal from BS is given
by
,,
DM
is sisifi
yPrhx
 
n (3)
with received SNR
0
2
,,
,
s
si si
DM
i
Ph r
N




where the superscript “DM” means direct multicast,
s
P
is the transmission power of BS, ,
s
i is the distance be-
tween BS and user , ,
r
i
s
i is the Rayleigh flat fading,
,
h
(0,1),
si
hCN
f
x
is a unit power symbol encoded by
fountain code, the i is the additive white Gaussian
noise with a variance of 0. If the received SNR is
higher than a threshold 0
nN

, we assumed the receiver is
able to decode the received message with a negligible
probability of error, otherwise we assume the receiver
cannot decode the packet correctly, and drop this packet.
Obviously, some users with good channel conditions will
receive enough correct encoded packets and decode the
multicast message before other users whose channel
conditions are not good. These good users have the abil-
ity to help the users that are still accumulating the mutual
information from BS; we called these users candidate
relays (CR). By relay selection, we choose some relays
out of candidate relays to forward message; we use set D
to represent relays. Then users in D can re-encode the
packets using the same rule as the BS, and simultane-
ously broadcast the encoded symbols with the BS on the
common channel. We assume all candidate relays in-
volved in cooperation are synchronized and the delay
spread of arriving signals is negligible, which is valid in
narrow-band wireless communication. When users in D
begin to forward packets, it is a multipoint-to-multipoint
cooperative multicast. The users with bad channel condi-
tions will receive message from BS and cooperative us-
ers, the received signal is given by
,, ,,
CM
issisiji ji jfi
jD
yPrh Prhx
 





n




(4)
with received SNR
2
,, ,,
0
ssisijij ij
jD
CM
i
Pr hPr h
N







(5)
where the superscript “CM” means cooperative multicast,
j
P is the transmission power of relay in set , the
value of
jD
j
P will be determined in power allocation sec-
tion. We do not use orthogonal channels to forward mes-
sage is based on the considerations of spectrum
efficiency and multicast channel sharing behavior.
If all decoded users forward message with the maxi-
mum transmission power for other users have not yet
decode the message, clearly, significant gain will be
achieved. But as the multicast session continue, more and
more users have decoded the message, let all of them
forward message for only several users is a waste of en-
ergy. And some users with good channel conditions for-
ward for others with maximum transmission power for a
long time, their residual energy will be running out
quickly, which is the situation we try to avoid. So we
proposed a relay selection and power allocation strategy
to coordinate with a higher energy efficiency and prolong
the network lifetime.
3. Relay Selection and Power Allocation
Strategy
If all decoded users forward message with the maximum
transmission power for other users have not yet decode
the message, clearly, significant gain will be achieved.
But as the multicast session continue, more and more
users have decoded the message, let all of them forward
message for only several users is a waste of energy. And
some users with good channel conditions forward for
others with maximum transmission power for a long time,
their residual energy will be running out quickly, which
is the situation we try to avoid. So we proposed a relay
selection and power allocation strategy to coordinate
with a higher energy efficiency and prolong the network
lifetime.
3.1. Relay Selection
In multicast scenario, there are some users that have de-
coded the message, but there are no other users located
nearby, or users nearby have decoded the message too. If
these users forward message, its cooperation gain is small.
So we proposed a relay selection procedure to filter these
kinds of users to improve the energy efficiency. The
procedure is shown in Figure 2, contains the following
steps.
1) BS encode the input packets with fountain code and
broadcast encoded symbols on multicast channel, multi-
cast group users receive the signals and accumulate cor-
rect fountain code symbols until they can decode the re-
ceived message. Then these users send out an ACK sig-
nal on a feedback channel to inform the BS that it has the
correctly receive the input packets and become candidate
relays (CR).

Q. H. DU ET AL. 147
Figure 2. Relay selection.
2) The users that have not decode the message listen
on the feedback channel to monitor the ACK signal. The
SNR of the received ACK signals are measured and
compare to a predetermined SNR threshold. If the SNR
of received ACK signal is lower than the threshold, it
indicates that the link conditions between the two users are
not good enough to forward message. If the SNR of re-
ceived ACK signal is greater than, the users will send a sig-
nal to the user who just completes the decoding for help,
which includes its own ID and the ID of candidate relay.
3) If a candidate relay hear the signal ask it for help, it
will recode the ID of the users who ask for help, and be-
came a relay involve in cooperation. It begins to
re-encode the decoded packets and forward them simul-
taneously with BS and other relays.
4) When the user completes the decoding with the help
of BS and relays, it will broadcast an ACK signal too.
The relays monitor the ACK signals on the feedback
channel and remove the ID, if the ACK signal is from the
user who asks for the relay to help before. When the re-
lay find all users it help have decode the message, this
relay will stop forwarding message and exit cooperation
in this multicast session.
3.2. Power Allocation
Let be the residual energy of user at the be-
ginning of the time slot, is the transmis-
sion power of useriin the period of the time slot.
So we can get iis and
[]
i
em i
-t h
,

-t hm
i
em
[]
i
Pm
[Pm
m
]T[1]
[]em
s
T is
the length of each time slot. At the beginning of a time
slot, the selected user determine the transmission power
by its residual energy. Assuming that there are L levels of
residual energy, 1liL ([Ee]m),

 
then the user
can obtain its energy index(EI) during time slot,
i, if 1l

[]
li
em
i
-thm
EI l[ ]m

 
, , 0
(1,lL)0


.
Accordingly, the transmission power of users involved in
cooperation can be equally divided into L discrete power
levels, We assume that and the
interval between two levels is
min
(Pm ,Pax)[]
i
Pm ,
,

max min
1
P 

EI
, ,.
P
L

 (6)
When the user i determine by compare its
residual energy to set 12
[ ]
im
.., ),
L
E=(


(E
 then the trans-
mission power in time slot is determined by -thm
min
[] []1)
ii
Pm Pm


 (7)
4. Simulation Results
The performance of the proposed cooperative multicast
scheme is evaluated using Monte Carlo simulations
through MATLAB. In our simulations, we randomly
generate M user locations uniformly distributed inside a
circle with 100.R

The path loss parameter is 2.6,


channel gains are generated independently following the
complex Gaussian distribution and vary inde-
pendently in each time-slot.
(0,1)CN
To analyze the network lifetime performance of pro-
posed scheme, we assume the transmission rate of multi-
cast data stream is 4 on unit bandwidth, and then if the
received SNR is higher than threshold 015,

 the us-
ers can accumulate a correct packet. And we assume if
users can accumulate 1000 packets, they can decode the
message. The transmission power of BS is 1, and remains
unchanged during the multicast session. The transmission
power range of a relay user is (0.01, 0.1), and there are
10L

power levels, so the interval of power levels is
0.01.


The initial residual energy of all user is 0
we assume 0
,e
100,e

the length of each time-slot
1,
s
T

the relay selection threshold is 115.

 We also
simulate the simple cooperative multicast scheme that
without relay selection and power allocation, the users
begin to forward the message with maximum transmis-
sion power after they have decoded the message. We set
100M,

and varied the 0s rom 64 dB to 70 dB.
The result is showed in Figure 3; the vertical axis is the
network lifetime normalized by the maximum value in
the simulations. Figure 3 shows that proposed multicast
scheme perform much better than simple cooperative
multicast scheme, the network lifetime is prolonged
about triple times in our scheme. This gain is attributed
to the relay selection and power allocation based on the
local channel state information and the local residual
energy information. By relay selection, the candidate
relays with bad link conditions connecting to other users
are not involved in the cooperation and the relays that
have help the users nearby decode the message will exit
cooperation too. By power allocation, the relays with
abundant residual energy can support the cooperative
transmission with high level of transmission power, and
the relays with poor residual energy will decrease the
power level to save energy.
/NP

Q. H. DU ET AL.
148
Figure 3. The performance of Network lifetime.
We also want to see the impact of the relay selection
and power allocation to the multicast transmission delay
performance. For comparison, the performance of the
direct multicast is simulated as well as two cooperative
multicast schemes under the same simulation conditions.
We know the multicast transmission delay is determined
by the time the worst user needs to decode the encoded
packets in a multicast session. When the users are with
different residual energy, the performance of transmis-
sion delay is different. At the beginning of simulation,
every user have the same initial residual energy, as the
multicast transmission continue, the user that running out
residual energy will emerge. This procedure will last
several multicast sessions; we record the multicast
transmission delay of every multicast session and calcu-
late the average transmission delay. The result is showed
in Figure 4 the vertical axis is the transmission delay
normalized by the maximum value in the simulations.
Benefiting form cooperative diversity, the two coopera-
tion schemes perform better than direct multicast scheme
in the aspect of transmission delay. The performance
gain is high especially in low SNR region, that is because
some users is hard to receive correct packets without help
of relays in low SNR region, and it seriously limit the
performance of transmission delay. When the SNR is
high, these users can receive the correct packets easier
even without the help of relays, so the cooperation gain is
decreased in high SNR region. Compare the two coop-
erative multicast schemes, we will find that the perform-
ance gap between them is relatively small. That means
the relay selection and power allocation have small im-
pact on the transmission delay in statistical average as-
pects. Some candidate relays are inefficient for other us-
ers, they cannot decrease too much performance gain if
we do not let them involve in the cooperation.
Figure 4. The performance of average transmission delay.
5. Conclusions
In this paper, we investigated a novel cooperative multi-
cast scheme based on fountain code over wireless net-
works. By introducing fountain code, the users with dif-
ferent packet loss rates will decode the message at dif-
ferent time slots. The user who decodes the message be-
comes candidate relay, and decides whether to join the
cooperation or not based on the link quality between it-
self and users. The transmission power of relays is de-
termined by the local residual energy information. The
simulation results show that our scheme can prolong
network lifetime of multicast system, at the price of
slight loss of average transmission delay performance.
REFERENCES
[1] J. She, P. Ho and L. Xie, “IPTV over WiMax: Key Suc-
cess Factors Challenges, and Solutions,” IEEE Commu-
nications Magazine, Vol. 45, No. 8, 2007, pp. 87-93.
doi:10.1109/MCOM.2007.4290319
[2] S. Kota, Y. Qian, E. Hossain and R. Ganesh, “Advances
in Mobile Multimedia Networking and QoS,” IEEE
Communications Magazine, Vol. 45, No. 8, 2007, pp.
52-53.doi:10.1109/MCOM.2007.4290314
[3] Y. Omar, M. Youssef and H. El Gamal, “ARQ Secrecy:
From Theory Practice,” in Proc. IEEE Information The-
ory Workshop ITW 2009, Oct.11-16, 2009, pp. 6-10.
[4] M. Luby, “LT Codes,” in Proc. 43rd Annual IEEE Symp.
Found. Computer Sci., Nov. 2002, pp. 271-280.
[5] A. Shokrollahi, “Raptor Codes on Symmetric Channels,”
in Proc. IEEE Int. Symp. Inform. Theory, June-July 2004,
p. 38.
[6] Universal Mobile Telecommunications System (UMTS)
Multimedia Broadcast Service (MBMS) Protocols and
Codecs (3Gpp TS 26.346 version6.12.0 Release 6), 2008,
pp. 90-105.

Q. H. DU ET AL.
149
[7] Digital Video Broadcasting (DVB) IP Datacast over
DVB-HContent Delivery Protocols (ETSI TS 102472
v1.1.1), June 2006, pp. 57-72.
[8] J. Castura and Y. Y. Mao, “Rateless Coding for Wireless
Relay Channels,” IEEE Transactions on Wireless Com-
munications, Vol. 6, No. 5, 2007, pp. 1638-1642.
[9] W. J. Lei, X. M. Li, X. Z. Xie and G. J. Li, “Wireless
Cooperative Relay System Using Digital Fountain
Codes,” International Conference on Communications,
Circuits and Systems, ICCCAS 2009.
[10] V. Shah-Mansouri and V. W. S. Wong, “Life-
time-Resource Tradeoff for Multicast Traffic in Wireless
Sensor Networks,” IEEE Transactions on Communica-
tions, Vol. 9, No. 6, 2010, pp. 1924-1934.
[11] R. Budde, S. Nowak and R. Kays, “Reliable Broadcast
Transmission in Vehicular Networks Based on Fountain
Codes,” IEEE 73rd Vehicular Technology Conference
(VTC Spring), 2011, pp. 1-5.
[12] A. Sendonaris, E. Erkip and B. Aazhang, “User coopera-
tion diversity. Part I. System Description,” IEEE Trans-
actions on Communications, Vol. 51, No. 11, 2003, pp.
1927-1938.doi:10.1109/TCOMM.2003.818096