Page 114 - ITUJournal Future and evolving technologies Volume 2 (2021), Issue 1

ITU Journal on Future and Evolving Technologies, Volume 2 (2021), Issue 1

Fig. 10 – Packet Delivery Ratio per node. Fig. 11 – Energy consumption per RAT.
works, multi‐hop networks...). In our scenario, nodes are 11. EXPERIMENTAL RESULTS
deployed throughout a ield used for cultivating crops.
The simulated setup is illustrated in Fig. 9. Five nodes Topology and Packet Delivery Ratio (PDR) are measured.
monitor environmental metrics useful for farmers. Nodes Nodes transmit at an interval randomly picked in [2; 4]
have to of load numerical data on a regular basis while seconds. We consider a population of 20 experiments
saving up power. They may have to send an alarm if a lasting 10 minutes each. A small population is suf icient
metric becomes off chart, putting the crops at risk (e.g., because of the low standard deviation. Longer experi‐
temperature). ments are not relevant because the network stabilizes af‐
ter a few messages have been exchanged. We do not di‐
Out of the ive RATs available on FiPy, we are using WiFi, rectly compare RODENT’s results to related works as the
LoRa and BLE in this scenario. Sigfox and LTE‐M/NB‐IoT heavy difference between proposals makes it irrelevant,
are not open technologies, so we could not use them di‐ and increased lexibility cannot be measured. In this sec‐
rectly. LoRa and BLE links are more interesting in terms tion we present the results obtained.
of energetic savings than WiFi. Each node ( ) is in a dif‐
ferent situation. 1 is the control node, it only has a WiFi 11.1 Topology
link with the WiFi BS. 2 can reach the WiFi BS and ben‐
e its from the LoRa link when RODENT is active. 3 has With the use of RODENT, the MTN’s topology changes. 1
to choose between reaching the WiFi BS directly at a high does not change its link because it can only reach the WiFi
energy cost or forwarding its data to its neighbor 1 via a BS. 2 uses the LoRa link instead of the WiFi link, because
BLE link. 4 needs to be able to send regular monitoring it costs less energy. 3 decides to use the BLE link to of‐
data as well as alarms, via WiFi or LoRa. 5 is an isolated load its data to 1, which in turn forwards it to the WiFi
node, deployed too far away to directly communicate with BS. 4 of loads its monitoring data to the LoRa BS, to re‐
the WiFi BS. Farms are usually located in wide rural en‐ duce energy consumption compared to WiFi. It can still
vironments, unfriendly to wireless waves because of tall use the WiFi link to forward alarms that needs a quicker
crops (e.g., corn). Thus white zones and isolated nodes RAT at the expense of a higher energy cost. 5 is not iso‐
are common. Using RODENT, 5 can forward its data to lated anymore, as it forwards its data to 4 through LoRa
its neighbor 4 using LoRa. which will of load it to the LoRa BS in turn.
We run three types of experiments. First, RODENT is not 11.2 Packet Delivery Ratio
active and nodes only use WiFi links, depicted in blue in
Fig. 9. Second, RODENT is active, which allows nodes to The Packet Delivery Ratio (PDR) is the ratio between the
switch to LoRa and BLE links, depicted in red and green total packets received and the total packets sent. The
in Fig. 9. Third, RODENT is active, each LoRa message is PDR of every node taking part in the MTN is depicted in
sent two times and each BLE message is sent three times, Fig. 10 along with its standard deviation. 1’s PDR does
which increases the network’s reliability. A video of an not change, as its route remains the same. Without RO‐
1
experiment running is available online . DENT, 5’s PDR is null as the node is isolated and can‐
not of load a single data packet. The PDR of 2, 4 and
5 is around 80% with RODENT which allows them to
use LoRa. It is not the same as WiFi because of frequent
1 http://chercheurs.lille.inria.fr/bfoubert/ressources/ collisions, as nodes do not use a proper MAC. 3’s PDR
rodent.mp4 is around 60% with RODENT. The node forwards its data

98 © International Telecommunication Union, 2021