You are on page 1of 12

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/268522289

Self-organising comprehensive handover strategy for multi-tier LTE-


Advanced heterogeneous networks

Article  in  IET Science, Measurement and Technology · September 2014


DOI: 10.1049/iet-smt.2014.0035

CITATIONS READS

7 169

5 authors, including:

Mohammadreza Behjati John Cosmas


Loughborough University Brunel University London
13 PUBLICATIONS   51 CITATIONS    257 PUBLICATIONS   1,655 CITATIONS   

SEE PROFILE SEE PROFILE

Rajagopal Nilavalan
Brunel University London
129 PUBLICATIONS   1,140 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

IoRL: Internet of Radio Light View project

MIMO antennas design View project

All content following this page was uploaded by Mohammadreza Behjati on 17 July 2015.

The user has requested enhancement of the downloaded file.


www.ietdl.org
Published in IET Science, Measurement and Technology
Received on 31st January 2014
Revised on 25th July 2014
Accepted on 27th July 2014
doi: 10.1049/iet-smt.2014.0035

ISSN 1751-8822

Self-organising comprehensive handover strategy for


multi-tier LTE-Advanced heterogeneous networks
Mohammadreza Behjati1, John Paul Cosmas1, Rajagopal Nilavalan1, Giuseppe Araniti2,
Massimo Condoluci2
1
Wireless Networks and Communications Centre, Brunel University, London UB8 3PH, UK
2
DIIES Department, University Medditeranea of Reggio Calabria, Reggio Calabria 89060, Italy
E-mail: mohammadreza.behjati@brunel.ac.uk

Abstract: Long term evolution (LTE)-advanced was introduced as real fourth generation (4G) with its new features and
additional functions, satisfying the growing demands of quality and network coverage for the network operators’ subscribers.
The term muti-tier has also been recently used with respect to the heterogeneity of the network by applying the various sub-
network cooperative systems and functionalities with self-organising capabilities. Using indoor short-range low-power cellular
base stations, for example, femtocells, in cooperation with existing long-range macrocells are considered as the key technical
challenge of this multi-tier configuration. Furthermore, shortage of network spectrum is a major concern for network
operators which forces them to spend additional attentions to overcome the degradation in performance and quality of
services in 4G HetNets. This study investigates handover between the different layers of a heterogeneous LTE-advanced
system, as a critical attribute to plan the best way of interactive coordination within the network for the proposed HetNet. The
proposed comprehensive handover algorithm takes multiple factors in both handover sensing and decision stages, based on
signal power reception, resource availability and handover optimisation, as well as prioritisation among macro and femto
stations, to obtain maximum signal quality while avoiding unnecessary handovers.

1 Introduction self-organising methods have recently been investigated


as reconfigurable technology to improve the spectrum
As a critical step towards improved network capacity and efficiency for the wireless access technologies, such as LTE
coverage, long term evolution (LTE)-Advanced (LTE-A) and LTE-A. Self-organising strategy extracts planned
was recently standardised in 3GPP new releases and approved coordination and interactions between network parts within
by International Telecommunication Union (ITU) and its diverse stages, leading to a self-aware and demand-based
International Mobile Telecommunications (IMT)-Advanced network with capability of self-correction to increase network
and is now considered as a true fourth generation (4G) system. efficiency, while taking into account the end-to-end goals [3].
Two important requirements of capacity and quality of The self-organisation network (SON) concept in this regard
network services are considered as the main objectives of contains a number of management methods, which are
network design and manufacture in telecommunication planned and executed by network entities through a
standards and networks, which have received the highest pre-defined algorithmic trend. The proposed SON strategy
priority in recent upgrades of LTE-A systems [1, 2]. The peak includes the processes of self-planning, self-configuration,
data rate of 1 Gb/s, realising 100 and 50 Mbps downlink self-optimisation, self-operation and self-correction, with
and uplink rates, respectively; aided by multiple-input responsibility of keeping the network QoS and capacity in the
multiple-output (MIMO) and orthogonal frequency-division satisfactory levels. As the preliminary stages of SON, the link
multiplexing (OFDM) techniques, as well as network configuration, capacity planning and authentication test will
cooperative plans are mentioned as added values of these new be followed by its operational processes such as QoS
releases. Capacity and quality of service (QoS) in the network optimisation, packet drop detection and re-transmission
should also be considered in network design by planning to attempts. Owing to the big demands of the wireless access
reduce the latency of packet transmission from the server and applications, the available unlicensed spectrum is
to the clients. The resultant capacity and QoS in the network reaching its limits and therefore the intelligent use of
are highly affected by growing demand and therefore spectrum is essential to avoid latency and other difficulties in
appropriate solutions are required to measure and manage the broadband, communications, such as congestion. However,
spectral efficiency of the network. On the other hand, new network heterogeneity could alleviate the discussed
releases of 4G systems in radio access network offer, using difficulties in capacity and QoS by the integration of various
self-organising trend, to deal with demand growing challenges sub-networks [4–6], for example, marcocell, picocell,
to reduce the cost and improve the service reliability. The femtocell, relay nodes etc. The multi-tier deployment in

IET Sci. Meas. Technol., pp. 1–11 1


doi: 10.1049/iet-smt.2014.0035 & The Institution of Engineering and Technology 2014
www.ietdl.org

Fig. 1 Multi-tier network deployment in heterogeneous networks

modern networks is considered as an advantageous way to connection of the user equipment (UE) and the source node
reach heterogeneity in networks, as depicted in Fig. 1. is broken before or exactly when the new connection to the
As explained briefly in the abstract, use of multi-tier target node is made. Hard HO is considered as an event
structure in networks also brings some new challenges, during the ongoing communication and requires the least
despite its advantages in improved capacity and QoS. The processing by the network providing system. On the other
time and conditions for handover (HO) [7, 8], the hand, in soft HO (make-before-break) the channel in the
transmission power configuration for various transmitter source node continues to be used in parallel with the new
types, inter-cell interference etc. are examples of those connection to the target node. Therefore the connection to
challenges [9]. This research presents a novel the target node is established for a while before the
comprehensive HO strategy for the proposed multi-tier connection to the source channel is fully broken. The soft
network construction in LTE-A, to facilitate mobile users’ HO is considered as a state during the ongoing
(MUs) movements between general and sub-general communication rather than an event. There can also be
network transmitters. It makes critical enhancements to the more than two parallel connections and the signals used
existing LTE networks, which improves cellular could also be combined to produce a stronger signal for
interference and spectrum usage. transmission, either in downlink, uplink paths or both
HO classifications: depending on what is most advantageous. Soft HOs are
HO can be classified as inter-cell and intra-cell HO. In possible only when the HO cells have a single cell site.
inter-cell HO, the source and target nodes are located on Also, HO could be classified depending on the type of
different cells, even if they are allocated to the same cell target and source nodes, as shown in Fig. 2.
site, whereas in intra-cell HO both the source and target Unnecessary HO in heterogeneous networks is considered
nodes belong to the same cell, and therefore the cell is not as a critical cause for network degradation in the systems with
changed during the HO process. different application layers. The main contribution of this
In other classifications, HO is classified as hard HO and paper is to address the unnecessary HO problem in two-tier
soft HO. In hard HO (break-before-make), the channel in HetNets by proposing of a novel algorithm for new releases
the source node is completely released first, and then the of LTE-A networks. The proposed two-way HO algorithm
channel in the target node is engaged. Therefore the contains received power calculation as its sensing process,

Fig. 2 Different HO classifications in heterogeneous networks

2 IET Sci. Meas. Technol., pp. 1–11


& The Institution of Engineering and Technology 2014 doi: 10.1049/iet-smt.2014.0035
www.ietdl.org
as well as resource availability, HO optimisation and The unnecessary HOs reduction is also investigated based
femtocell priority checks as parts of the HO decision process. on the CAC mechanism between WiMAX and femtocells
[16]. The femtocell capability of providing services in
shadowed areas of WiMAX cell coverage is used to relieve
2 Related work the main traffic from the macro network, as well as
reducing the costs for the network operators, service quality
The significant value of the HO latency is considered by an in indoor environments and capacity increase.
efficient multi-objective HO solution for LTE cellular The existing research in HO strategies for HetNets only
systems [10]. The proposed solution considers different considers a single parameter among resource availability,
parameters, for example, available bandwidth and signal signal strength or HO optimisation processes. Nevertheless,
strength in the selection of the optimal target cell. This the proposed comprehensive algorithm considers both
proposed multi-objective HO scheme results in a the resource availability and user residence duration to
considerable improvement in the session blocking rates, HO optimise the mobility management by avoiding the
latency, session queuing delay and throughput during HO. unnecessary HOs. Furthermore, the novel process of
Femtocells offload a large amount of traffic from the femtocell priority over the macrocell nodes offloads the
macrocellular network in cases of dense deployment of congestion of the users from the macro nodes to the
femto applications, but the HO among macro and femto femtocell nodes, which could improve in both cases of
parts of the network presents a key challenge. network capacity and QoS.
To plan HO management in high-dense networks, research
[11] considers intelligent femto/macro network architecture
as the main contribution. The neighbour cell list with a 3 Comprehensive HO algorithm
minimum number of femtocells and effective call admission
The proposed comprehensive HO strategy is based on a
control (CAC) are considered by proposing a novel
two-way HO for both macro–femto and femto–macro sides
algorithm for the HO. The algorithm aims to create a
and the general processes of the HO sensing and decision
neighbour cell list with a minimum, but appropriate number
are made in the source node, either eNB or HeNB. Since
of cells for HO, as well as novel HO procedure and traffic
there is no direct interface between macro and femto
model for macro/femto networks. The results show that the
stations, the communications between eNB and HeNB are
proposed CAC could be effective in handling of various calls.
implemented through the mobility management entity and
The signal measurement by UEs, whether located in or out
gateways. The table of abbreviations for the HO algorithm
of range, and also the large number of idle femtocells as a
is depicted in Table 1.
result of having a dense deployment of macrocell are
The focus of this work is to propose novel algorithms for
addressed as the two major challenges of HO [12]. The
both the sensing and decision processes within the HO
research proposes an efficient measurement procedure and
model. The proposed algorithms are based on different
appropriate solutions for the two above-mentioned
steps, called checks, which confirm the algorithm to
challenges. Seamless HO is also planned and simulated by
continue with the HO process, or return back to the
reactively multi-casting the data to both the source and
beginning on the algorithm. The proposed comprehensive
target cells after the HO is actually initiated, which results
HO process model is shown in Fig. 3.
in a reduction in the downlink service interruption time, as
well as avoiding the packet loss compared with the standard
3GPP, with only limited extra requirements [13]. 4 Technical considerations
A newly proposed three-dimensional (3D) Markov chain
model for indoor applications and LTE femtocell are also Most of the related works in HO formulation for LTE systems
discussed in [14, 15], respectively, which are required to consider one or two statistics to be checked on HO decision
define priority in the HO algorithm. On the other hand, step, but this work follows a more comprehensive strategy
different mobility patterns and dynamic network conditions to consider more parameters than beforehand. For this
might cause challenging situation for MUs, despite reason, some considerations have to be taken into account.
femtocells’ capability of providing services in shadowed As the first consideration, blocking a HO call in the system
areas by cell coverage enhancements. is not dropping that call. Also the HO from femto to macro

Table 1 Table of abbreviations for the HO algorithm


Abbreviation Full name Abbreviation Full name

MC measurement control RANAP radio access network application part


MR measurement report RANAP RD RANAP relocation detect
HOR handover request RANAP RC RANAP relocation release
HOA handover acknowledgment UE CR UE context release
HCM handover command ULA uplink allocation
PCR physical channel reconfiguration DLA downlink allocation
PCC physical channel complete determine determine the target eNB/HeNB
CINRT CINR target decision the handover decision
CINRS CINR source detach Detach from the source eNB/HeNB
HCN handover confirm release release the sources
ST status transfer U-plane UR user plane update request
PDF packet data forwarding U-plane UA user plane update acknowledge
PSR path switch request RSRP reference signal received power
PSA path switch acknowledgment RSRQ reference signal received quality
PD packet data RSSI received strength signal indicator

IET Sci. Meas. Technol., pp. 1–11 3


doi: 10.1049/iet-smt.2014.0035 & The Institution of Engineering and Technology 2014
www.ietdl.org
station (outbound HO) is not as complex as the HO from probability and queuing delay are anticipated when using
macro to femto (inbound HO), because there is no this HO operation, compared with the existing HO
other option than a HO to a macro station each time an mechanisms. Fig. 4 depicts the communication objects and
outbound HO is being made. In complete spectrum sharing, entities used for the HO process [9].
the total spectrum band is shared by macro and femto The reference signal received power (RSRP) and quality
stations, and in open access mode the femto stations are (RSRQ) values are measured by the UE and reported via
free to be arbitrarily used by any users. This expands the measurement report (MR) to the source eNB/HeNB, and
network capabilities and coverage with an inexpensive the available bandwidth (RBs) is reported by Target eNB/
solution for the network operators. In the proposed HeNB via handover acknowledgment to the source eNB/
algorithm, it is assumed that a UE with a new session HeNB. The proposed comprehensive HO algorithm is
which is waiting in the HO queue does not move from one enhanced by using multiple checks to firstly avoid
eNB to another. unnecessary HOs, and secondly prioritise low-power femto
The considered total channel bandwidth is 20 MHz up to stations over high-power macro base stations, which
100 MHz, which contain 100 and 500 resource blocks significantly increase the QoS over the LTE-A network.
(RBs), respectively. RB is the smallest time–frequency The algorithm mainly focuses on HO sensing and decision
resource, consisting of 12 subcarriers, which can be processes. The proposed multi-objective HO algorithm
allocated to an UE. Significant reduction in blocking initially considers the signal received power and quality as

Fig. 3 Comprehensive HO algorithm process model

4 IET Sci. Meas. Technol., pp. 1–11


& The Institution of Engineering and Technology 2014 doi: 10.1049/iet-smt.2014.0035
www.ietdl.org

Fig. 4 Macro–femto objects and entities to be used in HO process

the HO sensing process, followed by bandwidth availability The receive signal strength indicator (RSSI) and RSRQ are
check, user residence check and femto over macro priority also calculated based on the proposed algorithm definitions
(FoMP) check, and incorporated within the HO decision by the simulation platform. RSSI comprises the linear
process. The novelties of this algorithm consist of both its average of the total received power (in Watts), which is
sensing and decision processes, as well as its unique extra only observed in OFDM symbols containing reference
checks. The internal links and interfaces make the symbols, and is calculated by the simulation source codes.
communications between gateways, mobility management The RSRQ value also is calculated for N number of RBs of
and core network, as shown in Fig. 5. the E-UTRA carrier RSSI measurement bandwidth as (2)

5 System model RSRP


RSRQ = N × (2)
RSSI
5.1 Sensing process

The RSRP is measured and reported by UE and includes


Since the RSSI includes the noise generated in the receiver
antenna gain, pathloss, log-normal shadowing and fast
and also thermal noise within the bandwidth defined by the
fading, which are averaged over all the reference symbols
receiver pulse shaping filter in a specified timeslot, so
within the measurement bandwidth. Considering P as
RSRQ expresses the relation between signal and noise [17].
downlink-received power and Gk, j as estimated channel
The simulator uses both the RSRP and RSRQ values to
gain for the jth symbol of the kth eNodeB, the
calculate a parameter called as ‘Cell Preference Value’. The
downlink-received RSRP from the kth cell is estimated as
calculation process of the cell performance value is then
(1) [10]
followed by a condition test, which if it is between 0 and
 100 (inclusive), allows the candidate to be added to the list
RSRPk = P Gk, j (in which je all symbols) (1) of HOs. The sub-algorithm flowcharts for HO sensing
j process are depicted in Fig. 6.

Fig. 5 Macro–femto internal interfaces to be used in HO process

IET Sci. Meas. Technol., pp. 1–11 5


doi: 10.1049/iet-smt.2014.0035 & The Institution of Engineering and Technology 2014
www.ietdl.org

Fig. 6 HO sensing process sub-algorithm flowcharts and pseudo code

5.2 Decision process data transmission etc. Furthermore, the total number of RBs
(f) is fixed for any node across the entire frequency for a
5.2.1 Resource availability check: The first assessment particular channel bandwidth, and every eNodeB or
parameter of the HO decision process is to check the HeNodeB allocates a portion of these RBs between
available radio resources or wireless bandwidth in the target different users depending on its current channel conditions
node, and if the resource availability is satisfied, the process and cell load [10].
qualifies to start the HO optimisation check (Fig. 7). This
test is computed by the RBs available in the target node,
which results in the target node capable of offering the 5.2.2 HO optimisation check: Although having
maximum available resources being selected. low-power capabilities, femtocell indoor application
The fraction of the total available RBs is mathematically provides low range and limited coverage for the MU, which
calculated by (3). could result in a notable number of unnecessary HOs in
some circumstances. As a very possible scenario, a
h high-speed UE might enter the cell and be covered by
f− bi
gk (h, b) = i=1
(3) multiple femto stations, each for a short period of time,
f which causes multiple successive unnecessary HOs and
therefore the noticeable reduction in QoS. Therefore
where f is the total number of RBs of the target node, βi is the minimising the number of unnecessary HOs is considered
RBs consumed by the ith UE and η represents the number of as a dominant objective for the novel HO strategy. For this
active UEs in thekth target node. The distribution of the RBs, reason, a new CAC mechanism is proposed as part of the
given by b = hi=1 bi , could vary for each different HO decision process. The critical parameters that are
application through the network, for example, voice calls, considered in this check include the expected UE dwell

Fig. 7 Resource availability and HO optimisation checks

6 IET Sci. Meas. Technol., pp. 1–11


& The Institution of Engineering and Technology 2014 doi: 10.1049/iet-smt.2014.0035
www.ietdl.org
Table 2 Pre-defined UE residence states for different speed for handing over to a macro station (hand-out) in this
ranges algorithm (Fig. 9). In this figure, FDCH and MDCH stand
UE residence state UE speed, km/h for femto data channel and macro data channel, respectively.
As mentioned earlier, to obtain the best available cellular
low speed 0–15 quality and coverage with lower service charges, the
medium speed 15–30 femtocell nodes are preferred over the macrocell nodes to
high speed >30 be selected as the target node in HOs. This means the UE is
more likely to be assigned a femto channel, if available,
and then in case of no available femto channel, the macro
time in femto coverage area, by considering the UE speed (in channel will be assigned to the UE. If also no macro
kilometre/hour), in cooperation with the signal quality HO channel is available, the session queuing and session block
checks, to minimise any unnecessary HOs. Hence, the statuses have the next respective priorities.
pre-defined UE state is initially defined in Table 2 [16, 18]. When the system traffic is statistically stable, the traffic
The calculations for the optimised HO check need more intensity of the incoming HO is equal to that for ongoing
complexity to calculate the UE speed and consider the HO calls, that is, (4).
appropriate UE residence state, which the process is
designated after the resource availability check. The
flowchart for the HO optimisation check, which is based on To = au Tn (4)
the defined speed ranges for a MU, is depicted in Fig. 8.
The normalised HO rates in eNodeB are also used to calculate
5.2.3 FoMP check: As a technical comparison between the the normalised traffic intensities of new HO sessions in the
central macro and supportive femto applications, femto equation above. The possible sessions in the HO procedure
services deliver lower power requirements, higher quality of are defined as new session and HO session, and the major
signal and also encouraging cost of services, which parameters for the possible sessions are also defined as in
altogether make the use of femto stations a priority over the Table 3 [10].
macro. Therefore, as the final stage of HO decision process, Fig. 10a shows the HO sessions model for LTE systems. In
macro–femto HO (hand-in) is considered to have priority this regard, the HO session arrival rate seen by the eNodeB is

Fig. 8 HO optimisation check flowchart

Fig. 9 FoMP check sub-algorithm flowcharts

IET Sci. Meas. Technol., pp. 1–11 7


doi: 10.1049/iet-smt.2014.0035 & The Institution of Engineering and Technology 2014
www.ietdl.org
Table 3 Possible HO sessions Table 4 Major parameters and definitions for 3D Markov chain
l
Normalised traffic intensities of new Tn = n Symbol Explanation
sessions m
l
Normalised traffic intensities of HO To = o m number of MDCHs being used
sessions m
mu n number of FDCHs being used
Normalised HO rates in eNodeB au = b number of new session requests waiting in the queue
m
m M total number of MDCHs
Normalised HO rates in HeNodeB av = v N total number of FDCHs
m
ln new session arrival B total number of session requests
rate H number of MDCHs reserved for HO
lo HO session rate
µu HO service rate from
eNB
µω HO service rate from (a) Session completion at HeNodeB: The femto channel is
HeNB released (nj = ni − 1).
µ session service rate (b) Session completion at eNodeB: The macro channel is
released (mj = mi − 1). Now if the macro channel is
as in (5). available (mi − 1 < H), then it is assigned to a new session
waiting in the queue (bj = bi − 1 and mj = mi), otherwise bj
 nav = bi and mj = mi − 1.
(1 − c)lo + , if n , N (c) Incoming session to HeNodeB coverage area: If femto
leNB = 2 (5) channel is available (ni < N), then it is assigned (mj = mi and
lo , if n = N
nj = ni + 1). Otherwise, if macro channel is available then it
is assigned (mj = mi + 1 and nj = ni). Otherwise if the queue
is not full then wait on the queue (mj = mi, nj = ni and bj =
In which c represents the ratio of HO regions in eNB and bi + 1).
HeNB, respectively, n is the number of FDCHs being used (d) Incoming session out of the HeNodeB coverage area: If
and N represents the total number of FDCHs. It is clear that macro channel is available then it is assigned (mj = mi + 1
the above estimation depends on the value of n to calculate and nj = ni). Otherwise, if the queue is not full then wait on
the HO session arrival rate. the queue (mj = mi, nj = ni and bj = bi + 1).
To characterise the behaviour of the macro and femto (e) Outgoing session from HeNodeB: The femto channel is
channel allocation within the HO process, the HO released (nj = ni − 1). If another femto channel is available
transitions and probabilities estimations are introduced then it is assigned (nj = ni). Otherwise, if macro channel is
using the 3D discrete Markov chain [10]. In this regard, the available then (mj = mi + 1).
potential target node is selected from the nearby femto (f) Outgoing session from eNodeB: The macro channel is
nodes, if available [19]. If not, the target node is selected released (mj = mi − 1) and then the femto channel (if
from the available macro stations to enhance the signal available) is assigned nj = ni + 1. If any macro channel now
reception. A state in the 3D discrete Markov chain is available (mi − 1 < H) then it is assigned to a new session
defined as (m, n, b), 0 ≤ m ≤ M, 0 ≤ n ≤ N, 0 ≤ b ≤ B with waiting in the queue (bj = bi − 1 and mj = mi), otherwise bj
the major parameters and symbols which are described as in = bi and mj = mi − 1.
Table 4.
The state transition probabilities for the above scenario are (g) New session: Waits on the queue if the queue is not full
depicted as in Fig. 10b. At the time of HO process, when the (bj = bi + 1).
number of macro channels used is greater than the reserved
macro channels for HO, that is, H ≤ mi ≤ M, the state
transition from (mi, ni, bi) to (mj, nj, bj ) are as the following The mentioned hierarchical states consider the session
states. In this case, the allocation priority is with femto, completion, incoming session and outgoing session statuses
macro and HO queue (if available), respectively, as the with respect to the available macro and femto channels. The
following details [10]: priorities, as already explained, are to assign the femto and

Fig. 10 HO sessions model and state transition probabilities for LTE systems
a HO sessions model for LTE systems
b State transition probabilities for macro and femto

8 IET Sci. Meas. Technol., pp. 1–11


& The Institution of Engineering and Technology 2014 doi: 10.1049/iet-smt.2014.0035
www.ietdl.org
Table 5 Simulation parameters and characteristics 6 Simulation scenarios and results
Entire parameter Simulation value
6.1 Simulation model and considerations
antenna gain 15 dBi
maximum macro transmission power 26.99 dBm
To fully illustrate the impacts of the novel comprehensive HO
maximum femto transmission power 10 dBm strategy within a multi-tier LTE-A network, it is necessary to
terrain model and type urban, terrain type A consider both the effects of the novel algorithm and advanced
propagation model HATA macro–femto combination within a unique analysis. The
building percentage 31.6% simulation parameters and values are revealed in Table 5 [20].
PHY profile 20 MHz FDD
FDD uplink SC-FDMA Therefore three different scenarios have been planned,
FDD downlink OFDMA simulated and analysed by using OPNET modeller network
bandwidth 20 MHz simulator. The first scenario contains an existing
uplink base frequency 1920 MHz macro-only LTE network, in which the network users move
downlink base frequency 2110 MHz
between fixed macro base stations according to random
trajectories. On the second scenario, the cooperation among
macro channels, respectively, if available, or waiting on macro and femto stations is simulated to support LTE users
the queue subject to the queue availability. By using all the by operating the existing HO plans. As a further step
available femtocell coverage within the LTE-A, as the towards the network performance analysis, the third
initial priority, the maximum capability of network scenario comprises the novel comprehensive algorithm for
heterogeneity will be applied to obtain the channel two-tier LTE-A network to show the improvements over the
optimisation goal of when to use the available resources as existing algorithms. The ten LTE mobile nodes have been
part of the channel allocation procedure. allocated to four fixed eNodeBs and six fixed HeNodeBs,

Fig. 11 Comparison between simulation scenarios for


a Downlink SNR
b Downlink BLER
c Downlink delay
d Packet end-to-end delay

IET Sci. Meas. Technol., pp. 1–11 9


doi: 10.1049/iet-smt.2014.0035 & The Institution of Engineering and Technology 2014
www.ietdl.org

Fig. 12 MAC traffic received in downlink for different systems

which are managed by a central evolved packet core entity 6.2 Simulation results and analysis
through IP backbones and gateways.
For the packet transmission values of interference power The relevant function codes for process and node models for
(Pi), noise power (Pb) and the received power (Pr), the eNodeB and HeNodeB are being considered to apply for
value of signal to interference plus noise ratio (SINR) is different strategies of HOs. The access stratum (AS) is
calculated as in [3] by (6). being considered as a functional layer between LTE-A
network and UE, and the performance value and number of
  RBs are calculated through this node model. Furthermore,
Pr
SINR = 10 log10 (6) the non-AS is applied as a functional layer between UE and
Pb + Pi core network, where the UE’s regional tracking and time of
residence is saved. On the other hand, the calculations that
Further to the signalling values, the blocking rate of the apply FoMP check are also performed on the S1 and X2
network is also computed within the value of block error interfaces to select the most suitable interface for
rate (BLER). This depends on the value of the received transmission.
error over the total number of the blocks, and can be Considering the explained scenarios, the transmission values
calculated as in (7). for the three scenarios are depicted and compared in Fig. 11.
It can be observed that the average values of SNR and
BLER are more stable and improved by 40 and 65% for the
Number of erroneous blocks
BLER = (7) LTE-A system enhanced by the comprehensive HO
Total number of received blocks algorithm. In two cases of downlink and packet end-to-end
delays, the proposed HO algorithm affects the packet
The insufficient range of macrocell, when considering the delivery for the destination mobile node in downlink,
growing demands, leads to experiencing drops in QoS for therefore the anticipated delay for the packet delivery is
both the cases of SNR and BLER values. This has been improved.
taken into account while designing the femto network layer The received traffic on media access control (MAC) layer is
to alleviate the network capacity shortage problem in new also affected by the proposed comprehensive strategy. Fig. 12
releases of LTE-A architectures. shows that the value of downlink MAC traffic received for the

Fig. 13 Admitted downlink GBR capacity against


a Simulation time
b Number of femtocells

10 IET Sci. Meas. Technol., pp. 1–11


& The Institution of Engineering and Technology 2014 doi: 10.1049/iet-smt.2014.0035
www.ietdl.org
system is more satisfactory by 1.2% for the comprehensive applications for channel allocation, power control
HO algorithm. and interference coordination through the multi-layer
This higher value for the received traffic in MAC layer is network structure are the other new topics which are under
due to the mitigation of unnecessary HOs by applying the research.
multi-check HO architecture. The LTE-A node only
initiates the HO when the duration of the residence
threshold is met by the MU while moving. 8 References
A transport block is considered as the minimum allocation
size in LTE-A sub-frames, and the bandwidth capacity of 1 Gozalvez, J.: ‘First LTE-advanced commercial network deployed’, IEEE
Veh. Technol. Mag., 2013, 8, (1), pp. 10–17
20 MHz in the proposed LTE-A system has 100 transport 2 Shin, O.S., Elayoubi, S.E., Jeong, Y.K., Shin, Y.: ‘Advanced
blocks per sub-frame. Therefore, a total of 100 k transport technologies for LTE advanced’, EURASIP J. Wirel. Commun. Netw.,
blocks per second could be available, but only part of this 2013, 25, doi: 10.1186/1687-1499-2013-25
number might be used for data communication [20]. 3 Behjati, M., Cosmas, J.: ‘Multi-layer cell deployment strategy for
Fig. 13 shows the admitted downlink guaranteed bit rate self-organizing LTE-advanced networks’. IEEE Int. Wireless
Communications and Mobile Computing Conf. (IWCMC), Cagliari,
(DL GBR) capacity for different number of applied Italy, July 2013
HeNodeBs in transport packets per second, for the time of 4 Ling, J., Chizhik, D., Chen, C.S., Valenzuela, R.A.: ‘Capacity growth of
simulations. The simulator runs the projected scenarios for heterogeneous cellular networks’, Alcatel-Lucent Bell Labs Tech. J.,
a macro-only system (no femtocell applied), 5, 10, 15 and 2013, 18, (1), pp. 27–40
20 femtocells with same characteristics, respectively. The 5 Stanze, O., Weber, A.: ‘Heterogeneous networks with LTE-advanced
technologies’, Alcatel-Lucent, Bell Labs Tech. J., 2013, 18, (1),
results show that the scenarios with a larger number of pp. 41–58
femto stations (HeNodeBs) need a lower number of 6 Soh, Y.S., Quek, T.Q.S., Kountouris, M., Shin, H.: ‘Energy efficient
admitted DL GBR capacities, since the proposed HO heterogeneous cellular networks’, IEEE J. Sel. Areas Commun., 2013,
algorithm switches the serving femtocells accordingly, 31, (5), pp. 840–850
7 Kim, R.Y., Jung, L.: ‘Advanced handover scheme in IMT-advanced
while the mobile nodes are moving through the network. systems’, IEEE Commun. Mag., 2010, 48, (8), pp. 78–85
Furthermore, because of the big deviation of the second 8 Huang, J.W.: ‘Cognitive base stations in LTE/3GPP femtocells: a
scenario from the first scenario with no femtocell node, the correlated equilibrium game-theoretic approach’, IEEE Trans.
value of admitted downlink GBR capacity when using a Commun., 2011, 59, (12), pp. 3485–3493
larger number of femtocells are plotted closer to their prior 9 Behjati, M., Cosmas, J.: ‘Self-organizing network interference
coordination for future LTE-advanced networks’. IEEE Int. Symp.
scenarios in this figure. For this reason, the results also Broadband Multimedia Systems and Broadcasting (BMSB), London,
confirm the improvements in this capacity value, when the UK, June 2013
femtocell nodes are applied, compared with the macro-only 10 Roy, A., Shin, J., Saxena, N.: ‘Multi-objective handover in LTE macro/
scenario. femto-cell networks’, J. Commun. Netw., 2012, 14, (5), pp. 578–587
11 Chowdhury, M.Z., Jang, Y.M.: ‘Handover management in high-dense
femtocellular networks’, EURASIP J. Wirel. Commun. Netw., 2013, 6,
7 Conclusion and future work doi: 10.1186/1687-1499-2013-6
12 Chang, C.W., Lin, Y.H., Jan, R.H., Chen, C.: ‘Efficient measurement
In this paper, we proposed a comprehensive algorithm for HO procedure for handover in LTE femtocell networks’. IEEE Int.
Workshop on Mobile Ad-Hoc Wireless Networks, 2012
among macro and femto applications on LTE-A networks, by 13 Guo, T., Quddus, A.U., Tafazzoli, R.: ‘Seamless handover for LTE
using self-organising instructions. Multiple filters such as macro-femto networks based on reactive data bicasting’, IEEE
received signal power, available resources, user residence Commun. Lett., 2012, 16, (11), pp. 1788–1791
duration and node priorities have been designated within a 14 Bhattacharya, S., Qazi, B.R., Elmirghani, J.H.H.: ‘A 3-D Markov chain
unique widespread algorithm to sense and decide about the model for a multi-dimensional indoor environment’. IEEE Globecom
Proceedings, 2012
best option for a target station, while avoiding the 15 Amirrudin, N.A., Ariffin, S.H.S., Malik, N.N.N.A., Ghazali, N.E.:
unnecessary HOs. The novel algorithm was incorporated ‘Mobility prediction via Markov model in LTE femtocell’,
into a multi-layer macro–femto LTE-A network and the Int. J. Comput. Appl., 2013, 65, (18), pp. 40–44
simulation results were compared for the existing 16 Singoria, R., Oliveira, T., Agrawal, D.P.: ‘Reducing unnecessary
handovers: call admission control mechanism between WiMAX and
macro-only and macro–femto network scenarios. As the femtocells’. IEEE Global Telecommunications Conference
results show the capacity and QoS in a heterogeneous (GLOBECOM), 2011
network with macrocell and femtocell are improved when 17 Xian, H., Muqing, W., Jiansong, M., Cunyi, Z.: ‘The impact of channel
the unnecessary HOs are avoided by the suggested environment on the RSRP and RSRQ measurement of handover
performance’. IEEE Int. Symp. Wireless Communication Systems,
algorithm. The simulation results clearly confirm the 2008, pp. 703–707
enhancement in network performance statistics such as 18 Zhang, H., Wen, X., Wang, B., Zheng, W., Sun, Y.: ‘A novel handover
SNR, BLER, delay and admitted capacity, when the HO mechanism between femtocell and macrocell for LTE based networks’.
among macro and femto nodes are planned with a Second Int. Conf. Communication Software and Networks, 2010
self-organising method. 19 Azni, A.H., Ahmad, R., Noh, Z.A.M.: ‘A 3-D Markov chain correlated
algorithm for MANETS’. Int. Conf. Computer and Information Science
Future research will focus on the other categories of HO in (ICCIS), 2012
HetNet architectures, for example, femto–femto HO in 20 OPNET modeler 17.1 software and documentations, opnetwork labs and
LTE-A networks. Furthermore, the other SON algorithmic tutorials, OPNET-Riverbed

IET Sci. Meas. Technol., pp. 1–11 11


doi: 10.1049/iet-smt.2014.0035 & The Institution of Engineering and Technology 2014

View publication stats

You might also like