Content uploaded by Javier Silvestre-Blanes
Author content
All content in this area was uploaded by Javier Silvestre-Blanes on May 20, 2019
Content may be subject to copyright.
2 Telfor Journal, Vol. 10, No. 1, 2018.
1
Abstract—this paper introduces a novel quality estimator
that uses different metrics to decide the best path towards the
root in Wireless Sensor Networks. The different metrics are
assessed at medium access control layer (MAC), under the
IEEE 802.15.4 standard, and are used at network layer,
enhancing the best path selection process done by the routing
protocol, and at the application layer, enabling visual quality
indicators in the nodes. This quality function is used during
deployment stage; ensuring nodes are located optimally and
nimbly. This mechanism will help WSN’s adoption in
Industrial Internet of Things applications.
Keywords — Deployment, link quality, objective function,
WSN.
I. INTRODUCTION
NE of the enabling technologies of the emerging
Industrial Internet of Things (IIoT) are the Wireless
sensor networks (WSN): systems formed by a multitude of
sensors distributed in a geographical extension (to a
greater or lesser scale), that perform a distributed
collection of information (environmental, physical, gases,
biological...) and communicate it to a central system using
wireless technologies. This creates high granularity
sensory systems with a much lower cost than the wired
solutions. The possibilities offered by gathering large
amounts of data, in real time and at a minimal cost, open
the door to a wide range of applications for the WSN
regarding not only industry, but also environment, public
safety, smart cities, infrastructure, and many more.
Traditionally, for a WSN to operate reliably in harsh
environments, pre-deployment studies/tools and additional
management devices are necessary. Nevertheless, an easy,
rapid and more efficient network deployment avoiding this
1
Paper received March 20, 2018; revised April 23, 2018; accepted
April 25, 2018. Date of publication July 31, 2018. The associate editor
coordinating the review of this manuscript and approving it for
publication was Prof. Grozdan Petrović.
This paper is a revised and expanded version of the paper presented
at the 25th Telecommunications Forum TELFOR 2017 [16].
This work is supported by IVACE (Insituto Valenciano de
Competitividad Empresarial) through FEDER funding (exp.
IMDEEA/2017/103).
D. Todolí-Ferrandis, S. Santonja-Climent and J. Vera-Pérez are with
the Instituto Tecnológico de Informática (ITI), Camino de Vera s/n,
46022 Valencia, Spain (e-mail:[dtodoli,ssantonja,jvera]@iti.es).
J. Silvestre-Blanes and V. Sempere-Payá are with the Universitat
Politècnica de València (UPV), Camino de Vera s/n, 46022 Valencia,
Spain (e-mail: jsilves@disca.upv.es, vsempere@dcom.upv.es).
pre-deployment phase is bound to overcome the arguments
wielded by technology hesitant adopters. A solution
proposed is based in adding mechanisms and tools in each
node that enable triggering a connectivity and quality test,
and showing the results in a user-friendly manner during
the deployment.
The objective of this paper is to define a quality
estimator that allows characterizing the links and paths
between nodes during the first steps of deployment. This
allows the Routing Protocol for Lossy networks (RPL) to
create rapidly a robust topology using the quality estimator
as metric input for the objective function that resolves the
best paths for each node in the network. The quality
assessment is designed to consider and merge different
indicators, such as signal strength (RSSI), expected
transmission count (ETX) and hop count, and mapped
together using fuzzy logic and statistics, making the
impact of each variable tunable to enhance easier
optimization for a wide range of applications. In addition,
the resolution algorithm is embedded in the RPL
mechanism, avoiding any extra control packets overhead
and optimizing the software implementation. This quality
parameter also supports the operator during deployment
tasks, by enabling the node to represent the value of
connection quality and allowing finding optimal locations
to place sensor nodes. Therefore, it will reduce the time
taken to deploy the network, as well as simplifying
operator’s work, meaning staff without technical
knowledge is able to deploy a WSN without simulators,
planning software or any other external complex tools.
The rest of the paper is organized as follows: Section II
shows related work concerning multipurpose link quality
estimation used for the RPL objective functions. In
Section III, we define the chosen parameters for the link
quality estimator, and their mapping for the RPL objective
function. In Section IV, the different simulations and
results are summarized, comparing the response of the
network to different quality calculation configurations.
Finally, we conclude this paper in Section V.
II. RELATED WORK
The Internet Engineering Task Force (IETF)
ROLL(Routing Over Low power and Lossy networks) has
overseen developing and standardizing the RPL protocol
and has so far proven to be a powerful and flexible route
selection mechanism for Low Power and Lossy Networks
(LLNs). Therefore, both the scientific and industrial
community has focused on it and many studies have been
published evaluating their performance [1].
Path Quality Estimator for 802.15.4e TSCH Fast
Deployment Tool
José Vera-Pérez, David Todolí-Ferrandis, Salvador Santonja-Climent,
Javier Silvestre-Blanes, and Víctor Sempere-Paya
O
Vera-Pérez et al.: Path Quality Estimator for 802.15.4e TSCH Fast Deployment Tool 3
The protocol uses different metrics introduced in the
Objective Function (OF) to characterize each of the links
and to select the best of the present routes. The selection of
an optimum metric for resolving routes through the OF is
still a reason for study and that is why different
implementations have been developed evaluating the
characteristics of different metrics.
The IETF ROLL group has also been responsible for
developing two OFs that are used by default in the RPL
protocol, the Objective Function Zero (OF0) [2] and the
Minimum Rank with Hysteresis Objective Function
(MRHOF) [3]. There are different studies and analyses
that compare the performance of both OFs. In [4] the
performance of these two OFs is evaluated considering
several parameters for the two OFs defined in RPL, such
as network convergence time, power consumption, ETX,
number of jumps, PDR and latency. They conclude that
MRHOF has a better performance than OF0 in terms of
network quality, with MRHOF being more suitable for
applications where data delivery and network reliability
are the priority, while OF0 is more suitable for
applications that require rapid network formation and
lower power consumption. Kamgueu et al [5] compared
ETX with a solution based on the remaining energy of the
nodes that form the network, proving that it manages to
extend the useful life of the network. Kim et al [6] deal
with load balancing and congestion problems using RPL
information on packet queues of different nodes. In [7]
Gonizzi et al propose a metric that minimizes the delay,
comparing its performance again with ETX. In [8] the
IETF group defines some of the metrics used for the
calculation of routes that can be used in RPL and [9]
presents a survey that details different metrics that can be
used to evaluate link quality, differentiating between
hardware-based (RSSI, LQI, SNR) and software-based
metrics (PRR and ETX). This survey concludes that the
use of a single metric can only evaluate a single property
of the link, and provides functions that combine some of
these metrics.
Karkazis et al [10] propose some methods to quantify
different routing metrics and different ways of combining
them. Baccour et al [11] propose F-LQE, a link quality
estimation based on four properties: packet delivery,
asymmetry, stability and channel quality. The quality of
the link is specified based on fuzzy rules. The results
demonstrate a higher performance compared to other
existing solutions. In [12] Rekik et al point out that ETX
is not accurate enough and they propose an alternative
metric based on holistic estimation of link quality, where
several link metrics are combined. Gaddour et al [13]
propose OF-FL, an OF that combines a set of metrics to
provide a configurable routing decision.
The proposed solution for estimating communication
quality during WSN deployment introduced in this article
uses some of the aforementioned metrics, but includes new
sampling policies for some metrics to better adapt to the
MAC layer selected, the IEEE802.15e standard operating
in Time-Slotted Channel Hopping (TSCH) mode, while
avoiding the introduction of any overhead due to extra
control traffic.
III. PROPOSED SOLUTION
The solution proposed in this paper uses the link quality
estimator for different purposes. These values obtained
from the quality function can be used as an indicator in
those situations to find the best location for the devices
during the deployment phases of a wireless sensor network
(WSN). On the other hand, the quality metric is used in the
RPL OF to improve the decision-making of the routing
protocol, ensuring that the quality informed to the user
matches the underlying network topology.
Devices that have the ability to establish wireless
communications links should be able to assess the quality
of the path through which the information is routed, in
order to establish robust and highly reliable
communications, characteristics necessary in the
aggressive conditions found in industrial environments. As
a mechanism for evaluating the quality of
communications, the nodes will have the ability to
estimate the level of link quality with surrounding
neighbors, allowing the configurations and topologies that
are formed in the network to improve the reliability and
robustness of communications. There are important
aspects to optimize, such as network coverage, message
latency or topology stability, which are important to
maintain a certain link quality and the service provided by
the network. In practice, these metrics usually come into
conflict, forcing to reach a compromise when optimizing
the general operation of the sensor network.
This quality evaluation mechanism may be used as
follows:
First, allow the operator to know the most optimal
location for placing the nodes, representing in a simple
way different quality values depending on the data that the
node picks up during the deployment phase. Secondly,
quality levels are used as metrics to perform routing tasks,
in order to choose the paths that present a higher level of
quality and thus improve efficiency when retransmitting
messages through the mesh network.
The propagation of radio signals can be affected by
several factors that contribute to the degradation of their
quality. Some of these metrics will depend only on the
radio environment and the hardware configuration
available to the devices, and in other cases, it will depend
on the topology of the network and on the planning of
radio resources.
Among all these parameters, some have been chosen
that will be combined to obtain a multi-objective quality
estimator, which will allow evaluating different aspects of
the quality of the links:
The RSSI parameter, which represents the signal level
with which they are receiving the packages from the
different neighbors within the range of coverage, being a
faithful indicator of the quality of the RF communications
links.
The parameter ETX, which represents the number of
retransmissions necessary for a packet to be transmitted
correctly;
Finally, the number of jumps from one node to the root
has been used, which can be a proportional and
approximate parameter of the latency between a node and
the root, in a deterministic and synchronized network,
without any additional traffic.
4 Telfor Journal, Vol. 10, No. 1, 2018.
Using fuzzy logic operations, these metrics have been
sampled using a series of mapping functions, with the aim
of normalizing the different metrics within a same range
delimited between 128 and 512. For each of the metrics
used, filtering is performed. Exponential Weighted
Moving Average (EWMA) type to smooth changes and
improve stability, but ensuring that the averaged variance
is close to the most recent data. Once the samples for each
of the metrics are obtained, they are combined using a
weighted arithmetic average with different weights for
each metric, which will allow to easily modify the impact
of each of the metrics on the result of the quality estimate.
The mapping range for the different metrics has been
chosen to follow the same philosophy that is used in
current implementations of the ETX metric and the RPL
MRHOF objective function. These limits prevent the use
of floating point format values, which can overload
memory and CPU utilization in this type of restricted
devices typical of WSN. The objective function MRHOF
normalizes the ETX metric by multiplying it by a factor of
128 and since the minimum value of this metric is 1, the
ranges of values used by this function will move between
128 and infinity. As an upper limit, different studies
recommend not using links that have an ETX equal to or
greater than 4, corresponding to a normalized value of
512, since it would imply 75% of packets are being lost.
The process of estimating each metric is detailed below.
A. Received Signal Strength Indicator
The RSSI metric represents the signal level strength of a
received packet. It is a parameter directly measured by the
hardware of the device whenever a message is received.
There are other parameters such as the LQI or the SNR
that represent the quality of the channel, however RSSI is
chosen because it is faster to get a reliable channel
representation, whereas the LQI has a greater variance and
the SNR requires more stages and computational resources
when calculating the noise of the channel [9].
In the networks of wireless sensors deployed, under the
IEEE 802.15.4 standard, and using RPL as a dynamic
routing protocol, several types of messages are exchanged:
beacons and Keep-Alives, ICMPv6 DIO messages
(DODAG1Information Object), DIS messages (DODAG
Information Solicitation) and DAO messages (Destination
Advertisement Object), or data messages using UDP, as
well as all acknowledgment messages (ACK). All these
types of messages mentioned are used to obtain the level
of RSSI with which these messages arrive.
Unlike other solutions that use RSSI values to estimate
channel quality, this solution allows nodes to measure the
RSSI value for each of the channels involved in the
communication. This is relevant because the devices have
been developed under the IEEE 802.15.4e standard using
the TSCH mode that is specially designed to work in
industrial environments. This mode allows the
communication to alternate between different frequency
channels to mitigate negative effects of the channel, such
as interference or fading. As a result, the level of RSSI
will vary over time, because not only a given channel
changes over time, but the communication also jumps
1
Destination-Oriented DAG (Directed Acyclic Graph)
between channels over time. Therefore, when evaluating
the RSSI level of a received message, the channel through
which the message has been transmitted is also verified,
allowing the characterization of this metric to reflect this
time/frequency variability. In order to organize all this
information, in each of the nodes a series of matrixes have
been defined to store the RSSI measured values for each of
the neighbors that are within the range, and for each of the
channels used in TSCH. Fig. 1 shows a representation of
the process of updating values in the commented matrixes.
Each RSSI value for each channel (matrix cells) is
computed and statistically softened with the EWMA filter,
which allows considering past values, to improve stability,
and then the corresponding cell of the matrix is updated.
This type of filtering uses a structure like the one shown in
(1).
𝑅𝑆𝑆𝐼
𝛽 𝑟𝑠𝑠𝑖
1𝛽
𝑟𝑠𝑠𝑖
(1)
Fig. 1. RSSI metric value update diagram.
T
ABLE
1: RSSI
COMPUTING MATRIX FOR EACH NEIGHBOR IN A
NODE
.
𝑅𝑆𝑆𝐼
_
... 𝑅𝑆𝑆𝐼
_
𝑅𝑆𝑆𝐼𝑛 𝑛𝑢𝑚𝑏𝑒𝑟
𝑜𝑓 𝑐ℎ𝑎𝑛𝑛𝑒𝑙𝑠
Node A 𝑅𝑆𝑆𝐼
_
...
𝑅𝑆𝑆𝐼
_
∑𝑅𝑆𝑆𝐼
𝑛
Node B 𝑅𝑆𝑆𝐼
_
...
𝑅𝑆𝑆𝐼
_
∑𝑅𝑆𝑆𝐼
𝑛
Then, for each row (representing a link with a neighbor
in all available channels), the mean is applied to obtain a
meaningful but single value to feed the quality estimator
(Table 1). In this case, β is set between 0.3 and 0.15
depending on the freshness of the actual sample (time limit
set to 10 minutes). This allows to characterize each of the
channels involved in the communication, allowing, for
example, to perform a blacklisting of those channels that
are hindering communication. The mapping function to
introduce RSSI in the quality function is shown in
Fig. 2. Map function for RSSI metric
B. Expected transmission count
ETX is a link quality measure between two nodes in a
wireless network based on the exchange of data packets.
This metric is commonly used in routing algorithms in
mesh networks, for example in the MRHOF RPL objective
function, to minimize packet loss. The ETX is defined
mathematically as shown below (2), where N means the
number of packets sent or received, and subscripts refer to
node i as the sender, and j as the receiver:
Vera-Pérez et al.: Path Quality Estimator for 802.15.4e TSCH Fast Deployment Tool 5
𝐸𝑇𝑋
(2)
In this case, ETX values are not resolved for each
channel due to the very nature of the ETX metric. ETX
counts the number of retransmissions needed to transmit
correctly a packet over a link, but in TSCH,
retransmissions are made in different channels due to the
channel hopping, so it is not possible to give an accurate
value of ETX for a particular channel. ETX calculated
values are also filtered with EWMA. The mapping feature
for the ETX metric, used in the quality function is shown
in Fig. 3. As an alternative, PER (Packet error rate) could
be used to measure how many packets are lost in each of
the channels, however, this would cause the time needed
to process this metric to be proportionally larger,
depending on the percentage of channels used, compared
to the method of obtaining the ETX metric.
C. Number of hops
The metric number of hops in the quality function aims
to provide an approximate and simplified value of the
point-to-point delay between two nodes. An exact
implementation of this parameter would require an
additional exchange of information, in addition to keeping
track of the packets all the way through the WSN. This
would cause heavy network congestion and have a
negative impact on the complexity of the implementation.
Although it is a simple assumption, it can be justified
since the MAC layer follows a synchronized and
deterministic behavior governed by TSCH, with a data
packet generation rate (6 seconds/packet) much higher
than the duration of slot frames scheduled at MAC layer.
Each generated packet needs one slot per hop to be
transmitted. Since we use frames composed of 7 slots of
10 milliseconds, 85 slot frames fit within one period,
leaving a sufficient margin to forward any packets
remaining in the node’s queue. Therefore, a packet end-to
end delay can be assumed to correspond with one frame
duration per hop up to the root. It is true that this
assumption fails to consider retransmissions of failed
packets, but in the overall quality estimation, this is
palliated by the impact of ETX. The function of mapping
of the number of hops used in the quality function is
shown in Fig. 4.
To obtain the number of hops, it is necessary to use a
variable that is incremented with each hop and
retransmitted to the leaf nodes so that they know their
situation within the RPL topology. However, we want to
avoid transmitting additional control traffic that could
overload the network.
Therefore, we use the RPL control messages to include
information about this metric, in order to distribute it
through the topology. The RPL control messages (DIO,
DAO, DIS) are used to build and maintain an RPL
topology, with the aim of transmitting the necessary
information so that the nodes can choose the routes
through which to route the information. In this way, the
range is propagated and updated in one of the fields of
these control messages. The different RPL control
messages also allow to include information about different
metrics in headers that are embedded in messages, called
Metric Containers. In this way, we can include a variable
that is increasing a fixed value with each jump,
propagating these messages down the tree topology,
allowing each node to calculate the distance in hops to the
root.
Fig. 5 depicts a representation of the message exchange
for the calculation of the number of jumps. These discrete
values are not affected over time, with the exception of
changes in the topology; therefore, it will not be necessary
to perform smoothing as we do with the other two metrics.
D. Quality estimator
Equation (3) displays the calculation of the quality
estimation as a combination of the weighted values of the
three metrics used. LQS variable represents link quality, µi
represent the mapped values of the different metrics and αi
represent the weights assigned to each of the metrics.
𝐿𝑄𝑆 𝛼 𝜇 𝛼 𝜇 𝛼 𝜇 (3)
Fig. 2. Map function for RSSI metric.
Fig. 3. Map function for ETX metric.
Fig. 4. Map function for Hop metric.
E. Hysteresis Mechanism
Because the quality estimate will be used to make
decisions when choosing the best routes, it is necessary to
implement a hysteresis mechanism, just as it is done in the
MRHOF objective function [3]. This mechanism allows
6 Telfor Journal, Vol. 10, No. 1, 2018.
improved stability in parent changes in situations where
the metrics of several neighbors have a similar range.
Fig. 5. DIO message exchange mechanism.
In this implementation, having a quality estimator using
different metrics, using a single threshold does not reliably
represent a value that is consistent with the different
metrics. In order to implement this, we use the same
weights that have been used in (3), with thresholds applied
to each metric individually. The margins that have been
chosen will allow acting against changes with a certain
level of precision. In ETX a margin of 0.75 has been
chosen, and 2dB for the RSSI metric. For the number of
hops, given its nature, threshold is 1 hops, which is the
same as no hysteresis.
IV. S
IMULATION AND
R
ESULTS
In order to perform tests and experiments, the Cooja
simulator integrated in Contiki has been used. The nodes
selected are Zolertia Z1, whose msp430 microcontroller
can be emulated in Cooja by the underlying tool MSPSIM,
with some modifications to enable TSCH with the initially
incompatible CC2420 transceiver. The nodes are
configured with a TSCH schedule based on a modification
of Minimal 6TiSCH [14], using slot frames of 7 slots size
with 3 active slots.
The topology chosen sets forces a minimum of 3 hops
in the farthest nodes, preventing a star configuration and
ensuring a multi-hop network. This scenario can be seen in
Fig. 6, with a 100x100m grid, each node has a 50m RF
range, and the separation between nodes in range is of
30m or 42’2m. Normally, a node connects directly with a
parent of its top level, but there are cases in which the
number of hops to the sink exceeds three if the
accumulated metric of the top-level nodes within range is
worse than the metric of neighboring nodes of the same
level. To compare the packet loss, and delay of the
network and test the implemented RPL’s quality estimator
objective function represented in Equation (3), different
weight (αETX, αRSSI, αHops) configurations have been set up,
and 20 simulation iterations for each configuration have
been run.
The following Fig. 7 and Fig. 8 show the results
obtained for 5 different configurations: MRHOF is used
for comparison as is the preconfigured one in Contiki.
Fig. 6. Simulated network.
Then the proposed LQS OF is set with αi in order to
check the impact of each parameter; and giving equal
weight to each parameter in order to test the improvements
of combined metrics.
Fig. 7. Packet loss for the 5 OF configuration simulated.
Fig. 8. Delay performance in every node for the 5 OF
configuration simulated.
Smaller losses are obtained in the LQS configuration
(𝛼 1) if we compare it with the MRHOF function,
whose behaviour should be similar, since both use only the
ETX metric. The only difference between these two
configurations (MRHOF and LQS (𝛼)) is the
maximum limit allowed for the link metric, configured as
1024 in the implementation of MRHOF, and 512 for LQS,
as indicated in RFC6719. This causes the same metric to
behave slightly differently since MRHOF presents some
more losses because it allows communication through
links with lower quality.
10
210
410
610
810
1010
1210
1410
1610
1810
2010
LQS(αHops=1)
MRHOF
LQS(α=1/3)
LQS(αHops=1)
MRHOF
LQS(α=1/3)
LQS(αHops=1)
MRHOF
LQS(α=1/3)
LQS(αHops=1)
MRHOF
LQS(α=1/3)
LQS(αHops=1)
MRHOF
LQS(α=1/3)
LQS(αHops=1)
MRHOF
LQS(α=1/3)
LQS(αHops=1)
MRHOF
LQS(α=1/3)
Node1 Node2 Node3 Node4 Node5 Node6 Node7
milliseconds
Delayforallnodes
min mean max
Vera-Pérez et al.: Path Quality Estimator for 802.15.4e TSCH Fast Deployment Tool 7
The results show that the configuration that combines
the three metrics (𝛼1/3) presents a reduction in the
percentage of packet loss, while maintaining a lower
average delay than for the rest of the configurations (Fig. 8
shows the statistics of delay for the 7 nodes of the network
in the 5 simulated configurations). The values of packet
loss that have been obtained are quite low in all cases, due
to the deterministic medium access (TSCH), but still
getting improvements close to 1% in that interval with
respect to the rest of configurations.
Regarding the average delay, if we focus only on the
node 7, which is the spatially furthest from the coordinator
and therefore with the most critical values, the
configuration that uses a combination of the three metrics
presents a result approximately 3 milliseconds less than
the rest of configurations. In addition, analyzing the results
obtained for LQS (𝛼 1), which should be the ones
with the lowest delay value, it has been concluded that this
metric should be replaced by a better approximation
representing the delay point a point, which will be
reflected in a shorter delay for the LQS objective function
that balances the three metrics.
Finally, in this last graph of Fig. 9 the parent changes
that occur during the simulations of the different
configurations are shown to adapt the topology to changes
in quality. As can be seen in the figures, the configuration
that uses a combination of the three metrics (LQS (𝛼
1/3)) is more sensitive to the quality changes of the links,
allowing to adapt better to choose the paths that present a
more optimal level. In the case of the configuration that
uses only the number of hops, we see that there are no
changes of parent. This is because the metric of the
number of hops is not a variable that changes with the
quality of the link, but it will always choose the shortest
paths, except for the addition of new nodes that allow
reaching the coordinator in fewer jumps.
Fig. 9. Number of parent changes per node for the 5 OF
configuration simulated.
V. CONCLUSION
This paper presented a quality estimator and its
application as routing metric for RPL networks to support
fast deployments. The objective of the proposed metric is
to ensure reliable end-to-end delivery in industrial
environments through efficient link quality estimation.
The proposed metric combines three properties, namely
packet retransmissions, hops and channel RSSI, without
requiring extra control traffic. The metric was evaluated
through COOJA simulations, demonstrating its
outperformance over using single metrics, and the impact
each parameter can have in the overall path’s quality, in
terms of packet loss ratio, end-to-end delay. Further
experiments will comprise the addition of energy metrics
for longer network lifetimes in steady operation, a better
metric to characterize end-to-end delay, and changing OFs
during runtime, as described in [15].
REFERENCES
[1] O. Gaddour, A. Koubâa, S. Chaudhry, M. Tezeghdanti, R. Chaari
and M. Abid, "Simulation and Performance Evaluation of DAG
Construction with RPL," in IEEE Third International Conference
on Communications and Networking (ComNet), pp. 1-8, 2012.
[2] IETF, "RFC 6552 - Objective Function Zero for the Routing
Protocol for Low-Power and Lossy Networks (RPL)," 2012.
[3] IETF, "RFC 6719 - The Minimum Rank with Hysteresis Objective
Function," 2012.
[4] N. Pradeska, Widyawan, W. Najib and S. S. Kusumawardani,
"Performance Analysis of Objective Function MRHOF and OF0 in
Routing Protocol RPL IPv6 Over Low Power Wireless Personal
Area Networks (6LoWPAN)," in 8th International Conference on
Information Technology and Electrical Engineering (ICITEE),
Yogyakarta, Indonesia, 2016.
[5] P. O. Kamgueu, E. Nataf, T. D. Ndié and O. Festor, "Energy-based
routing metric for RPL," Doctoral dissertation, INRIA, 2013.
[6] H.-S. Kim, J. Paek and S. Bahk, "QU-RPL: Queue utilization based
RPL for load balancing in large scale industrial applications," in
12th Annual IEEE International Conference on Sensing,
Communication and Networking (SECON), Seattle, WA, USA,
2015.
[7] P. Gonizzi, R. Monica and G. Ferrari, "Design and evaluation of a
delay-efficient RPL routing metric," in 9th International Wireless
Communication and Mobile Computing Conference (IWCMC),
Sardinia, Italy, 2013.
[8] IETF, "RFC 6551 - Routing Metrics Used for Path Calculation in
Low-Power and Lossy Networks," 2012.
[9] N. Baccour, A. Koubâa, L. Mottola, M. A. Zúñiga, H. Youssef, C.
A. Boano and M. Alves, "Radio link quality estimation in wireless
sensor networks: A survey," ACM Transactions on Sensor
Networks (TOSN), vol. 8 (4), 2012.
[10] P. Karkazis, H. C. Leligou, L. Sarakis, T. Zahariadis, P. Trakadas,
T. H. Velivassaki and C. Capsalis, "Design of primary and
composite routing metrics for RPL-compliant Wireless Sensor
Networks," in International Conference on Telecommunications
and Multimedia (TEMU), Chania, Greece, 2012.
[11] N. Baccour, A. Koubâa, H. Youssef, M. B. Jamâa, D. d. Rosário,
M. Alves and L. B. Becker, "F-LQE: A Fuzzy Link Quality
Estimator for Wireless Sensor Networks," in European Conference
on Wireless Sensor Networks (EWSN), Coimbra, Portugal, 2010.
[12] S. Rekik, N. Baccour, M. Jmaiel and K. Drira, "Holistic link quality
estimation-based routing metric for RPL networks in smart grids,"
in IEEE 27th Annual International Symposium on Personal, Indoor,
and Mobile Radio Communications (PIMRC), Valencia, Spain,
2016.
[13] O. Gaddour, A. Koubaa, N. Baccour and M. Abid, "OF-FL: QoS-
aware fuzzy logic objective function for the RPL routing protocol,"
in 12th International Symposium on Modeling and Optimization in
Mobile, Ad Hoc, and Wireless Networks (WiOpt), Hammamet,
Tunisia, 2014.
[14] IETF, "RFC 8180 - Minimal IPv6 over TSCH Mode of IEEE
802.15.4e (6TiSCH) Configuration," 2017.
[15] M. G. Amor, A. Koubâa, E. Tovar and M. Khalgui, "Cyber-OF: An
Adaptative Cyber-Physical Objective Function for Smart Cities
Applications," in 28th Euromicro Conference on Real-Time
Systems (ECRTS), Toulouse, France, 2016.
[16] J. Vera-Pérez, D. Todolí-Ferrandis, J. Silvestre-Blanes, S. Santonja-
Climent and V. Sempere-Paya, "Path quality estimator for wireless
sensor networks fast deployment tool," 2017 25th
Telecommunication Forum (TELFOR), Belgrade, 2017, pp. 1-4.
0
38
1718
0 00040 000
46
00
76
18
202
32
12
67
167
180
0
69
24
73
92
1
38
66
246
95
0
50
100
150
200
250
300
LQS(αHops=1)
MRHOF
LQS(α=1/3)
LQS(αHops=1)
MRHOF
LQS(α=1/3)
LQS(αHops=1)
MRHOF
LQS(α=1/3)
LQS(αHops=1)
MRHOF
LQS(α=1/3)
LQS(αHops=1)
MRHOF
LQS(α=1/3)
LQS(αHops=1)
MRHOF
LQS(α=1/3)
LQS(αHops=1)
MRHOF
LQS(α=1/3)
Node1 Node2 Node3 Node4 Node5 Node6 Node7
parentchanges
Numberofparentchangespernode
