Académique Documents
Professionnel Documents
Culture Documents
in LTE systems
G. Araniti
, M. Condoluci
, A. Molinaro
, A. Iera
, J. Cosmas
sS
rs = R
5: Compute
1
=
sS
bsrs|Us|
6: t = 2
7: while t |S| do
8: Find i, j S that maximise t
9: t =
t1
b
i
r
i
|U
i
| b
j
r
j
|U
j
| + bnrn|Un|
10: where n = arg min(b
i
, b
j
), rn = r
i
+ r
j
, and Un = U
i
U
j
11: if t >
t1
then
12: Update S = S {i, j} {n}
13: Update R
14: rn = r
i
+ r
j
, with n = arg min(b
i
, b
j
)
15: r
i
= 0, r
j
= 0
16: t = t + 1
17: else
18: Stop
19: end if
20: end while
Moreover, compared to SMS, L-SMS guarantees to perform
subgroup formation also when few frequency resources are
available for multicast services. The latter proposed policy,
aimed at achieving better near-optimal performance compared
to L-SMS, deals with the RB assignment phase in order to
improve the throughput experienced by MBMS users.
A. The L-SMS
In analogy to SMS, L-SMS manages the subgroup formation
according to the channel conditions experienced by multicast
users, i.e., the CQI feedback in LTE systems. Obviously, L-
SMS is designed to maximize the sum of data rates assigned
to all the multicast users in a group, i.e., the ADR. L-SMS
proceeds by iterative steps; at each step it tries to maximize the
ADR of the enabled subgroups, and it stops when no further
improvement is achieved or no other subgroups are available.
L-SMS is summarized in Table II. At the rst iteration, L-
SMS gathers all the users reporting the same CQI value in
the same subgroup. Such set of users is denoted with U
s
, i.e.,
U
s
contains all the users experiencing a CQI value equal to s.
The number of created subgroups is equal to the number of
different measured CQI levels, i.e., S, and it is bounded by the
maximum number of CQI levels dened by LTE system, i.e.,
S 15. The created subgroup conguration is denoted with
the R vector. If r
s
R (with s = 1, 2, . . . , S) is greater than
zero, the subgroup related to the s-th MCS is enabled and r
s
represents the number of RBs assigned to such subgroup. The
data rate achieved by each subgroup depends on the transport
block size of the MCS related to the subgroup, i.e., b
s
[1],
and the number of assigned resources, i.e., r
s
. At the end of
this step, the initial value of the target cost function, the ADR,
is computed. L-SMS searches through all the combinations of
two subgroups to merge and selects the one with the highest
ADR. If by merging the two selected subgroups the target
function increases compared to the previous step, then L-
SMS will merge the users and the resources of the selected
subgroups into a new subgroup. This process is iterated until
no further improvement in the cost function value is achieved
or no subgroup to merge is available.
The overall complexity of L-SMS is equal to O(S
4
);
therefore, L-SMS is more feasible for real implementation
compared to the ESS, which requires O(S
R
) [10], where R
is the overall number of available resources. It is worth noting
that, differently from SMS [10], the computational cost of L-
SMS does not depend on the number of users in the multicast
group, hence it is more scalable than SMS.
B. The L-eSMS
Another key issue in the subgroup merging approach is re-
lated to the distribution of the R available frequency resources
among the subgroups enabled at the rst iteration. Both SMS
and L-SMS randomly distribute the resources among the
formed subgroups (refer to line 4 in Table II). This choice may
cause inefciency especially in scenarios where the multicast
users experience heterogeneous channel qualities. With the aim
of improving the ADR of the subgroup conguration enabled
at the rst step, we extend the proposed L-SMS scheme by
dening the L-eSMS, tailored to better exploit the multi-user
diversity in the resource allocation phase.
The novel resource distribution strategy is based on the
denition of a weight
s
, dened as follows:
s
=
b
s
|U
s
|
iS
b
i
|U
i
|
(1)
Once
s
is computed for each subgroup, the RB assignment is
performed as follows. At each group is assigned at every sub-
group enabled in the rst iteration. The remaining resources,
i.e., RS, are allocated to the subgroup with the highest
s
value, i.e, s
= arg max
s
. The RB allocation of step 4 in
Table II can be recast as follows:
r
s
=
R S + 1 if s = arg max
s
1 otherwise
(2)
where S 15 is the number of subgroups enabled at the rst
step. Eq. (2) guarantees that each enabled subgroup is served
with at least one frequency resource, whereas the subgroup
which achieves the higher amount of resources is selected in
order to guarantee improved ADR performance in the initial
conguration. As a consequence, L-eSMS aims at assigning
more resources to more populated subgroups or to subgroups
supporting higher MCSs.
The overall complexity of L-eSMS is equal to O(S
4
), i.e.,
the proposed RB assignment strategy does not change the
complexity cost compared to L-SMS.
V. PERFORMANCE ANALYSIS
The performance analysis has been conducted in accordance
with the guidelines dened in [17]. The channel quality of
each multicast member is evaluated in terms of Signal to
Interference and Noise Ratio (SINR) experienced over each
sub-carrier [18]:
SINR
i
=
P
0
PL
0
h
0
NBS
j=1
(P
j
PL
j
h
j
) +N
o
(3)
where P
j
, PL
j
and h
j
are the transmission power, the path
loss, and the small scale fast fading of the link between the
UE and the j-th base station (j = 0 indicates the serving base
station, j > 0 the interfering ones); N
o
is the noise power.
Once the SINR value for all the sub-carriers is collected, the
effective SINR is obtained through the Exponential Effective
SIR Mapping (EESM):
SINR
eff
= ln
1
N
sub
N
sub
i=1
e
SINR
i
(4)
being N
sub
the overall number of sub-carriers and a scaling
factor used to adjust the mismatch between the actual and
the predicted block error rate (BLER). Finally, the effective
SINR is mapped onto the CQI level related to the MCS which
ensures a BLER smaller than 10% [18]. More details on the
LTE system settings are listed in Table III.
TABLE III
MAIN SIMULATION ASSUMPTION
Parameters Value
Cell layout 3GPP Macro-cell case #1
Cell radius 500 m
Distance attenuation 128.1+37.6*log(d), d [km]
Shadow fading Log-normal,0 mean, = 8 [dB]
Fast Fading ITU-R PedB (extended for OFDM)
Scheduling frame 10 ms
TTI 1 ms
Carrier frequency 2 GHz
eNodeB transmit power 20 W, 13 dB
Maximum antenna gain 11.5 dB
Thermal Noise -100 dBm
Fig. 1. User distribution within the cell.
We compared the performance of the proposed L-SMS and
L-eSMS policies with that of ESS and CMS. Simulations have
been carried out by addressing a uniform user distribution
scenario, where the users are uniformly distributed within the
cell coverage area (Fig. 1). We consider two simulation cases:
Scenario A, where we considered a variable number of
users (from 10 to 100) and a xed bandwidth deployment
scenario of 100 RBs;
Scenario B, where we varied the number of RBs (from
15 to 100) in a multicast group conguration with 100
members.
10 20 30 40 50 60 70 80 90 100
0
250
500
750
1000
1250
1500
A
D
R
[
M
b
p
s
]
Multicast Group Size
CMS
ESS
LSMS
LeSMS
(a) Aggregate Data Rate
10 20 30 40 50 60 70 80 90 100
0
20
40
60
80
100
A
D
R
[
%
]
Multicast Group Size
LSMS
LeSMS
(b)
ADR
Fig. 2. Simulation results in Scenario A.
Outputs are achieved by averaging a sufcient number of
simulation results to obtain 95% condence intervals.
Fig. 2 shows the comparison of CMS, ESS, L-SMS, and
L-eSMS schemes in the Scenario A. Subgrouping strategies
considerably improve the ADR performance, as depicted in
Fig. 2(a), with respect to the CMS. Indeed, the ADR of CMS
varies from 50 up to 260 Mbps, whereas that of ESS increases
from 157.2 to 1305 Mbps. The L-SMS achieves an ADR
performance which varies from 108 to 838 Mbps. Finally,
the ADR of L-eSMS increases from 154 Mbps up to 1247
Mbps. It emerges that L-eSMS offers performance closer to
optimal compared to L-SMS. This is more evident in Fig. 2(b),
which shows the results in terms of the mismatch between the
optimal ADR value (that achieved by ESS policy) and the one
of the two addressed low complexity approaches, i.e.,
ADR
.
The performance of L-SMS varies from 31% to 35%, and
this demonstrates that L-SMS does not achieve performance
very close to the optimal value. Focusing on the proposed L-
eSMS, the
ADR
varies from a minimum value of 2% up to
a maximum value of 4.5%. This underlines that L-eSMS can
considerably improve the L-SMS approach in order to achieve
near-optimal performance.
Fig. 3 shows the comparison of the addressed policies in the
Scenario B. As shown in Fig. 3(a), CMS offers an ADR which
varies from 39 up to 260 Mbps. Again, the ESS guarantees
the highest ADR results. Focusing on the sub-optimal policies,
the L-SMS achieves an ADR which varies from 125 to 838
Mbps, whereas the ADR of L-eSMS increases from 130 up
to 1247 Mbps. Also in this scenario, L-eSMS outperforms the
L-SMS. Fig. 3(b) depicts the performance in terms of
ADR
of the two proposed near-optimal schemes. The performance
of L-SMS varies from 32% to 36%. Focusing on the proposed
L-eSMS, the
ADR
decreases from a maximum value of
30% up to a minimum value of 4.5%. It is worth noting
that
ADR
of L-SMS does not meaningfully vary as the
available channel bandwidth increases, whereas the one of
L-eSMS decreases when the number of RBs becomes large.
These results demonstrate that, also in the Scenario B, L-SMS
does not achieve performance very close to the optimal value
whereas the L-eSMS approach is well designed in order to
achieve near-optimal performance compared to L-SMS and the
gain is more evident when the number of available resources
is large.
VI. CONCLUSIONS AND FUTURE WORK
In this paper we addressed the design of resource allocation
schemes for multicast data delivery in LTE systems. We
focused on the subgrouping approach, tailored to overcome the
limitations of the conventional approaches while guaranteeing
system capacity maximization. We considered the issue to
reduce the computational burden in link adaptation procedures.
We proposed two low-complexity strategies, i.e., L-SMS and
L-eSMS, that efciently exploit the potentialities of LTE
in order to achieve better scalability than other approaches
proposed for OFDMA-based networks. Indeed, L-SMS and L-
eSMS guarantee a computational cost which is not inuenced
by the multicast group size and by the number of resources
available for MBMS session delivery. Through simulations,
the effectiveness of the proposed L-eSMS in achieving near-
optimal performance compared to schemes in literature is
underlined in different cell deployment and multicast group
conguration scenarios.
Future enhancement of this work is related to the design of a
novel low-complex subgrouping strategy tailored to decrease
the computational burden and to achieve performance more
close to the optimal one.
ACKNOWLEDGEMENT
The research of Massimo Condoluci is supported by Eu-
ropean Union, European Social Fund and Calabria Regional
Government. This paper reects the views only of the authors,
and the EU, and the Calabria Regional Government cannot
be held responsible for any use which may be made of the
information contained therein.
15 25 50 75 100
0
250
500
750
1000
1250
1500
A
D
R
[
M
b
p
s
]
Number of RBs
CMS
ESS
LSMS
LeSMS
(a) Aggregate Data Rate
15 25 50 75 100
0
20
40
60
80
100
A
D
R
[
%
]
Number of RBs
LSMS
LeSMS
(b)
ADR
Fig. 3. Simulation results in Scenario B.
REFERENCES
[1] 3GPP, TS 36.300, Evolved Universal Terrestrial Radio Access (E-
UTRA) and Evolved Universal Terrestrial Radio Access Network (E-
UTRAN), Rel. 11, September 2012.
[2] A. Iera, A. Molinaro, S. Polito, and G. Ruggeri, Coordinated Multi-
hop Scheduling in IEEE 802.11e Wireless Ad Hoc Networks, IEEE
17th International Symposium on Personal, Indoor and Mobile Radio
Communications (PIMRC), pp. 1-5, September 2006.
[3] S. Frattasi, R. L. Olsen, M. De Sanctis, F. H. P. Fitzek, and R. Prasad,
Heterogeneous services and architectures for next-generation wireless
networks, 2nd International Symposium on Wireless Communication
Systems (ISWCS), pp. 213-217, September 2005.
[4] E. Cianca, M. De Sanctis, and M. Ruggieri, Convergence Towards 4G:
a Novel View of Integration, Wireless Personal Communications, vol.
33, no. 3-4, pp. 327-336, June 2005.
[5] 3GPP, TS 36.440, General aspects and principles for interfaces sup-
porting Multimedia Broadcast Multicast Service (MBMS) within E-
UTRAN, Rel. 11, September 2012.
[6] W. Rhee, and J. Ciof, Increase in capacity of multiuser OFDM systems
using dynamic subchannel allocation, IEEE 51st Vehicular Technology
Conference (VTC-Spring), vol. 2, pp. 1085-1089, 2000.
[7] P. K. Gopala, and H. E. Gamal, Opportunistic multicasting, Thirty-
Eighth Asilomar Conference on Signals, Systems and Computers,
November 2004.
[8] G. Araniti, V. Scordamaglia, A. Molinaro, A. Iera, G. Interdonato, and
F. Span` o, Optimizing point-to-multipoint transmissions in high speed
packet access networks, IEEE International Symposium on Broadband
Multimedia Systems and Broadcasting (BMSB), pp. 1-5, June 2011.
[9] G. Araniti, M. Condoluci, and A. Iera, Adaptive multicast scheduling
for HSDPA networks in mobile scenarios, IEEE International Sympo-
sium on Broadband Multimedia Systems and Broadcasting (BMSB), pp.
1-5, June 2012.
[10] C. K. Tan, T. C. Chuah, and S. W. Tan, Adaptive multicast scheme for
OFDMA-based multicast wireless systems, Electronics Letters, vol. 47,
no. 9, pp. 570-572, April 2011.
[11] I. Bisio, A. Delno, G. Luzzati, F. Lavagetto, M. Marchese, C. Fra,
and M. Valla, Opportunistic estimation of television audience through
smartphones, International Symposium on Performance Evaluation of
Computer and Telecommunication Systems (SPECTS), pp. 1-5, July
2012.
[12] I. Bisio, F. Lavagetto, M. Marchese, and A. Sciarrone, Smartphone-
based user activity recognition method for health remote monitoring
applications, 2nd International Conference on Pervasive Embedded
Computing and Communication Systems (PECCS), pp. 200-205, 2012.
[13] 3GPP, TS 36.213, Evolved Universal Terrestrial Radio Access (E-
UTRA): Physical layer procedures, Rel. 10, March 2012.
[14] L. Militano, M. Condoluci, G. Araniti, and A. Iera, Bargaining Solu-
tions for Multicast Subgroup Formation in LTE, IEEE 76th Vehicular
Technology Conference (VTC-Fall), pp. 1-5, September 2012.
[15] G. Araniti, M. Condoluci, A. Molinaro, and S. Pizzi, Radio-Aware Sub-
groups Formation for Multicast Trafc Delivery in WiMAX Networks,
IEEE 23rd International Symposium on Personal, Indoor and Mobile
Radio Communications (PIMRC), pp. 477-482, September 2012.
[16] L. Militano, M. Condoluci, G. Araniti, and A. Iera, Multicast Service
Delivery Solutions in LTE-Advanced Systems, IEEE International
Conference on Communications (ICC), June 2013.
[17] 3GPP, TS 36.201, LTE physical layer; General description, Rel. 11,
December 2012.
[18] C. Mehlf uhrer, M. Wrulich, J. Ikuno, B. Colom, D. Bosanska, and
M. Rupp, Simulating the long term evolution physical layer, 17th
European Signal Processing Conference (EUSIPCO), pp. 1471-1478,
August 2009.