HAL Id: hal-00863169
https://hal.archives-ouvertes.fr/hal-00863169
Submitted on 20 Sep 2013
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of sci-
entic research documents, whether they are pub-
lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diusion de documents
scientiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
Reliability for Emergency Applications in Internet of
Things
Nourhene Maalel, Enrico Natalizio, Abdelmadjid Bouabdallah, Pierre Roux,
Mounir Kellil
To cite this version:
Nourhene Maalel, Enrico Natalizio, Abdelmadjid Bouabdallah, Pierre Roux, Mounir Kellil. Reliabil-
ity for Emergency Applications in Internet of Things. IEEE International Conference on Distributed
Computing in Sensor Systems DCOSS, May 2013, Cambridge, MA„ United States. pp.361-366,
�10.1109/DCOSS.2013.40�. �hal-00863169�
Reliability for emergency applications in
Internet of Things
Nourhene Maalel*
†
, Enrico Natalizio
†,
Abdelmadjid Bouabdallah
†,
Pierre Roux*, Mounir Kellil*
*CEA, LIST, Communicating Systems Laboratory, F-91191 Gif-sur-Yvette, France
†
UTC, Heudiasyc Laboratory, UMR CNRS 6599, Compiègne, France
*name.surname@cea.fr,
†
name.surname@hds.utc.fr
Abstract— This paper addresses the Internet
of Things (IoT) paradigm, which is gaining
substantial ground in modern wireless
telecommunications. The IoT describes a vision
where heterogeneous objects like computers,
sensors, Radio-Frequency IDentification (RFID)
tags or mobile phones are able to communicate
and cooperate efficiently to achieve common goals
thanks to a common IP addressing scheme. This
paper focuses on the reliability of emergency
applications under IoT technology. These
applications’ success is contingent upon the
delivery of high-priority events from many
scattered objects to one or more objects without
packet loss. Thus, the network has to be self-
adaptive and resilient to errors by providing
efficient mechanisms for information distribution
especially in the multi-hop scenario. As future
perspective, we propose a lightweight and energy-
efficient joint mechanism, called AJIA (Adaptive
Joint protocol based on Implicit ACK), for packet
loss recovery and route quality evaluation in the
IoT. In this protocol, we use the overhearing
feature, characterizing the wireless channels, as
an implicit ACK mechanism. In addition, the
protocol allows for an adaptive selection of the
routing path based on the link quality.
Keywords—IoT; routing protocol; reliability
I.
I
NTRODUCTION
Thanks to technology innovation, wireless
broadband connectivity is turning out to be affordable
and ubiquitous. This advance brings about the
proliferation of connected devices through internet
which interact and collaborate to accomplish a
common task. This new paradigm is called Internet of
Things (IoT) [1]. The major strength of the IoT idea
is incontestably the high impact it will have on our
daily life. Its application covers a variety of domains
ranging from transport, agriculture, tracking, and
defense to smart environments like homes and
buildings [2].
However, “Things” may face a number of
challenges that can hamper their widespread
exploitation [3]. A network of things has to be self-
adaptive and resilient to communication errors by
providing efficient mechanisms for information
distribution especially in the multi-hop scenario.
These requirements have to be satisfied in an
architecture that can be constrained by limited
processing capabilities, scarce energy resources and
unreliable communication channels [4]. In particular,
in a typical harsh environment, the radio signal is
often affected by interference; medium access
conflicts, multipath fading, shadowing etc. These
problems may result in significant packet losses.
Moreover, the success of any application (particularly
mission-critical ones like life-care data and alarms)
requires the delivery of high-priority events to sinks
without any loss on the path from the original sources
to the final destination [5]. These constraints
emphasize the need for an energy efficient, scalable
and reliable data transport system.
Data retransmission has been considered as one of
the most common schemes [6-7] for improving
transmission reliability. The usage of ACK/NACK
messages is the basic method used to assess the
necessity of retransmission. Nevertheless, such a
method generates an extra traffic, which causes an
additional overhead that is not suitable in a highly
constrained and error prone environment.
Accordingly, an alternative solution should be found
to deal with retransmissions without wasting
bandwidth.
In this paper, we define a joint reliable and
energy-efficient mechanism, called AJIA (Adaptive
Joint protocol based on Implicit ACK), for packet
loss recovery and route quality evaluation. In this
protocol, we transform the overhearing feature, which
is usually considered as a drawback [8], because it
provokes battery depletion, into an advantage for
implicit ACK. We elaborate a lightweight protocol
for data loss recovery, which allows high resource
constrained nodes to achieve reliable data
transmission. Our energy efficient approach uses the
most consistent link and exploits the resource
diversity of the IoT.
The remainder of this paper is organized as
follows: the next section gives an overview of the
concept of IoT. The next one summarizes the
background and the challenges of the IoT paradigm,
section 4 details our proposed protocol AJIA, and
finally section 5 concludes this work.
II. I
NTERNET OF
T
HINGS
A. Concept
The IoT designates a novel paradigm that
connects the pervasive heterogeneous variety of
things around us to the Internet and among
themselves. The IoT is foreseen to be ‘a self-
configured dynamic global network infrastructure
with standards and interoperable communication
protocols where physical and virtual things have
identities, physical attributes,
and virtual
personalities, and are seamlessly integrated into the
information infrastructure’ [9].
Since this concept has emerged, we have seen the
deployment of a new generation of networked smart
objects with powerful abilities (typically
communication, sensing and action). These
capabilities enable to cover various applications
ranging from healthcare to smart environments, from
monitoring to transportation [10-12], etc. All these
applications’ success relies on the data collected from
distributed smart ‘network enabled’ objects and the
reliability of the used infrastructure for data
transmission.
B. IoT in emergency applications
As mentioned above, emergency applications
require an immediate response to any alarm which
involves a continuous supervision of the alarm state.
Communicating objects in the IoT provide complete
visibility of the resources to the administrator of the
system. In building monitoring, for example, such
visibility enables instant reaction to any event by
transferring real-time information about the
occurrence and extension of an accident (such as a
fire) outside the building [2]. The reliable
transmission of parameters, such as temperature and
smoke ingress, can greatly increase awareness and
reactivity of the first responders [2].
In summary, the inclusion of the IoT concept into
emergency application will open new perspectives by
identifying the ‘things’, achieving sensing tasks and
building low cost and reliable solutions and services.
III. B
ACKGROUND
&
CHALLENGES
Currently, many challenging topics have to be
addressed in the IoT. We can summarize these
challenges by:
• Allowing a full interoperability of the
interconnected devices despite their
heterogeneity.
• Enabling an autonomous behavior and a
self adaptation to the environment and network
unreliability by providing a high degree of
smartness.
• Guaranteeing privacy and security issues for
all applications.
• Elaborating energy efficient and scalable
solutions.
Several standards are currently involved in the
development of solutions for IoTs fulfilling the
highlighted technological requirements and acting as
a bridge between the physical world and the Internet
for the IoT. The most used are ZigBee [14] and
6LowPAN [13]. Both of them are implemented on
top of the IEEE 802.15.4 standard [15] which is a
protocol designed for low data rate, and low power
consumption network. In this paper, we give special
consideration to reliability issues and more
particularly to the reliable data transmission
paradigm, which can be extremely critical in some
emergency application. Indeed, a rapid response is
required in critical situations (fire detection, terrorist
attack, etc) to avoid serious damages or even loss of
human lives. That is the reason why prompt
awareness and reliable decision-making support are
important factors for minimizing such dangers. More
concretely, the varying environment and
requirements during an emergency require the ability
to dynamically react in a rapid and correct way. The
wrong transmission of such relevant data, due to link
failure or congestion, by nodes can provoke harmful
consequences. Therefore, adequate mechanisms need
to be developed and implemented in order to
reconstruct new paths when established ones break.
IV. P
ROPOSED PROTOCOL
C. Overview of the mechanism
Throughout this paper, we focus on elaborating
an efficient packet loss control mechanism with
implicit acknowledgments. Our protocol tackles the
link failure and packet loss problem, by proposing a
reliable and energy-efficient error control protocol in
a limited computational resources environment. Our
idea stems from the overhearing characteristic of
wireless communication as shown in Figure 1. When
a node transmits a packet, nodes in its neighborhood
overhear the packet transmission even if there are not
the intended recipients, due to the broadcast nature of
the wireless channel.
Our solution uses this overhearing characteristic
instead of the acknowledgment messages to
guarantee reliability in the network. Moreover, when
a packet loss is detected, retransmission is carried out
on the most reliable link between the node that sent
the (lost) packet and its one-hop neighbors. The
reliability of links is defined through a metric based
on the link history and the link quality indicator
(LQI) which will be detailed in the next subsection.
Besides, the device resource heterogeneity in IoT is
exploited to load balance the traffic and share the
current workload. Indeed, nodes with available
resources are the most involved in retransmission
issues.
To achieve these goals we have taken a different
approach in comparison to traditional end-to-end
error recovery mechanisms, where only the final
destination node is responsible for detecting loss and
requesting retransmission. We propose a hop-by-hop
packet loss recovery mechanism, in which
intermediate nodes also take responsibility for loss
detection and recovery. This approach essentially
segments multi-hop forwarding operations into a
series of single hop transmissions that eliminate error
accumulation. Intuitively, the hop-by-hop approach is
more scalable and capable to recover from loss.
Figure 1 Implicit ACK for the transmission from node A to B
D. Network model and assumptions
We consider a dense set of randomly distributed
objects. Our objective is to provide a reliable
retransmission scheme that takes into consideration
the link consistency without inducing the extra
overhead caused by acknowledgement messages.
Before discussing the details of our protocol, we need
to clarify our assumptions:
-We assume that the network is composed of
heterogeneous objects such as sensors, actuators,
mobile phones, etc, which have different transceiver
characteristics and are randomly distributed in a
limited environment.
- All the devices have sufficient resources to
perform sensing, computing, and communication
activities.
- We consider a many-to-one traffic pattern where
source nodes send data to the sink.
- Data packets are randomly generated and
transmitted to the sink node.
- We assume that nodes are stationary during their
lifetime and able to record the performance of the
link between themselves and their neighboring nodes,
in terms of the ratio packet lost/packets sent.
- We adopt a basic routing scheme, where a
packet gets one-hop closer to the sink after each
transmission. This is made possible by assigning to
each node a level corresponding to its hop-distance
from the sink. Data is transmitted level by level
toward the sink. When different nodes are at the
same distance from the sink, the next hop is chosen
according to its resources: in the routing decision,
nodes will privilege neighbors whose energy is not
limited by their battery (electrical plugged devices),
in order to limit the energy expenditure of
electrically unplugged devices.
- We assume that the collection of data from a
node to the sink must be completed within a specified
time. If the packet does not reach the sink within this
time limit, it is dropped and considered as it has been
lost.
- We assume to have a symmetrical channel, so
that both the endpoints of the channel will keep an
identical history of link performance. Link state is
set to up or down after each packet transmission
depending on the reception/loss of the packet at the
receiving endpoint of the link. We consider that a
link state is up if a packet is correctly transmitted
through it, and down otherwise. The channel state
history is made up by the set of evaluations
performed on the state of a channel. Since the
channel state is binary, a simple count of the number
of states (up/down) suffices to fully describe its
history.
E. Algorithm steps
Our work proposes a reliable data transfer by
using link reliability based on implicit
acknowledgments. The motivation behind our
protocol is to ensure low latency, while minimizing
the loss recovery cost by using localized data
recovery among one hop neighbors.
Our algorithm comprises an initialization phase
followed by three steps: message relaying, lost
message detection and selective recovery:
1) Initialization phase
The initialization phase starts with the
construction of a spanning tree for ordinary routing
operations. To perform this task, we assign levels to
nodes. A level corresponds to the hop-distance of the
node from the sink and data will be transmitted level
by level toward it.
Besides, we maintain a consistent history of the
state of the different links. The channel history will
be used to elaborate our AJIA metric used in the
recovery mechanism, which will be explained in the
next subsection. By using the mentioned
assumptions, we can characterize the link reliability
by the probability of link failure based on the
history. Let A and B be two communicating nodes,
n
up
be the number of successful transmissions
through the link AB and n
down
be the number of
failed transmissions. We denote the link failure
probability, P
hist
, as it follows:
(1)
2) Message relaying
Packets are forwarded hop by hop to get closer to
the sink. Intermediate nodes keep the packet
Id
in
their buffer during T
col
corresponding to the
maximum delay allowed to transmit a packet from a
source to the sink. Neighbors that receive this packet
check their local data cache to discard any replica. If
the packet is received for the second time, it is
automatically dropped to avoid message duplication.
3) Lost message detection
The lost message detection is achieved thanks to the
overhearing mechanism. Let us assume that node A
sends a packet packet
Id1
to node B. After the
transmission, node A awaits T
wait
instants of time
overhearing node B’s transmission, in order to check
whether the packet has been well transmitted or not.
By localizing loss events, this mechanism segments
the multi-hop forwarding operations into a series of
single hop transmission processes, which is effective
in highly error-prone environments. Upon sensing
the channel, if node A does not hear node B
transmitting the packet packet
Id1
to its next hop, it
becomes aware that the packet has been lost and that
a retransmission is required.
4) Selective recovery
Once a packet loss is detected, the AJIA
mechanism relies on its routing metric to choose the
best next hop for the packet retransmission. AJIA
metric evaluates and assigns costs to network links,
and then it determines the most suitable node to act
as a relay. The metric component of AJIA evaluates
links according to 3 parameters:
-The history of the link presented in the previous
subsection.
-The Link Quality Indicator (LQI) is a metric of
the current received signal quality. By using the
802.15.4 standard [15], we have access to the LQI of
the channel between neighboring nodes in the
network. This measurement is included into the
MAC header of each received packet. The use of the
LQI score function ensures a certain level of
adaptability to the environmental conditions. In fact,
it expresses the real quality of the link. The
implementation of our adaptive routing is
accomplished by considering the LQI, which is
considered as a so-called “stigmergic” variable (i.e. a
variable that contains the information used by nodes
to communicate indirectly). The LQI variables are
defined and used by nodes to adaptively change the
way they build routing path. Any change in this
value may induce a change in the preferred direction
towards the sink.
-The resources of the node: the heterogeneous
nature of the IoT infers various level of resources
consumption for different devices. While sensor
nodes are constrained by their batteries, other
devices like computers or phones are not limited by
resources scarcity. This is the reason why we pay
special attention to this parameter when we choose
the best next hop and we exploit the network
diversity to preserve objects energy.
Besides, LQI experiences frequent fluctuations in
highly interfered environment. Hence, we consider
statistics (average number of lost packet per link) as
a basis to assess the reliability of links. For this
reason, we have decided to weight the AJIA metric
by the link failure probability given by our
