A CoMP Simulation Model Based on UE-Centric Soft Handover Grouping Algorithm

- 1 -
A CoMP Simulation Model Based on UE-Centric
Soft Handover Grouping Algorithm
Li Chen，Zhang Li
School of Information and Communication Engineering, Beijing University of Posts and
Telecommunications, Beijing（100876)
E-mail:dawn_lee@sina.com zhangli408@gmail.com
Abstract
In downlink CoMP transmission, we consider a UE-centric scheme called SHO (Soft Handover)
grouping algorithm, which is basically different from recent cell-centric scheme. It could afford more
freedom in UEs and cells grouping for collaboration, and provide significant performance gains.
Simulation results are provided to show the considerable gains of the proposed algorithm, which have
been scaled to compare with single-cell mode.
Keywords: Coordinated Multiple Point (CoMP);UE-centric;Soft Handover;grouping algorithm;
MIMO;Linear Minimum Mean Squared Error (LMMSE).
1 Introduction
It is well known that Multiple Input Multiple Output (MIMO) transmission can significantly
enlarge channel capacity in an end-to-end scenario. However, in multi-cell scenario of a cellular
system, the achieved gain is limited due to inter-cell interference (ICI). Coordinated Multiple
Point transmission (CoMP), also known as Collaborative MIMO (C-MIMO) or network MIMO,
can provide significant performance gains especially for cell-edge UEs (User Equipment). It has
been included in protocols for LTE-Advanced of 3GPP [2-3]. Collaborative area (CA), a basic
concept of CoMP, is defined as follows:
Definition: Collaborative area – A transmitted and received set of cells and UEs in the same
CoMP group, in which all the cells transmit useful signals for all the UEs. Each CA contains one
UE set and one sector set. Other cells outside CA are considered as interferences for UEs in CA.
Current researches on downlink CoMP are mostly based on cell-centric scheme, which means
CoMP cells are fixed, then drop UEs served by them in CoMP mode. In other words, CAs are
determined by cells. In some researches, they assumed a pre-set network based CA, i.e. UEs
connected to a pre-selected cell are considered to be in a CA form with 2 other pre-selected cells
belonging to the neighboring sites. It’s easily realistic, but hard to provide obvious gains, even for
cell-edge UEs.
In order to achieve improvements for CoMP transmission, we searched into some other
different techniques. A circular array system has been studied in [4], in which more than 2 cells
are arranged on a circle, and using a multi-cell cooperative zero-forcing beamforming (ZFBF)
scheme combined with a simple user selection procedure, it could provide near-optimal
performance for a moderate number of UEs per cell. A highly sectorized system with 12 sectors
per site has been studied in [5], which could prominently increase the throughput gain. In [6], an
efficient cooperative transmission using simple channel quality identifiers has been proposed and
demonstrated potential capacity gains.
In this paper, we present an entirely new technique called SHO grouping algorithm
mentioned in [1], which is based on UE-centric scheme. Firstly, we divide UEs into groups based
on their active sets which store strongest received signals. UEs in one group are neither more nor
less than the UEs in one CA. Secondly, CoMP cells of each CA could be determined by active sets
of UEs in the CA. Though seriously increased difficulty for scheduling, it could achieve
http://www.paper.edu.cn中国科技论文在线

- 2 -
considerable gains due to more freedom in CA grouping, which would provide new thoughts for
CoMP studying. Simulation results show the significant gains.
Transmitted and received model of CoMP, as well as the precoding scheme, is described
below, while simulator is described in [7].
2 Preparing For Work
Interference cancellation is the main goal for CoMP transmission and reception. In our
simulation, we should choose one method suitable for interference cancellation, considering both
performance gain and system spending.
One UE’s received signal can be expressed as
( )y W P PG H s i n= ⋅ ⋅ + + (1)
where W is precoding matrix, P is transmitted power matrix, PG is path gain matrix, H is
channel transfer function matrix, s is vector of transmitted signal from transmitted antennas of
base station, i is vector of interference signal, n stands for noise vector, which is Gaussian white
noise here. indicates dot multiplication, which means two matrixes multiply element by
element.
From (1), we can see that one received signal is mainly determined by transmission power,
path gain and channel transfer function. In each TTI, only channel transfer function changes
because of fast fading. As signals intensity changed, if we want to remove the strongest
interference during the simulation precisely, we need to calculate and sort their order in each TTI.
It may cost much more simulated time.
For tradeoff, we compared performances of precise and “rough” interference cancelled
scheme, where “rough” scheme means calculate and sort interferences’ order only once during
each simulation run, not each TTI. Results appear, the “rough” scheme, or we called “static”
interference cancelled scheme, indeed not so well as the precise one for gains, but in acceptable
range. Moreover, it saves a great deal of simulation time. Therefore, we choose static interference
cancelled scheme as a basic method in our simulation.
Furthermore, using static interference cancelled scheme, we tried on removing some
strongest interferences for each UE, and have found that when two strongest interferences
removed, system performance would achieve great gain, which means in downlink CoMP,
consider performance and spending tradeoff, two strongest interferences removed is just the main
goal we struggle to achieve.
3 Grouping Algorithm
In this chapter, first, let’s discuss CoMP CA with 3 sectors.
In the system model called 3GPP Macro Case 1, one site is divided into 3 sectors, which
means base station put in the middle of cell, transmitting signals to 3 different directions. So, in
our simulation, sector is just UEs’ serving unit. When UEs have been dropped in simulated area,
each of them would receive signals from all sectors. Only one is the serving sector, others are all
interfered sectors. Now, if one CA contains 3 sectors, for UEs of this CA, one should have 1
serving sector, and 2 interfered sectors. On the premise of both 2 strongest interferences in this CA,
through collaboration, make them removed, then ideal performance gain achieved.
Now two schemes appearing for chosen:
Sector-centric scheme: If sectors in CAs determined before dropping UE, which called
sector-centric scheme, it’s easy to be realized in real mobile communication system, but for UE, it

- 3 -
cannot ensure that interfered sectors in CA are exactly the strongest ones. In other words, even
though an excellent collaboration algorithm adapted, we cannot assure it can reach the ideal
performance gain. Actually, it’s nearly impossible to reach the ideal gain with sector-centric
scheme.
UE-centric scheme: If CAs determined by UEs, which called UE-centric scheme, it’s hard to
put into real system, for the whole network scheduling is quite difficult, which greatly enlarge
system complexity. But with UE-centric scheme, we can make sure that most UEs could be
grouped with their 1 or 2 strongest interference(s). It’s extraordinarily useful for us to analysis the
performance of CoMP.
From our point of view, we prefer UE-centric scheme, because it’s an excited way of free
grouping for CAs, and could break through the limited gain.
Our original simulation system is a 2 × 2 MIMO system. When transform to CoMP
transmission, one principle should be followed: in each CA, UEs’ number should be equal to the
number of sector. For example, 3 UEs’ CA could only allocate 3 sectors, 2 UEs’ CA could only
allocate 2 sectors, and single UE could only “paired” with its serving sector. That’s because no
other than this principle followed, it could ensure that taking average of all, the number of UEs
served by each sector in CoMP mode is equal to that in original. Only in this way, its results can
compare with original ones.
Endeavored to remove 2 strongest interferences for each UE, doesn’t mean it has to. If
strictly group each 3 UEs into CAs, there should be a good many UEs cannot be grouped into any
CA, due to many UEs don’t have the same strongest interferences. Then, they cannot achieve any
performance gain. Consider these cases, we propose one UE grouping scheme called SHO (Soft
Handover) grouping algorithm, which idea comes from [1].
As mentioned in chapter II, interference cancelled scheme needn’t work in each TTI; it’s
enough to work in each simulation run. That’s the same with SHO grouping algorithm.
3.1 Primary SHO Grouping Algorithm
Steps of SHO grouping algorithm described as follows.
1) Sorted order: Rearrange all received signals for each UE, sort them from strongest to
weakest. Note that in our simulation, the strongest signal of each UE always comes from the
serving sector of this UE.
2) Active sets: Establish one null set for each UE called active set, which size is 3, to store
sector indecies. When signal sorting completed, put first 3 signals into active set of each UE.
Active set stores not these signals, but the indecies of sectors which transmitted these 3
strongest signals. Then for one UE, e.g. U1, in its active set, there are index of U1’s serving
sector S1, and indecies of U1’s 2 strongest interfered sectors, for example S2 and S3.
3) UE grouping: Starting from one UE, e.g. U1, check U1’s active set, and search rest UEs for
these which have the same elements in their active sets with U1. Where “same” means, the 3
sector indecies in different UEs’ active sets are same, forget the order of each 3 elements.
4) Counted numbers: Put UEs with same active set elements into one “big” group, count and
save the total number of these UEs in each “big” group.
5) Divided “big” groups into “small”: To build UE set of each CA, according to UEs’ number u
in each “big” group, divide these “big”s to “small” ones as follow. One “small” group is just
the UE set of relative CA.
u = 2 or u = 3: No need to divide these “big” groups which contain 2 or 3 UEs.

- 4 -
u divided by 3 remaining 1: Here u = 3k + 2×2, k is a nonnegative integer. Then we get
k+2 “small” groups from the “big” one.
u divided by 3 remaining 2: Here u = 3k + 2×1, k is a positive integer. Then we get k+1
“small” groups from the “big” one.
u divided by 3 remaining 0: Here u = 3k, k is a positive integer except 1. Then we get k
“small” groups from the “big” one.
Single UE: If one UE cannot be grouped into any “big” group, we call it “single UE”,
consider it’s the only element of a single group. So, one single UE and its serving
sector form one CA.
6) Stored sector indices: When grouping UEs, store indices of relative sectors of these UEs.
UE group: For UEs in 3 UE groups, one’s serving sector and 2 strongest interfered
sectors, namely 3 sectors in one UE’s active set form the sector set of this CA. The 3
UEs’ serving sectors might be either the same or not.
2 UE group: two cases as follow.
a) If different serving sectors for the 2 UEs, the 2 serving sectors form the sector set of
this CA.
b) If same serving sector for the 2 UEs, the serving sector must be 1 sector in this
CA’s sector set, another one should be one UE’s strongest interfered sector, which
might be another UE’s strongest or second strongest interfered sector.
In conclusion, when we get a 3 UE group or 2 UE group, we say that these UEs and their
relative sectors form a whole 6×6 or 4×4 CoMP MIMO system in their own CAs, when
transmitted and received antennas are separately equal to 2. A Single UE works with its serving
sector, which form a 2×2 MIMO system.
From primary simulation results, we found that UEs in same group are near to each other in
geographic location, which means SHO grouping algorithm is reasonable and suitable for CoMP
MIMO system.
When UEs grouped, they should work in CoMP mode, and in CoMP-SHO system, strangest
interferences could be cancelled, by which grouped UEs could enhance their performance.
Nevertheless, with the primary SHO grouping, there are more than 40% single UEs of the
whole network, which means these UEs cannot be grouped, and cannot get any gains. That’s
unacceptable for us. So it’s quite necessary to make some reasonable improvement for SHO
grouping algorithm.
3.2 Advanced SHO Grouping Algorithm
Through primary SHO grouping algorithm, we found it assuredly reasonable for CoMP
transmission, while some problems existed. Based on the primary grouping scheme, we make
some improvement, and propose the advanced SHO grouping algorithm. The main improvement,
namely adding re-grouping for single UEs, showed as follows.
1) Single UEs re-grouping: For single UEs, group them once more. As we said, each UE’s
active set contains 3 indecies of “strong” sectors. If in 2 UEs’ active sets, there are 2 same
sector indecies, the 2 UEs could be put into the same 2 UE group as UE set of this CA. Then
store 2 sectors indecies as sector set of this CA. It also contains two cases as follows.
Different serving sectors for the 2 UEs: The 2 serving sectors form the sector set of this
CA.
Same serving sectors for the 2 UEs: The serving sector must be 1 sector in this CA’s
sector set, another one should be one UE’s strongest interfered sector, which might be
another UE’s strongest or second strongest interfered sector.

- 5 -
2) Single UE: If one UE still cannot be grouped into any group, we call it “single UE”, consider
it’s the only element of a single group. So, one single UE and its serving sector form one
CA.
Until now, CA grouping is completed. All UEs and sectors have become part of 3 UE CoMP,
2 UE CoMP or single UE group. Note that one UE could appear in only one CA, but one sector
could appear in different CA.
As described above, advanced SHO grouping algorithm re-grouped single UEs into 2 UE
CoMP, which could greatly reduce single UEs number.
3.3 Fair PRB Allocation
When grouping finished, we are ready to allocate physical resources to CoMP UEs and single
UEs.
In LTE and LTE-A, OFDM resources distribute in both frequency and time domain. Our
changes for CoMP simulation only base on frequency domain, so we only need to discuss
frequency resource allocation.
In OFDM, the basic unit of frequency resource is called physical resource block (PRB). Each
base station contains same number of PRBs. When each sector dropped same number of UEs,
every UE would be allocated same number of PRBs by its serving sector. But in CoMP mode,
things appear differently. Taking a 3 UE CoMP as example, all the 3 sectors need to allocate their
PRBs to the 3 UEs, e.g. 3 UEs: Ua, Ub and Uc in same 3 UE CoMP, there are three possible cases:
1) Ua, Ub and Uc have the same serving sector S1.
2) 2 of the 3 UEs have the same serving sector S1, the rest one has its own serving sector S2.
3) 3 UEs have 3 different serving sectors: S1, S2 and S3.
In first two cases, sector S1 only need to allocate resource blocks once; but if in normal mode,
each UE should be allocated resource blocks once, then sector S1 has to allocate PRBs twice or
three times, which means some PRBs of S1 should be allocated, but in CoMP mode not. Actually,
sector S2 and S3 allocated ones instead. It’s unfair for each sector. That’s why we need fair PRB
allocation.
Now let’s explain the “fair” scheme: When we determine sectors of each 2 UE CoMP, if the
2 UEs have the same serving sector, we need to check another 2 sectors in one UE’s active set. If
one sector has allocated more PRBs, we choose another as the second sector of the 2 UE CoMP.
Nevertheless, this method cannot completely avoid “unfair” allocations. As a result, a little
few single UEs and CoMP UEs cannot be allocated any resources; in other words, they have to be
discarded. Similarly, a few PRBs cannot be used. It’s not good news, which would make an
impact on the system performance.
4 CoMP MIMO System with SHO Grouping Scheme (CoMP-SHO)
Key points of collaborative MIMO:
1) Determined collaborative areas for ideal system performance, including both sectors and
UEs. It has been realized by our SHO grouping algorithm.
2) Pretreatment for transmitted signals of sectors in CA, just as precoding, is needed, which
would facilitate interference cancelled of CoMP UEs.
In this chapter, we’ll construct the whole CoMP MIMO system with SHO grouping
algorithm, and calculate signal-to-interferer-noise-ratio (SINR) and throughput, to see how much
performance gain we can achieve. We call it COMP-SHO scheme.
We build transmitter and receiver model with the method, which assume base stations have

- 6 -
known channel state information of all UEs. Each sector contains 2 transmitted antennas, while
each UE equipped 2 received antennas. For single UEs, each of them works in a 2×2 MIMO
system with the same precoding scheme.
For easier description, we take a 3 UE CoMP as example. In this CA, 3 UEs are separately
UE 1, 2 and 3, with 3 sectors A, B and C.
4.1 Transmitter Model
Signal model for one UE in a 3 UE CoMP (6Tx-2Rx) transmission can be expressed as
( ),u r u TXu u uy W P PG H s i n= ⋅ ⋅ + + (2)
where u = 1, 2 or 3 is UE index, uy is received signal vector, ,r uW is received precoding
matrix, TXuP is transmitted power matrix, uPG is path gain matrix, uH is channel transfer
function matrix, s is vector of transmitted signal from transmitted antennas of base station, i is
vector of interference signal, all of which are on Rx antennas of user u. n is noise vector, which
is Gaussian white noise here. indicates dot multiplication, which means two matrixes multiply
element by element. The matrix operators:
11 12 11 12 11 12
21 22 21 22 21 22
A A B B C C
A B C
u A A B B C C
h h h h h h
H H H H
h h h h h h
⎡ ⎤
⎡ ⎤= =⎢ ⎥ ⎣ ⎦
⎣ ⎦
(3)
11 12 11 12 11 12
21 22 21 22 21 22
A A B B C C
A B C
u A A B B C C
pg pg pg pg pg pg
PG PG PG PG
pg pg pg pg pg pg
⎡ ⎤
⎡ ⎤= =⎢ ⎥ ⎣ ⎦
⎣ ⎦
(4)
11 12 11 12 11 12
21 22 21 22 21 22
A A B B C C
A B C
TXu TX TX TXA A B B C C
p p p p p p
P P P P
p p p p p p
⎡ ⎤
⎡ ⎤= =⎢ ⎥ ⎣ ⎦
⎣ ⎦
(5)
where k
rth is the complex channel transfer function, k
rtpg is the total path gain, k
rtp is the
total transmitted power, all of which are from sector k transmitted antenna t to UE u received
antenna r.
The expression of transmitted signal vector s:
11 12 131
21 22 232
1
11 12 131
2
21 22 232
3
11 12 131
21 22 232
A A AA
T T T
A A AA
T T T
B B BB
T T T
TX B B BB
T T T
C C CC
T T T
C C CC
T T T
w w ws
w w ws
x
w w ws
s W X x
w w ws
x
w w ws
w w ws
⎡ ⎤⎡ ⎤
⎢ ⎥⎢ ⎥
⎢ ⎥⎢ ⎥ ⎡ ⎤
⎢ ⎥⎢ ⎥ ⎢ ⎥= = ⋅ = ⋅⎢ ⎥⎢ ⎥ ⎢ ⎥
⎢ ⎥⎢ ⎥ ⎢ ⎥⎣ ⎦⎢ ⎥⎢ ⎥
⎢ ⎥⎢ ⎥
⎢ ⎥⎢ ⎥⎣ ⎦ ⎣ ⎦
(6)
where k
ts is the precoded transmitted signal from transmitted antenna t of sector k, k
Ttuw is
the precoded weight from sector k transmitted antenna t to UE u, ux is the useful signal of UE u.
Note that dimensions of transmitted precoding matrix TXW is [ ]t uN N× , where tN and uN
are separately the total number of transmitted antennas (here is 6) and the total number of UEs
(here is 3). The column ,r uW of matrix TXW corresponds to a precoding vector for UE u, with
dimension [ ]1tN × .
We apply zero-force (ZF) precoding to calculate these precoding weights. First, we calculate
the equivalent channel ,eq uH of each UE based on SVD decomposition, and select the direction
that corresponds to the largest eigenvalue:

- 7 -
H
u u u uH U V= ⋅∑ ⋅ (7)
( ), 1 :,1H
eq u u uH Vσ= ⋅ (8)
where 1uσ is the first element of the diagonal matrix u∑ , namely the largest eigenvalue;
and ( ):,1H
uV is the first column of the unitary matrix H
uV .
Then the precoding matrix tuW is calculated by means of the Moore-Penrose pseudo-inverse
as:
( )
1H H
tu eq eq eqW H H H
−
= ⋅ ⋅ (9)
where
,1 ,2 ,3
H
eq eq eq eqH H H H⎡ ⎤= ⎣ ⎦ (10)
4.2 Receiver Model
For receiving signals, 2Rx linear minimum mean squared error (LMMSE) filter is applied.
The received filter for UE u:
( )( ) ( )
1
:,
H H
RXu RXu RXu RXu I NW H u H H R
−
+= ⋅ ⋅ + (11)
where ( )RXu TXu u u tuH P PG H W= ⋅ is the equivalent precoded channel of UE u,
( )
{ }
2
0, , ,
ocN
H
I N k k N
k k A B C
R i i Iσ+
= ∉
= ⋅ +∑ is the interference + noise covariance matrix for ocN other
sectors out of CA (not A, B or C) interfering signals, 1
2
n
n
n
i
i
i
⎡ ⎤
= ⎢ ⎥
⎣ ⎦
is interferences on the two
received antennas of UE 1.
The filtered received signal:
( )RXu RXu RXuY W H i n= ⋅ + + (12)
for UE 1, it can be expressed as
( )
( )
( )
( ) [ ]
1 1 2 3 1
1 1
1 2 3 11 21
2 2
1,1
2,2
I N
RX mu mu RX
I N
oc
mu mu
oc
R
Y y i i W
R
i n
y i i w w
i n
+
+
⎡ ⎤
= + + + ⋅⎢ ⎥
⎢ ⎥⎣ ⎦
+⎡ ⎤
= + + + ⋅⎢ ⎥
+⎣ ⎦
(13)
where 1y is the useful signal of UE 1, 2mui and 3mui are interferences separately from UE
2 and 3 in the same CA, 1rw is the precoded received signal of received filter 1RXW on received
antenna r of UE 1, ,oc ri is the interference from sectors out of CA on received antenna r of UE 1,
rn is the noise on received antenna r of UE 1.
Finally, we get the sub-carrier SINR of UE 1:
( ) ( ) ( )
( ) ( )
2
1
1 3 2 22
, 1 ,
2 1
2
1
3 2 22
, 1 ,
2 1
:,
RX
mu u RX oc r r
u r
mu u r oc r r
u r
y
SINR
i W r i n
y
i w i n
= =
= =
=
⎡ ⎤+ ⋅ +
⎢ ⎥⎣ ⎦
=
⎡ ⎤+ ⋅ +
⎢ ⎥⎣ ⎦
∑ ∑
∑ ∑
(14)
In our transmitted and received model, multi-user interference within CA is cancelled, so

- 8 -
interferences mainly come from sectors out of CA.
As to UEs in 2 UE CoMP and single UEs, the same models are applied. From descriptions
above, we know that with CoMP-SHO scheme, CoMP UEs can availably cancel strong
interference, which is just the goal of it.
5 Numerical Results
In chapter 3 and 4, the whole CoMP-SHO scheme has been explained in detail. In this
chapter, we present numerical results for UE grouping distribution, and performance gain of SINR
and throughput.
Our simulator is presented in [7]. Scenario is 3GPP Macro Case 1, with TU 20 channel. No
HARQ or link adaption.
Figure 1: Percentage of UE distribution

- 9 -
Figure 2: CDF of subcarrier SINR
Figure 3: CDF of throughput
Figure 1 shows the percentage of UE distribution in 5 simulation runs, using SHO grouping
algorithm. As statistic, 28.3% UEs work in 3 UE CoMP, when 56.4% in 2 UE CoMP, and 15.3%
single. 5.75% UEs have to be discarded. In other words, nearly 85% UEs randomly dropped in the
whole network can be grouped, and only 5-6% UEs have to be discarded. It’s acceptable to us.
Figure 2 and 3 separately show the CDF of subcarrier SINR and throughput, which have been

- 10 -
scaled to compare with single-cell mode. With CoMP-SHO scheme, we can achieve 1.5-2 dB
SINR gain, 14.0% mean throughput gain when didn’t statistic discarded UEs, and 7.4% when
included them. The results could reach our satisfaction.
6 Conclusion
In this paper, we proposed a UE-specific grouping scheme called SHO grouping algorithm,
to create collaborative areas with sectors and UEs for CoMP system, which could afford good
grouping results of CA. Under the transmitter and receiver model, we achieved satisfying
performance gains. Though enlarge system complexity in realization, the CoMP-SHO scheme has
high potential for CoMP researches in LTE-Advanced.
References
[1] A. Tolli, M. Codreanu, M. Juntti (2008) Cooperative MIMO-OFDM Cellular System with Soft Handover
between Distributed Base Station Antennas. IEEE Transactions on Wireless Communications, 7(4), pp. 1428-1440
[2] 3GPP TR36.814 V0.4.1 (2009) Further Advancements for E-UTRA Physical Layer Aspects. Release 9
[3] 3GPP TR36.913 V8.0.1 (2009) Requirements for Further Advancements for E-UTRA (LTE-Advanced).
Release 8
[4] O. Somekh, O. Simeone, Y. Bar-Ness, et al (2009) Cooperative Multicell Zero-Forcing Beamforming in
Cellular Downlink Channels. IEEE Transactions on Information Theory, 55(7): pp. 3206-3219
[5] H. Huang, M. Trivellato, A. Hottinen, et al (2009) Increasing Downlink Cellular Troughput with Limited
Network MIMO Coordination. IEEE Transactions on Wireless Communications, 8(6): pp. 2983-2989
[6] L. Thiele, M. Schellmann, T. Wirth, et al (2008) Cooperative Multi-User MIMO Based on Limited Feedback
in Downlink OFDM Systems. 42nd Asilomar Conference on Signals, Systems, and Computers. Proceedings. Vol.4,
pp. 2063-2067
[7] N. Wei, A. Pokhariyal, T. B. Sorensen, et al (2008) Performance of Spatial Division Multiplexing MIMO
with Frequency Domain Packet Scheduling: From Theory to Practise. IEEE Journal on Selected Areas in
Communications, 26(6), pp. 890-900

A CoMP Simulation Model Based on UE-Centric Soft Handover Grouping Algorithm

More Related Content

What's hot (18)

Similar to A CoMP Simulation Model Based on UE-Centric Soft Handover Grouping Algorithm (20)

A CoMP Simulation Model Based on UE-Centric Soft Handover Grouping Algorithm