Page 106 - ITUJournal Future and evolving technologies Volume 2 (2021), Issue 1
P. 106
ITU Journal on Future and Evolving Technologies, Volume 2 (2021), Issue 1
decrease the devices’ lifetimes and may overload the de‐ Section 8 details RODENT’s inner workings. Section 9
vices’ limited memories which leads to hardware failure. presents the hardware used and irmware implemented
We address those issues in this paper, by proposing a for our MTN prototype. Section 10 introduces the exper‐
lightweight TOPSIS‐based NIS method optimized for WSN imental setup and scenario. Section 11 details the exper‐
devices. Furthermore, our method simpli ies TOPSIS iments’ results. Section 12 concludes this article and lists
computations and completely eliminates rank reversal by future work.
modifying the TOPSIS normalization algorithm. This re‐
sults in less complexity and provides time and energy sav‐ 2. RELATED WORK
ings.
Currently available routing protocols are not suited for Several works have been conducted to mitigate rank re‐
versal in TOPSIS or to apply TOPSIS to NIS. To the best
MTN. In this article, we introduce a novel Routing Over
of our knowledge, only few works exist in the literature
Different Existing Network Technologies protocol (RO‐
about multi‐technology network. This section presents
DENT) designed for MTN leveraging our custom TOPSIS
related work about TOPSIS and multi‐RAT devices.
method. Our contribution takes every RAT of each node
into account for the route selection. Every node has a list
of available links between itself and its neighbors. Links 2.1 TOPSIS method
have associated costs and performances, in terms of de‐ [4] proposes an iterative TOPSIS method, where TOPSIS
lay, energy consumption etc. A node constructs its routes is executed, then the worst alternative is removed from
based on its links’ values and the routes’ values shared by the ranking, and TOPSIS is re‐executed, as long as there
its neighbors. Criteria for the best route depend on the is more than one alternative in the ranking. The remain‐
use case and the requirements data has to meet (e.g., data ing one is selected as a communication technology. [5]
size, deadline). combines TOPSIS with fuzzy logic, in order to improve
RODENT is implemented and its performances are as‐ how uncertain attributes are taken into account. [6] in‐
sessed through experimental evaluation. Results show troduces alternative methods based on TOPSIS, but with
that RODENT increases network lexibility and reliability, different normalization algorithms using maximum and
decreases energy consumption and enables better con‐ minimum values of the attributes. [7] compares several
sideration of the data requirements while maintaining a NIS methods applied to heterogeneous WSN. [8] intro‐
good Packet Delivery Ratio (PDR). Compared to related duces an original MADM method along with an in‐depth
work, RODENT offers a lexible and dynamic way to over‐ analysis of TOPSIS. [9] proposes a new Service‐based In‐
come WSN’s limitations without the need of a dedicated terface Selection Scheme algorithm based on TOPSIS to
infrastructure other than multi‐RAT nodes. enable NIS applied to vehicle‐to‐vehicle communications
The contributions of this paper can be summarized as fol‐ scenarios. [10] details a fast TOPSIS‐based NIS technique
lows:
for vertical handover in heterogeneous emergency com‐
munication systems.
• We designed a lightweight selection method for WSN
Overall, those propositions reduce the probability of oc‐
based on TOPSIS, free of rank reversal which shows
currence of rank reversal, but does not nullify it because
an improvement in the computation time of around
the euclidean normalization is still used. Furthermore,
38%, which in turn results in energy savings, while
the technology selection is equivalent to using the some of the proposed modi ications tend to increase the
classic TOPSIS method in 82% of the experiments. complexity of the TOPSIS method. This would increase
the execution time of TOPSIS and in turn the energy con‐
• We designed a multi‐technology routing protocol for sumption of the nodes, thus reducing their lifetime.
WSN based on our custom selection method. It is ca‐ To the extent of the authors knowledge, no works has
pable of handling multi‐technology devices and se‐ been conducted to propose a rank reversal free TOPSIS‐
lecting the best route and technologies for speci ic based method for NIS speci ically for energy constrained
data requirements. devices. Thus, in this in paper we introduce a lightweight
TOPSIS‐based NIS method that aims not only to reduce
• We designed and developed an MTN prototype com‐ the complexity and energy consumption of TOPSIS, but
posed of Pycom FiPy devices running a custom im‐ also to completely eliminate rank reversal.
plementation of RODENT.
2.2 Multi‐technology networks
The rest of this paper is organized as follows: Section 2 in‐
troduces the work related to MTN and TOPSIS. Section 3 The authors of [11] propose an IoT architecture for multi‐
presents the background about MADM and TOPSIS. Sec‐ RAT devices. This architecture is based on a network con‐
tion 4 explains what issues have to be faced with TOP‐ vergence layer in charge of the multi‐RAT management
SIS. Section 5 details our lightweight TOPSIS method. Sec‐ in nodes, and a heterogeneous network controller on the
tion 6 presents our experiments on the selection method network operator side. It also proposes a hardware plat‐
and the results we have obtained. Section 7 exposes the form for the nodes, a polling scheme as well as a compres‐
network model and assumptions RODENT is based on. sion scheme based on Static Context Header Compres‐
90 © International Telecommunication Union, 2021