Heterogeneous Networks: Mobility Management Challenges and Solutions

This article provides a summarized review of mobility management challenges in heterogeneous networks and proposes solutions to improve performance. It discusses topics such as handover process, mobility management, and ping-pongs. The article also analyzes the use of distributed mobility management and resource management in 5G networks.

HETEROGENEOUS NETWORKS
By Name
Course
Instructor
Institution
Location
Date
Abstract -This article gives a summarized review
of five papers, that is, Mobility Management
Challenges in 3GPP Heterogeneous Networks, [1],
Resource and Mobility Management in the
Network Layer of 5G Cellular Ultra-Dense
Networks [2], Distributed Mobility Management
for Future 5G Networks [3], Network slicing
based 5G and future mobile network [4],
Distributed Mobility Management: Current
practices and gap analysis [5].
In particular, a review on such topics as
handover process of heterogeneous networks,
mobility management, mobility management
challenges, hangover failures, ping-pongs amongst
others has been given. The mobility performance of
HetNets is evaluated here using 3GPP as well as
eICIC features like ABSFs. A review of MB-ICIC
scheme proposal has also been done, showing how
pico-cells configure coordinated resources. In
addition a simulation assumptions in 3GPP are
carried out to investigate MB-ICIC performance
under various conditions.
Similarly, this paper did a summary on
management of mobility and resource in the network
layer of 5G ultra-dense cellular networks through cell
densification that plays a big role owing to the fact
that it can increase the spatial reuse of resources. Due
to the challenges involved, this paper analyses some
Heterogeneous Networks 1

of the solutions for the most relevant network
problems of cell densification, that is, interference,
mobility management and resource.
Moreover, this paper discusses the use of
distributed mobility management approach for
networks analysing the contemporary solutions. This
is because distributed mobility management is
coming up as a valid framework for designing
network architectures that will take care of large
traffic requirements as well as the emergence of
highly dense wireless networks.
Index terms
ABSFs, API, BSs, Cell densification, CoMP, DMM,
DMM-GW, GSM, HetNets, LMA, MB-ICIC, MG,
MN, MUEs, NC, PBA, PBU, PDDCH, Ping-pongs,
PMIPv6, RATs, RSRP, SDN, SCs, SNR, TTTs, UEs,
UMTS, 3GPP, 3GPP RAN 2 SI, 5G
Introduction
Mobility management refers to a
functionality that enables operations of mobile
devices in Global Systems for Mobile
Communications (GSM) or Universal Mobile
Telecommunication Systems (UMTS). It traces
locations of subscribers to provide such services as
Short Message Services (SMS), calls and other
mobile phone services.
Network densification appearing in
deployments of SCs is becoming a trend in typical
cellular networks. SCs perform well with large traffic
demands. The fifth generation of cellular networks
technology, 5G is expected to offer a system that will
cater for the extravagant and massive user
requirements to justify its expected huge cost. As
seen in the existing cellular technology, satisfying
growing demand of users requires that an
improvement in spectral efficiency should be coupled
with further cell densification, more so in the densely
populated areas.
Mobile data traffic growth is exponential
and in order to meet this growth, more network
infrastructure is being deployed to bring cellular
networks closer to UEs, thus increasing spatial use
and efficiency of spectrum. Thus, HetNets, which
comprise of co-existing macrocells and low power
nodes like femtocells, relay nodes and pico-cells,
Heterogeneous Networks 2

have been deemed as the most promising way of
solving the performance challenges and offer a major
leap in performance. However, operators are facing
new challenges such as back haul provisioning,
mobility management and intercell interference
coordination, as they try to realize the capacity
benefits and potential coverages of HetNets.
Mobility management challenges in 3GPP
Heterogeneous networks
A comprehensive investigation into the
mobility challenges in HetNets has been done in this
article, proposing a mobility enhancement scheme,
known as MB-ICIC to curb ping-pongs as well as
handover failures. In MB-ICIC, picocells perform
configuration of coordinated resources, that
macrocells utilize to reschedule their high-mobility
UEs [9]. A simulation on ping-pong rates and
handover failures for a wide range of channel and
system parameters that are based on assumptions of
simulation in a 3GPP RAN 2 SI. In comparison to
implementation of ABSFs only at macrocells, a
reduction in ping-pong rate and handover failures for
high mobility UEs occurs by approximately 12% and
13%, in that order [5].
Handover process
Handover process in radio communication
networks are usually conducted between different
carriers, radio access technologies (RATs) and cells
[5],[7]. However this article has given emphasis on
intra-RAT intra-carrier handovers. Particularly, much
focus was put on 3GPP LTE hard handovers whereby
UEs make a new connection with the target cell
immediately after disconnecting with the source cell
[7].
The handover process in 3GPP LTE
typically has four phases, that is, measurement,
processing, preparation and execution. The first two
phases, that is, handover measurement and handover
processing are performed by UE in physical layer, 1
and network layer, 3. Handover measurements
typically rely on downlink reference signal received
power (RSRP) estimates, thereafter, processing
follows the process of handover measurement by
filtering the effects of estimation imperfections and
fading in the previous process.
After the processing phase, given that a
certain handover event entry threshold is achieved,
the UE informs the serving cell and through a report
of the measurement feeds back handover
measurements. Thereafter, the preparation phase
begins, whereby the serving cell invokes the
handover process and makes preparation for
handover execution as well as the target cell [6].
Eventually, during the execution phase, with the help
of the UE, the target cells and the serving cells
Heterogeneous Networks 3

perform vital network procedures to transfer its
connection from the serving cells to the target cells.
Figure 1. L1 and L3 filtering procedures, radio link
monitor process, and handover process in 3GPP LT
Handover failures and ping-pongs
For a handover failure to occur, the following three
conditions must be met:
● RLF occurs in the period between receiving
a handover command and satisfying event
A3 condition
● When timer T310 has been triggered and
continues to run even after a handover
command has been sent.
● When the UE wideband SINR Qout,u is
less than Qout after and handover complete
message has been sent.
If the conditions, 2 or 3 happens, it is called
a packet data control channel (PDCCH) failure. A
ping-pong occurrence is determined by the period a
UE is directly connected to a cell after a handover
process, known as time-of-stay [5]. The time-of-stay
is controlled by the UE. When the UE sends a
handover complete message to the cell, the time-of-
Heterogeneous Networks 4

stay starts and is ended when the UE sends the
message to another cell. Thus, a ping-pong is an
unnecessary handover that terminates the time-of-
stay that occurs when the UE gets a time-of-stay that
doesn't meet the threshold and the new target cell is
similar to the source cell when handing over to the
current serving cell.
Figure 2. Simulated handover failure and ping-pong
rates (with four randomly deployed picocells per
sector, 8 dB range expansion bias, and MB-ICIC): a)
Handover failure rates under MB-ICIC; and b) ping-
pong rates under MB-ICIC.
With the MB-ICIC the handover failures is
reduced significantly. The MB-ICIC enables
macrocells to allocate high speed UEs in coordinated
subframes in order to avoid their strong pico-to-
macro interference as well as their macro-to-pico
handovers. In this manner, the number of RLFs are
reduced, reducing the handover failures. However, in
order to achieve such an improvement in
performance, some resources must be released by the
picocells. Thus, to reduce throughput picocells loss,
high-velocity MUEs must be assigned to pico-cells
ABSFs, while low-velocity MUEs must be handled
through optimization of handover by use of large
TTTs to inhibit ping-pongs [8]. Similarly, it is
possible to adjust the duty cycle of ABSFs semi-
dynamically at a pico-cell with regards to the
percentage of high-mobility UEs within a particular
time frame.
Similarly, the proposed MB-ICIC
significantly reduces the rates of ping-pongs. This is
owing to the fact that there is avoidance of handovers
for high-mobility UEs through cooperative radio
resource management, whereas handover for low-
mobility UEs undergo the standard procedure of
Heterogeneous Networks 5

handover process but with longer TTTs leading to a
reduction in ping-pongs [4].
In comparison to implementation of ABSFs
only at macrocells, a reduction in ping-pong rate and
handover failures for high mobility UEs occurs by
approximately 12% and 13%, in that order. By
allowing a fair comparison of performance the gains
of the proposed MB-ICIC are quantified based on the
case of pico-cell eICIC for sset-3, TTT of 160ms and
pico-cell range expansion that has been proven to
offer the best tradeoff performance of handover
failure and ping-pong.
Resource and Mobility management in the
network layer of 5G cellular ultra-dense networks
This article analyses and presents the most
feasible solutions realized during project METIS for
the most relevant challenges in networks for cell
densification such as mobility management,
interference and resource. Currently, in the field of
contemporary cellular networks, there is an ongoing
trend of network densification in small cells SCs
[11]. Nevertheless, the 2G and 3G technologies
already had the SCs.
Resource and interference management in the
network layer
There are various coordination
options/techniques to inhibit the impact of
interference. The coordination of neighbour base
stations (BSs) for these techniques is essential in
preventing the reuse of resources during critical
situations. Thus, these techniques influence the
resource allocation and the same time, take care of
the allocation of the resources to users.
A simulation of a 2-user, 2-BS scenario
shows that at small signal-to-noise ratio (SNR),
sharing similar frequency resources amongst users is
superior with respect to long term throughput over
continuous transmission of data [13]. This is owing to
the fact that the rate achievable is limited majorly by
noise as compared to interference that has a lower
impact.
At higher SNRs, dedicating resources is
much better, due to the fact that in this situation noise
experienced is very low and thus becomes negligible
unlike interference, and the long term throughput for
resource sharing stops depending on the SNR.
There are three techniques/classes of interference
coordination, that is;
1. Standalone techniques whereby BSs remove
interference autonomously
2. Techniques whereby BSs decide
autonomously to coordinate with neighbours
before transmitting in some resources
3. Centralized techniques with central entities
Heterogeneous Networks 6

Clustering for CoMP
Future 5G networks must have CoMP which
is an important technology whereby several BSs
serve a group of users cooperatively. The cooperation
is beneficial especially to cell edge users, due to the
fact that it improves the useful signal power and
decreases interference. In spite of CoMP being a
promising MAC layer interference coordination
scheme, it is vulnerable to backhauling needs and
large computational burden. Thus, this section is
significant as it analyses the techniques of cell
clustering that reduces the cooperating points, hence
enabling manageable CoMP.
To mitigate the limitations associated with
current clustering, two clustering techniques are
proposed. The first one refers to user-eccentric
dynamic clustering that gives each user a cluster of
BSs. to achieve this the power that BSs use in the
transmission of each user is selected with the
objective of maximizing users’ fairness while
minimizing interference. This technique utilizes
shadowing and path loss reports for performing the
optimization, thus becomes dynamic and adaptable to
changes in network. Similarly, it removes cluster
edge users, due to the fact that all them are in the
middle of their own cluster [12].
The second technique enables sets of
overlapped clusters that are fixed. In this approach,
the clustering was conducted in a specific number of
phases, say N, whereby non-overlapped clusters were
developed in each one. To design this non-
overlapped clustering, a clustering toolbox was
designed using graph partitioning. Particularly, a
network graph in which the graph nodes represents
the cells is generated from the toolbox, and an edge
links two nodes if they share some coverage area.
In the case of overlapped clustering, users’ fairness is
maximized by each cluster using joint transmission
and dirty paper coding (DPC) that decreases
interference within individual clusters [11],[13]. The
difference in performance between the dynamic and
overlapped approach is as a result of DPC
interference mitigation in the overlapped clustering.
Heterogeneous Networks 7

Figure 3. Spectral efficiency achieved by the 95% of
users with different clustering techniques.
SCs Discovery
The future 5G networks will have separate
pieces of millimeter and centimetre bands in the
available spectrum. BSs may operate only on those
pieces in order to reduce filter design complexity.
Nevertheless, there is a challenge of detecting SCs in
many frequency bands. Thus this proposal tends to
address this problem by proposing a solution where
the network with information consisting of samples
Heterogeneous Networks 8

of radio fingerprints corresponding to SC locations
assist the users.
These radio fingerprint samples are lists of
cell-IDs. Users conduct measurements of neighbour
cells and do a comparison with radio fingerprint
samples, as part of the normal operation. Any
fingerprint match found is reported to the network
and configuring the users with measurement targeted
to get the SC that corresponds to it. This approach is
beneficial in the sense that it enables the network and
the users to save energy, since unnecessary
measurements are avoided if no SC is in range, and
the SCs are activated by the network only when users
are present.
Figure 3. Simulation results showing trade-off
between UE’s inter-frequency
measurement power saving and fingerprint (FP)
coverage (percentage of
area with a fingerprint match out of total SC
coverage). The reference for
UE power saving is periodic measurements of inter-
frequency SC carrier
according to LTE-A assumptions.
From the experiment conducted using 3-12
samples of fingerprints per SC, the energy
consumption of the user is decreased by 70-80%
without affecting the 95% accuracy of the SC
coverage. For a particular number of samples of
fingerprint per SC, different trade-offs between
power saving and fingerprint coverage were acquired
by decreasing or increasing the interval of RSRP
where fingerprints exhibit a match [15].
In conclusion, the most feasible solutions
realized during project METIS for the most relevant
challenges in networks for cell densification have
Heterogeneous Networks 9

been presented to aid in the design of future 5G
technology.
Resource and interference management
options and techniques of various implementation
challenges such as centralized, decentralized and
standalone have been presented. Also, a proposal has
been made for clustering techniques for enabling
advanced CoMP comms, together with used context
information for designing a mobility management
that is energy-efficient and reliable [14].
From the results, PRA improves the CoMP
combined with an overlapped clustering by 300% and
by 1000% it improves the throughput of users
experiencing the worst channel conditions. The SC
activation and deactivation techniques ensures that
45% of the energy is saved. The efficiency is further
improved with the SC discovery technique whereby
70-80 percent of energy can be saved.
Therefore, a combined application of these
techniques will definitely enhance a technology
design which will exude robustness against
unprecedented dense deployments and cell and user
mobility.
Distributed Mobility Management for Future 5G
Networks
There is need to expand the capacity of
network to provide an increased bandwidth. This has
been addressed by the deployment of extremely
dense radio networks. This development is facilitated
by the new developments in the IEEE 802.11 family
as well as cellular pico cells and femto cells.
Several approaches have been proposed so
far, including clean-slate solutions and extension of
the existing methods. However, only three methods
from the three main families have been described in
this section. They are the most significant types of
approaches since they eliminate a single mobility
anchor.
PMIPv6-Based DMM Solution
PMIPv6 is a centralized mobility protocol
local mobility anchor (LMA), a core entity develops
bidirectional tunnels located in the access networks
with mobility access gateways (MAGs). The MAG
collects upstream data packets of corresponding users
and via the tunnel are sent to the LMA that send them
to the internet in turns. Similarly, the LMA receives
the downstream packets first, then dispatches them
via the tunnel terminating at the MAG to which the
MN is attached currently.
The PMIPv6 protocol coordinates the
network status, through the use of dedicated signaling
messages, known as proxy binding acknowledgment
(PBA) and proxy binding updates (PBU) between the
LMA and the MAG. this lets the LMA know where
each MN is connected in the MAG for properly
routing its traffic. However, in this manner, in order
Heterogeneous Networks 10

to convey all subscribers’ traffic, the LMA must be
provided with high-speed links and redundancy to the
MAGs, since the data path may become suboptimal
[17].
The PMIPv6-based DMM solution proposed
in this article has its MAG replaced with a DMM
gateway (DMM-GW) that is an evolution of MAG
with links to the internet that do not traverse the
LMA, thus, acting as a plain access router to forward
packets to the internet and from the internet. I
addition, a DMM-GW exhibits mobility anchoring
functions, managing to forward without disrupting
the flows of IP started by an MN while attached to it
before going to another DMM-GW thereafter.
With regards to the DMM terminology, this
protocol is categorized under partially distributed
category. The control plane is maintained centralized,
attached to the CMD role, while the data plane is
always distributed within the DMM-GWs.
Figure 5. PMIPv6-based DMM: overview and
operations.
SDN-Based DMM Solution
This is a DMM solution base on network
defined networking, which is a paradigm in
networking that facilitates separation of data and
control forwarding planes. The separation provides
for a faster configuration and provisioning of
networks. SDN enables the network administrators to
program both the network and traffic behaviours in a
centralized manner, without needing independent
accessing and having to consider each of the
hardware devices of the networks [16].
This approach makes it possible to decouple
the system that makes decisions on the destinations
Heterogeneous Networks 11

of the traffic from the underlying system responsible
for forwarding the traffic to the selected final
destination. Thus networking is simplified together
with deployment of new applications and protocols.
Moreover, by providing for programmability on the
devices and traffic, an SDN network is likely to be
efficient and flexible compared to the traditional
approach.
The network controller is the most
significant entity in the SDN environment as it bears
the responsibility of configuring the network nodes
through a common application programming
interface (API), also known as the southbound API.
Openflow is an API which can be used by an external
software application for programming the network
devices’ forwarding plane [19].
The SDN-based DMM solution in this paper
the network controller (NC) which is a core entity use
the OpenFlow 1.3 API to perform the configuration
of forwarding rules on the DMM-GW access routers
that play the role of anchors. After an MN attaches to
an access point, the NC is informed by the DMM-
GW that assigns a network prefix to the MN. upon
detecting the attachment, the NC performs a
configuration of the OpenFlow rules in each of the
DMM-GW that an MN visits.
To achieve mobility, forwarding and
translation rules are combined on the DMM-GWs.
When an anchored flow packet gets to a visited
DMM-GW, firstly, the anchor rewrites address of the
IP destination with the last known IP address of MN
before redirecting traffic in the new location of MN.
Upon the traffic reaching the last visited DMM-GW,
the DMM-GW reverses the translation of the IP
address first, to restore the old address of IP
destination, then forwards traffic to the corresponding
MN. Contrary to the PMIPv6-based DMM solution,
this solution doesn’t involve IP tunnels.
However, like the PMIPv6-based DMM
solution, SDN-based solution is partially distributed.
The control plane is centralized at the network
controller, no central gateways for traffic but the data
plane is distributed.
Heterogeneous Networks 12

Figure 6. SDN-based DMM: overview and
operations.
Routing -based DMM Solution
This solution applies the concept of freeing
the architecture of an anchor, allowing all the
network nodes develop a new routing map when
terminals start moving by IP routing protocols. This
solution is distributed fully, since both the control
and the data planes are not routed to a particular
centralized node, but are rather operated by routers in
a distributed manner [17].
Figure 7. Routing based DMM: overview and
operations
Therefore, distributed mobility management,
(DMM) is a feasible candidate framework in 5G
networks mobility management. It analyzed the
DMM solution space by putting into perspective
Heterogeneous Networks 13

three major solution families for the distribution of
mobility management on a flat architecture for
mobile networks.
It's important to note that the three major
solutions take different approaches. The first solution
is an improvement of a standard mobility protocol for
the Evolved packet System, better known as Proxy
Mobile IPv6 [18]. It involves the modification and
extension of the original protocol for accommodating
a new batch of functions/operations in order to
distribute the protocol. The second solution is alike to
the first one, but with an exception that it follows an
approach that is defined by a software.
The third and the final mechanism uses the
protocol of BGP routing to operate the mobility
functions needed for delivering packets to and fro
mobile users. The above mentioned three solutions
were evaluated using Linux-based prototypes with
data obtained from real field experiments [20]. The
findings affirm the feeling that the first and the
second proposals respond quicker to stimuli in the
network, however they need specialized entities and
dedicated signaling to manage the required
operations.
On the other hand, the third solution needs a
well established routing protocol, inheriting the
factors associated with signaling overhead and high
convergence latency when performed on domains of
large networks.
Conclusion
In the 5G network design, however, some of
the puzzles that need to be solved consist of the span
of interference and resource management done in the
network and MAC layers together with the optimum
context information available for exchange so as to
benefit from it and keep a low signaling overhead
levels.
The most feasible solutions realized during
project METIS for the most relevant challenges in
networks for cell densification have been presented to
aid in the design of future 5G technology. Resource
and interference management options and techniques
of various implementation challenges such as
centralized, decentralized and standalone have been
presented. Also, a proposal has been made for
clustering techniques for enabling advanced CoMP
comms, together with used context information for
designing a mobility management that is energy-
efficient and reliable.
From the results, PRA improves the CoMP
combined with an overlapped clustering by 300% and
by 1000% it improves the throughput of users
experiencing the worst channel conditions. The SC
activation and deactivation techniques ensures that
45% of the energy is saved. The efficiency is further
Heterogeneous Networks 14

improved with the SC discovery technique whereby
70-80 percent of energy can be saved.
Therefore, a combined application of these
techniques will definitely enhance a technology
design which will exude robustness against
unprecedented dense deployments and cell and user
mobility.
Distributed mobility management, (DMM)
is a feasible candidate framework in 5G networks
mobility management. It analyzed the DMM solution
space by putting into perspective three major solution
families for the distribution of mobility management
on a flat architecture for mobile networks.
Three solutions were evaluated using Linux-based
prototypes with data obtained from real field
experiments. The findings affirm the feeling that the
first and the second proposals respond quicker to
stimuli in the network, however they need specialized
entities and dedicated signaling to manage the
required operations.
On the other hand, the third solution needs a
well established routing protocol, inheriting the
factors associated with signaling overhead and high
convergence latency when performed on domains of
large networks.
Heterogeneous Networks 15

References
[1] D. Lopez-Perez, I. Guvenc, and X. Chu,
“Mobility management challenges in 3GPP
heterogeneous networks,” IEEE Communications
Magazine, vol. 50, no. 12, pp. 70–78, 2012.
[2] D. Calabuig, S. Barmpounakis, S. Gimenez, A.
Kousaridas, T. R. Lakshmana, J. Lorca, P. Lunden,
Z. Ren, P. Sroka, E. Ternon, V. Venkatasubramanian,
and M. Maternia, “Resource and Mobility
Management in the Network Layer of 5G Cellular
Ultra-Dense Networks,” IEEE Communications
Magazine, vol. 55, no. 6, pp. 162–169, 2017.
[3] F. Giust, L. Cominardi, and C. Bernardos,
“Distributed mobility management for future 5G
networks: overview and analysis of existing
approaches,” IEEE Communications Magazine, vol.
53, no. 1, pp. 142–149, 2015.
[4] H. Zhang, N. Liu, X. Chu, K. Long, A.-H.
Aghvami, and V. C. M. Leung, “Network Slicing
Based 5G and Future Mobile Networks: Mobility,
Resource Management, and Challenges,” IEEE
Communications Magazine, vol. 55, no. 8, pp. 138–
145, 2017.
[5] B. Pehrson, “MultiG- distributed multimedia
applications in high-speed and mobile
networks,” First IEEE Symposium on Global Data
Networking.
[6] “Heterogeneous Networks(HetNets) -
SlideShare.” [Online]. Available:
https://www.slideshare.net/dev464898/ppt-het-net.
[Accessed: 07-Jun-2019].
[7] “Inter-Cell Interference Coordination for Control
Channels in LTE Heterogeneous
Networks,” IEEE/ACM Transactions on Networking
(TON). [Online]. Available:
https://dl.acm.org/citation.cfm?id=3041686.
[Accessed: 07-Jun-2019].
[8] D. Aziz and R. Sigle, “Improvement of LTE
Handover Performance Through Interference
Coordination,” Proc. IEEE Vehic. Tech. Conf.
(VTC), Barcelona, Spain, Apr. 2009, pp. 1–5.
[9] Z. Becvar, J. Plachy, and P. Mach, “Path selection
using handover in mobile networks with cloud-
enabled small cells,” 2014 IEEE 25th Annual
International Symposium on Personal, Indoor, and
Mobile Radio Communication (PIMRC), 2014.
[10] M. Olsson, S. Sultana, S. Rommer, L. Frid, and
C. Mulligan, “Mobile Broadband and the Core
Network Evolution,” EPC and 4G Packet Networks,
pp. 3–13, 2013.
[11] M. Nieddu, G. Boatto, M. A. Pirisi, and G.
Dessì, “Determination of four thiophenethylamine
designer drugs (2C-T-4, 2C-T-8, 2C-T-13, 2C-T-17)
in human urine by capillary electrophoresis/mass
spectrometry,” Rapid Communications in Mass
Heterogeneous Networks 16

Spectrometry, vol. 24, no. 16, pp. 2357–2362,
2010.Techniques,” Procedia Computer Science, 10-
Aug-2016. [Online]. Available:
https://www.sciencedirect.com/science/article/pii/S1
877050916317896. [Accessed: 07-Jun-2019].
[12] E. Negishi and G. Wang, “1,4-Disubstituted ,-
1,3-Dienes by [2C 2C] Alkynyl–Alkynyl Coupling
via Boron- or Zirconium-Mediated Migratory
Insertion,” 1,3-Dienes, p. 1, 2009.
[13] A. Prasad, P. Lunden, M. Moisio, M. A.
Uusitalo, and Z. Li, “Efficient mobility and traffic
management for delay tolerant cloud data in 5G
networks,” 2015 IEEE 26th Annual International
Symposium on Personal, Indoor, and Mobile Radio
Communications (PIMRC), 2015.
[15] Z.-Q. Luo and S. Zhang, “Dynamic Spectrum
Management: Complexity and Duality,” IEEE
Journal of Selected Topics in Signal Processing, vol.
2, no. 1, pp. 57–73, 2008.
[16] “www.science.gov.” [Online]. Available:
https://www.science.gov/topicpages/a/area network
bus.html. [Accessed: 07-Jun-2019].
[17] P. Sroka and A. Kliks, “Distributed interference
mitigation in two-tier wireless networks using
correlated equilibrium and regret-matching
learning,” 2014 European Conference on Networks
and Communications (EuCNC), 2014.
[18] M. Nieddu, G. Boatto, M. A. Pirisi, and G.
Dessì, “Determination of four thiophenethylamine.
[19] F. Giust, C. J. Bernardos, and A. D. L. Oliva,
“Analytic Evaluation and Experimental Validation of
a Network-Based IPv6 Distributed Mobility
Management Solution,” IEEE Transactions on
Mobile Computing, vol. 13, no. 11, pp. 2484–2497,
2014.
[20] J. Cartmell, “Meeting lawful interception
requirements for selected IP traffic offload and local
IP access traffic,” 2013 IEEE International
Conference on Technologies for Homeland Security
(HST), 2013.
Heterogeneous Networks 17