Connectivity Maintenance in IoT-based Mobile Networks: Approaches and Challenges

Full text

Connectivity Maintenance in IoT-based Mobile
Networks: Approaches and Challenges
Vahid Khalilpour Akram
International Computer Institute
Ege University
Izmir, Bornova
vahid.akram@ege.edu.tr
Moharram Challenger
University of Antwerp
and Flanders Make
Flanders, Belgium
moharram.challenger@uantwerpen.be
Abstract—Connectivity is an important requirement in almost
all IoT-based wireless networks. The multi-hop networks use
intermediate nodes to create a communication path between other
nodes. Hence losing some nodes may cut off all communication
paths between other active nodes. Generally, the connectivity
of a partitioned network can be restored by adding new or
activating redundant nodes, moving available nodes to the new
location, and increasing the wireless communication range of
nodes. The restoration problem may have many constraints
and sub-problems. The network may initially be disconnected,
the nodes may be heterogeneous, reliable connections may be
required between the nodes, we may have unreachable locations
in the network area to put the new nodes or move exciting
nodes, more than one node may fail at the same time and
the expected coverage area may complicate the connectivity
restoration problem. In this paper, we study the main challenges
and methods of connectivity restoration in IoT-based wireless
networks.
Index Terms—Internet of Things, Connectivity, Multi-hop
Wireless Network, Mobile Networks.
I. INT ROD UC TI ON
INTERNET of Things (IoT) is one of the fastest-growing
and promising technologies that already formed a revolu-
tion in daily human life. In recent years, the new generation
of smart buildings, structures, vehicles, clothes and almost all
types of objects that every day are used by people benefit
from IoT technologies [1], [2]. Technically, IoT is a set of
small, low-energy electronic devices that can connect to the
Internet over wired or wireless communication platforms [3],
[4]. These devices may have different types of capabilities
such as processing, sensing, and data storage. Recent advances
in electronic and hardware technologies allow the generation
of a wide range of tiny, low-cost, low-energy devices that
support local processing, sensing, and various communication
methods. The diversity and capabilities of IoT devices grow
exponentially day by day which allows people to use them in
different application areas. Tracking the status and location of
patients and health care devices in hospitals [5], automation of
activities and increasing the quality and efficiency of products
in agriculture [6], tracking a mobile object in indoor or outdoor
environments [7], controlling the objects in smart homes [8],
automation of fabrication in factories [9], fast and efficient
rescue systems [10], real-time monitoring systems of critical
infrastructures [11], and providing ad-hoc or mobile commu-
nication platforms [12] are a few samples of IoT applications.
Connectivity is a critical necessity in all sorts of networks,
including wired local area networks, wireless ad-hoc networks,
mobile networks, and the Internet of Things. Ideally, all
available devices in a network should be able to commu-
nicate with other devices in the network. In other words,
the network must keep the connectivity between all available
devices. In some types of networks, such as wired local area
networks, preserving the connectivity between the nodes is
almost straightforward. As long as the routers, switches, and
cables work properly, any connected device may communicate
with other devices under predefined security policies. In these
networks, the status of endpoint devices has no effect on the
connectivity of the network. For example, if a device stop
working, the connectivity of other nodes will not be affected.
However, preserving the connectivity in ad-hoc wireless net-
works may be much more complicated. In a wireless ad-hoc
network, the nodes communicate with other remote nodes
over multi-hop links. Using the ad-hoc routing protocols, each
node forwards the received message to its neighbors which
allows the nodes to remote nodes which are outside of their
communication range. Therefore, the connectivity of nodes
relies on the proper working of available intermediate nodes
in the network. Consequently, if a node stop working, we
may lose the connectivity between other working nodes. The
problem will be much more complicated if the nodes are
mobile. If a node changes its initial location, the connectivity
between some other nodes may be completely destroyed. In
a vehicle or drone network, if a mobile node changes its
location, the communication paths between its neighbors will
be changed. In the worst case, if there is no other redundant
path, moving or losing a node may cut the communication
paths to a large set of working nodes and waste many active
resources.
The diversity of device and communication technologies
allows establishing ad-hoc networks almost everywhere even
in harsh environments such as mountains, sea-bed, and forests.
In these networks, the nodes may use hybrid communication
technologies such as Bluetooth, WiFi, GSM, LTE, LoRa, and
Zigbee. Also, some nodes may be static with a fixed location
and some other nodes may be mobile. For example, for real-
Position and Communication Papers of the 16th Conference on
Computer Science and Intelligence Systems pp. 145±149
DOI: 10.15439/2021F102
ISSN 2300-5963 ACSIS, Vol. 26
©2021, PTI 145

time monitoring of an environment, we may distribute some
sensor nodes in the environment and collect their sensing data
over multi-hop links, mobile drones, or mobile vehicles (Fig.
1).
Fig. 1: Sample network for collecting sensed data from envi-
ronment
A wide range of hardware and sensors are available for es-
tablishing a network similar to Fig. 1. For example, an ESP32
device support both WiFi, low energy Bluetooth communi-
cation technologies and have enough memory and processing
power for most of monitoring applications. This device may be
equipped with different types of sensors to gather various data
from the environment. The new generation of drones [13] have
more than one hour fly time and wide communication range
which allows them to reach far locations miles away from
the base station. However, preserving continues and reliable
connectivity in wireless ad-hoc networks still is a challenging
problem. In this paper we, focus on the applications and
different challenges of connectivity maintenance in IoT based
mobile ad-hoc networks. The remaining parts of this paper
has been organized as follow; Section II provides a formal
definition for connectivity problem and its different variants.
Section III focuses on the open challenges and research
problems on the efficient connectivity maintenance in mobile
networks. Finally, Section IV provides the conclusion and
future works.
II. PRO BL EM FO RM UL ATIO N
We can model an ad-hoc network as graph G(V , E)where
Vis the set of nodes and Eis the set of edges between the
nodes. For example, Fig. 2a shows a sample mobile ad-hoc
network with 4 mobile nodes and 15 static nodes. Fig. 2b
shows the graph model of this network where V={0,1, ..18}
and E={(0,7),(1,3),(1,7), ...}is the set of links between
the nodes. In Fig 2b triangles show the mobile nodes and
circles show the static nodes. We assume that node 0, (the filled
black node) is the base station of the network. The dashed big
circles in Fig. 2b shows the communication range of the node
which may differ based on the node types.
Generally, a network is called connected if there is at least a
communication path between every pair of nodes. Connectivity
is one of the most important requirements in all networks.
In wireless ad-hoc networks, where the network connectivity
relies on the proper working of nodes, different strategies have
been developed to increase connectivity robustness. Placing
redundant nodes, creating alternate paths between the nodes,
and increasing the radio range of nodes are some of these
strategies which have their own advantages and disadvantages.
Placing redundant nodes in the environment is a simple and
feasible approach but increases the network cost. Increasing
the radio power of node allows them to connect more nodes
but at the same time increase the energy consumption of
nodes which are not desirable in the battery-powered networks.
Creating and maintaining alternate paths between the nodes
needs complex algorithms and real-time topology control
which may be hard to implement.
Formally, a network is called k-connected if there is at
least kpath between every pair of nodes. Therefore in a 1-
connected, there is at least one path and in a 3-connected
network, there are at least 3 disjoint paths between every
pair of nodes. Higher kvalues increase the reliability of the
network but need precise nodes deployment and restoration
strategies. Generally, challenges and problems on network
connectivity can be classified into 2 groups as connectivity
detection and connectivity restoration problems which are
discussed in more detail in the following subsections.
(a)
(b)
Fig. 2: a) Sample mobile network, b) Graph model of the
network.
146 POSITION AND COMMUNICATION PAPERS OF THE FEDCSIS. ONLINE, 2021

A. Connectivity Detection
Connectivity detection is the problem of finding the connec-
tivity status and reliability of connections between the nodes.
In the simplest case of the connectivity detection problem,
the aim is to determine whether all nodes in the network are
connected. In most applications, we need to ensure that all
nodes have at least one communication path to each other
which leads to the simplest form of connectivity detection
problem. There are many central and distributed algorithms for
the connectivity detection problem [14]. The central connec-
tivity detection algorithms may use different methods such as
depth-first search, network flow, path traversal, and matching
to find the connectivity of the network.
Existing of a communication path between all nodes is a
required condition in most applications, but in most cases
is not enough. In wireless ad-hoc networks, 1-connectivity
usually is considered unreliable because losing some nodes or
links may disconnect a large number of nodes from the others.
For example, Fig. 3a shows a sample 2-connected network
that can tolerate any node or links failure without losing its
connectivity. In contrast, Fig. 3b shows a 1-connected network
with many critical links (orange color) and nodes (filled with
orange) that losing each one destroy the network connectivity.
A node whose failure destroys the network connectivity is
called a critical node. Similarly, a link whose failure destroys
the network connectivity is called a critical link or bridge.
Detecting the critical nodes and bridges may help to improve
connectivity reliability. For example, Fig. 3a and Fig. 3b show
that adding only two links to the graph can resolve all critical
nodes and links.
(a) (b)
Fig. 3: a) a sample 2-connected network, b) a sample 1-
connected network with critical nodes and critical bridges.
Besides the bridges and critical nodes, we may find the
minimum cut edges and minimum cut vertex of a network to
measure its connectivity reliability. The minimum edge cut
of a network is the smallest set of edges whose removal,
destroys the connectivity of the network. For example, in Fig.
3a a minimum edge cut of the network is {(1,5)(2,7)}which
their removal disconnects node {2,5}from the other nodes.
Similarly, a minimum vertex cut of presented network in Fig.
3a is {1,9}. A network may have more than one minimum
edge or minimum vertex cut. Finding the minimum vertex and
edge cuts reveals the weak points and connectivity robustness
of the network.
B. Connectivity Restoration
Network connectivity restoration is the process of increasing
the reliability of network connectivity by reconnecting the
disconnected nodes [15]. In some applications, the connec-
tivity restoration is started after failure in some nodes that
disconnect some working nodes from the others. However,
some applications require continuous and reliable connectivity.
In these applications, the connectivity restoration process
must be started before complete disconnection to reinforce
the unreliable connections. So, the connectivity restoration
strategies can be classified into proactive and reactive groups.
The proactive methods start after each node or links failure and
reinforce the connectivity if required. For example, in the k-
connectivity restoration methods [16], if a node failure reduces
the kvalue, the restoration algorithm tries to increase the k
value by moving other nodes or activating redundant nodes.
The reactive methods start after network disconnection and try
to reconnect the disconnected parts.
The connectivity restoration algorithms usually rely on the
connectivity detection algorithms to determine the current
connectivity status and decide about the required actions.
Generally, the main approaches for connectivity restoration
are moving the available mobile nodes to the new locations,
activating or placing new nodes in the network environ-
ment, and increasing the radio communication of the nodes.
Each approach has its own advantages and disadvantages.
The movement-based methods use available resources in the
network but require mobile nodes which are not feasible in
some applications. Also moving the nodes from their initial
location may disconnect some other links which complicate
the connectivity restoration process.
Placing new nodes or activating redundant nodes simplifies
the connectivity restoration process but requires additional
resources. Also placing new nodes in the desired locations may
not be possible in some harsh environments. Increasing the ra-
dio communication range of reaming nodes is another solution
that may reconnect the disconnected parts. But increasing the
radio communication range increases the energy consumption
of nodes and may reduce the network lifetime. Besides these
issues and constraints, the connectivity restoration problem has
some other difficulties and challenges which are discussed in
the next section.
III. CHA LL EN GE S
In this section we discuss about the main challenges of
connectivty restoration in mobile ad-hoc networks.
A. Initial Connectivity
A network can be initially connected or it can be discon-
nected after deployment. For example, after distributing a large
set of sensor nodes to a forest using an airplane, with a high
probability the resulting network will be disconnected. Some
researchers assume that the network is initially connected and
VAHID KHALILPOUR AKRAM, MOHARRAM CHALLENGER: CONNECTIVITY MAINTENANCE IN IOT-BASED MOBILE NETWORKS 147

the connectivity restoration may start after failure or moving
of nodes. This assumption simplifies the restoration problem
as we ensure that restoring the disconnected links is enough
for establishing the network connection. Connecting all nodes
in a network that is initially disconnected is a hard problem
because the set of possible solutions is very large. In the
movement-based restoration, selecting the candidate nodes for
moving, selecting the direction of movement, and calculating
the movement distance is a hard problem because usually, the
optimal solution needs a combination of different movements.
For example, Fig. 4a shows the movement-based connectivity
restoration in a network that is initially connected and Fig 4b
shows another network that is initially disconnected. Similarly,
connectivity restoration by placing new nodes or activating
redundant nodes is much harder in the networks which are
initially disconnected.
(a) (b)
Fig. 4: a) Connectivity restoration when network is initially
connected, b) Connectivity restoration when network is ini-
tially disconnected.
B. Heterogeneity
An IoT network may include a set of similar nodes with
the same hardware and software properties. In such a ho-
mogeneous network all nodes have almost the same commu-
nication range, processing power, memory capacity, moving
capability, etc. In contrast, in a heterogeneous network, the
nodes may have different hardware and communication ranges.
When the nodes have different communication ranges, some
nodes may connect to a large number of nodes and some
nodes may only have a limited set of neighbors. Also in a
heterogeneous network, we may have uni-directed links which
only allow one-way communications. Connectivity restoration
in heterogeneous networks is much harder than homogeneous
networks because the communication range of each node and
the direction of links should be considered in graph model
[17]. Most of the existing researches in connectivity restoration
assume that all nodes have the same communication range.
C. k-connectivity
The aim of k-connectivity restoration is preserving the k
value of a given network [14]. For example, in a 3-connected
network, we want to preserve the 3-connectivity after losing
some nodes. For k= 1 the problem is converted to the
traditional connectivity restoration but for higher kvalues
the problem will be much more complicated because moving
every node in the network may affect the kvalue. In a 1-
connected network, moving most of the nodes have no effect
on the connectivity. For example in Fig. 4a moving each of
the nodes {1,2,4}does not affect the connectivity. However,
in a k-connected network the set of candidate nodes that can
leave their position without affecting kis limited, and finding
these nodes needs some computation.
D. Target Positions
In the movement or deployment-based methods, we may
assume that any position in the network area can be selected
as a target position for moving the nodes or placing new nodes.
Most of the existing research assumes that all nodes can move
to their desired location or we may put the redundant or new
nodes to the desired location. However, this assumption is not
true for most real-world applications. Due to environmental
conditions and obstacles, the nodes may not move to some
location or we may not put the new nodes in the desired loca-
tions. To simplifies the restoration problem, some researchers
assume that the new nodes can be only added to the location
of exciting nodes or the nodes can only move to the location
of existing nodes. This assumption simplifies the problem and
converts it to a polynomial-time problem.
E. Single vs. Multiple Failure
Restoring the connectivity after a single node failure is
generally simpler than the multiple nodes failure. After the
failure of a single node its neighbor nodes may change their
location to restore the connectivity because all of them may
know the exact location of the failed node. However, in
multiple nodes failure, a node and it’s all neighbors may stop
working at the same time. In this case, some of the failed nodes
may be undetectable, or moving multiple nodes is impossible.
Despite that the multiple node failures can happen in most
real-world application, the researches that consider this case is
limited and the number of proposed solutions is restricted [18].
F. Coverage
In some applications, the IoT nodes collect various data
from enshrinement using different sensors. Losing a node in
an IoT-based network or moving a node to a new location
may lead to some coverage lost in the network. The coverage
lost is not acceptable in some applications hence during the
connectivity restoration we should preserve the maximal cov-
erage. Restoring the connectivity and preserving the maximal
coverage at the same time complicate the restoration process
[19]. Especially in movement-based connectivity restoration,
the nodes which have the minimal effect of total coverage area
should be selected for movement. Generally, the coverage-
aware connectivity restoration methods try to find the nodes
which their covered are is also covered by the other nodes.
IV. CON CL US IO N
Connectivity is one of the most important properties in most
IoT-based wireless networks and robust connectivity is a vital
requirement in most applications. In multi-hop networks, the
148 POSITION AND COMMUNICATION PAPERS OF THE FEDCSIS. ONLINE, 2021

connectivity of the network relies on the proper working of the
nodes, and losing some nodes may destroy the connectivity.
In this paper, we surveyed the main challenges and methods
of connectivity restoration in IoT-based wireless networks.
Generally, the connectivity of a partitioned network can be
restored by adding new or activating redundant nodes, moving
available nodes to new locations, and increasing the wire-
less communication range of nodes. The restoration problem
may have many constraints and sub-problems. Restoring the
connectivity of a network that is initially connected is much
simpler than connectivity all nodes in a network that is initially
disconnected.
In a homogeneous network in which all nodes have the same
hardware and software capabilities, the connectivity restoration
is simpler than a heterogeneous network. In a heterogeneous
network, the communication range and moving capabilities
of each node may be different from the other nodes which
complicate the restoration process. While the 1-connectivity
allows the nodes to communicate with each other, the 1-
connected networks are usually considered unreliable because
losing a single node may destroy the connectivity. The k-
connectivity restoration process tries to preserve kdisjoint
paths between every pair of nodes.
In some applications, the nodes in the network may go
to every desired location or we may add new nodes to
the desired location. However, in some other networks, the
environmental conditions do not allow to put the new nodes
or move the existing nodes to the desired locations. The
connectivity restoration after a single failure can be simpler
than the connectivity restoration after multiple failures because
losing a node and its neighbors may complicate the restoration
process. Finally losing a node in the network may lead to
some coverage loss which may be not acceptable in some
applications. Hence coverage-aware connectivity restoration
algorithm tries to reconnect the connectivity while preserving
the maximal coverage.
As future works, we will focus on the discussed challenges
of the restoration problem to find efficient approaches that
consider more than one criteria at the same time. For example,
proposing a comprehensive approach that can handle multiple
failures, maximize the coverage, preserve the k-connectivity,
support heterogeneous nodes, and allow flexible target position
selection can be very useful in many real-world applications.
Also developing platform-specific languages and frameworks
to support the deployment and connectivity restoration of dif-
ferent mobile and flying nodes under the discussed constraints
can simplify the development and maintaining of complex IoT-
based applications [20], [21].
REF ER EN CE S
[1] S. Arslan, M. Challenger, and O. Dagdeviren, “Wireless Sensor Network
based Fire Detection System for Libraries,” in 2017 International
Conference on Computer Science and Engineering (UBMK). IEEE,
2017, pp. 271–276.
[2] L. ¨
Ozg¨
ur, V. K. Akram, M. Challenger, and O. Da˘
gdeviren, “An IoT
based Smart Thermostat,” in 2018 5th International Conference on
Electrical and Electronic Engineering (ICEEE). IEEE, 2018, pp. 252–
256.
[3] B. Karaduman, T. As¸ıcı, M. Challenger, and R. Eslampanah, “A cloud
and Contiki based Fire Detection System using Multi-hop Wireless Sen-
sor Networks,” in Proceedings of the Fourth International Conference
on Engineering & MIS 2018, 2018, pp. 1–5.
[4] B. Karaduman, M. Challenger, and R. Eslampanah, “ContikiOS based
Library Fire Detection System,” in 2018 5th International Conference
on Electrical and Electronic Engineering (ICEEE). IEEE, 2018, pp.
247–251.
[5] N. Karimpour, B. Karaduman, A. Ural, M. Challenger, and O. Dagde-
viren, “IoT based Hand Hygiene Compliance Monitoring,” in 2019
International Symposium on Networks, Computers and Communications
(ISNCC). IEEE, 2019, pp. 1–6.
[6] M. S. Mekala and P. Viswanathan, “A survey: Smart agriculture iot with
cloud computing,” in 2017 international conference on microelectronic
devices, circuits and systems (ICMDCS). IEEE, 2017, pp. 1–7.
[7] S. Shao, A. Khreishah, and I. Khalil, “Enabling real-time indoor tracking
of iot devices through visible light retroreflection,” IEEE Transactions
on Mobile Computing, vol. 19, no. 4, pp. 836–851, 2019.
[8] Y. Jie, J. Y. Pei, L. Jun, G. Yun, and X. Wei, “Smart home system based
on iot technologies,” in 2013 International conference on computational
and information sciences. IEEE, 2013, pp. 1789–1791.
[9] I. E. Etim and J. Lota, “Power control in cognitive radios, internet-of
things (iot) for factories and industrial automation,” in IECON 2016-
42nd Annual Conference of the IEEE Industrial Electronics Society.
IEEE, 2016, pp. 4701–4705.
[10] T. Ahn, J. Seok, I. Lee, and J. Han, “Reliable flying iot networks for
uav disaster rescue operations,” Mobile Information Systems, vol. 2018,
2018.
[11] S. L. Ullo and G. Sinha, “Advances in smart environment monitoring
systems using iot and sensors,” Sensors, vol. 20, no. 11, p. 3113, 2020.
[12] N. H. Motlagh, M. Bagaa, and T. Taleb, “Uav-based iot platform: A
crowd surveillance use case,” IEEE Communications Magazine, vol. 55,
no. 2, pp. 128–134, 2017.
[13] M. Hassanalian and A. Abdelkefi, “Classifications, applications, and
design challenges of drones: A review,” Progress in Aerospace Sciences,
vol. 91, pp. 99–131, 2017.
[14] V. K. Akram and O. Dagdeviren, “Deck: A distributed, asynchronous and
exact k-connectivity detection algorithm for wireless sensor networks,”
Computer Communications, vol. 116, pp. 9–20, 2018.
[15] Y. Zhang, J. Wang, and G. Hao, “An autonomous connectivity restora-
tion algorithm based on finite state machine for wireless sensor-actor
networks,” Sensors, vol. 18, no. 1, p. 153, 2018.
[16] V. K. Akram and O. DA ˘
GDEV˙
IREN, “Tapu: Test and pick up-based k-
connectivity restoration algorithm for wireless sensor networks,” Turkish
Journal of Electrical Engineering & Computer Sciences, vol. 27, no. 2,
pp. 985–997, 2019.
[17] Y. Zeng, L. Xu, and Z. Chen, “Fault-tolerant algorithms for connectivity
restoration in wireless sensor networks,” Sensors, vol. 16, no. 1, p. 3,
2016.
[18] M. Imran, M. Younis, A. M. Said, and H. Hasbullah, “Localized motion-
based connectivity restoration algorithms for wireless sensor and actor
networks,” Journal of Network and Computer Applications, vol. 35,
no. 2, pp. 844–856, 2012.
[19] N. Tamboli and M. Younis, “Coverage-aware connectivity restoration in
mobile sensor networks,” Journal of network and computer applications,
vol. 33, no. 4, pp. 363–374, 2010.
[20] H. M. Marah, R. Eslampanah, and M. Challenger, “DSML4TinyOS:
Code Generation for Wireless Devices,” in 2nd International Workshop
on Model-Driven Engineering for the Internet-of-Things (MDE4IoT),
21st International Conference on Model Driven Engineering Languages
and Systems (MODELS2018). Copenhagen, Denmark, 2018.
[21] T. Z. Asici, B. Karaduman, R. Eslampanah, M. Challenger, J. Denil, and
H. Vangheluwe, “Applying Model Driven Engineering Techniques to the
Development of Contiki-based IoT Systems,” in 2019 IEEE/ACM 1st
International Workshop on Software Engineering Research & Practices
for the Internet of Things (SERP4IoT). IEEE, 2019, pp. 25–32.
VAHID KHALILPOUR AKRAM, MOHARRAM CHALLENGER: CONNECTIVITY MAINTENANCE IN IOT-BASED MOBILE NETWORKS 149