Page 124 - ITU Journal Future and evolving technologies Volume 2 (2021), Issue 3 – Internet of Bio-Nano Things for health applications

ITU Journal on Future and Evolving Technologies, Volume 2 (2021), Issue 3

or confrontation, regarding communication ef iciency in • The sensor nodes have an inite number of
MCN. Game theory has also been used to describe the be‑ molecules, while the relay nodes have a inite size
havioral dynamics of natural MCNs for sharing common molecule reservoir.
resources such as bacteria populations [18] and plant mi‑
crobiomes [19]. • The nodes have a good knowledge of each other and
the location of the sinks, i.e., nodes use a localization
To the best of our knowledge, evolutionary game theory algorithm prior to their operation.
itself has not been considered in the literature for the
MCN resource allocation problem. The success of evolu‑ • The existing nodes do not cause any obstruction for
tionary game theory in application to biological problems, the nodes behind them.
including resource allocation in organisms [20] shows its
applicability to the nature inspired MCN.
The general operation of the organism is depicted in Algo‑
In this paper, we apply evolutionary game theory to the rithm 1. Firstly, one of the sensor nodes emits a message
resource allocation problem in MCNs. The evolution pro‑ (lines 10‑12). When a relay node receives the message,
cedure relies on selecting successful MCNs, where the it relays the message to another node until the message
success criterion is the total number of successful trans‑ reaches the sink (line 26). If several relay nodes receive
missions, followed by creating their offspring iteratively. the message, a random one among them, which is closer
By this approach, we have a population of MCNs which to the sink than the previously emitting relay node (lines
is generally better in terms of transmission count as we 14‑17), transmits the message (lines 18‑19). The organ‑
increase the iteration number. In a way, this approach re‑ ism is considered dead if messages are dropped con‑
sembles machine learning. In other words, our simulation secutively or if reservoirs of % of the relay nodes are ex‑
behaves as an evolutionary machine learning algorithm hausted (line 9). The latter condition is included to sup‑
based on mechanisms of evolutionary game theory. To il‑ port the former. Without this condition, if the consecutive
lustrate the evolutionary approach, we use a toy problem drop cycle is broken by a lucky transmission, at least
for resource management in an MCN and then provide re‑ more attempts are made, which increases the drop counts
spective analytical and evolutionary solutions compara‑ and overall performance parameters are disrupted.
tively. We also demonstrate our simulation using a ran‑
domly generated MCN. 2.2 Molecular communication model

The rest of the paper is organized as follows. In Section In this work, we use a simplistic model capturing all the
2, we describe the system model, i.e., the operation of the essential mechanisms of MC without the computational
MCN and evolution procedure. In Section 3, we present an burden. As with any MC model, our model comprises in‑
analytical and an evolutionary solution for a toy problem formation carriers, medium, transmission, and reception,
comparatively to illustrate how evolution works. In Sec‑ as described in detail as follows.
tion 4, we simulate a randomly generated MCN, using the
evolutionary approach, and then discuss its performance. 2.2.1 Information carrier
Finally, we conclude the paper in Section 5.
As the name suggests, the information carrier is a
2. SYSTEM MODEL molecule that can diffuse and propagate in a medium. The
molecules do not interact with each other, and they have
We use an organism as an MCN. The organism consists of a constant, isotropic diffusion constant, . The half‑life of
nodes communicating with each other using MC. In Sec‑ the molecules is assumed long compared to the propaga‑
tion 2.1 we describe the organism in detail and in Section tiontimeandshortcomparedtothetimebetweenconsec‑
2.2 we will present the system model for MC. Finally, in utive transmissions of the nodes. Therefore, the molecule
Section 2.3, we will illustrate the evolution procedure. count does not drop during propagation to other nodes.
Moreover, we assume that the channel is cleared between
2.1 Organism consecutive transmissions.

An organism has three types of nodes: sensor nodes col‑
lecting information about their surroundings, sink(s) op‑ 6 0
erating as gateways to the Internet, and relay nodes trans‑ 1
mitting the information they received from the sensor 5 4 8
nodes to the sink(s). Because of the extremely small sizes 2 3
of the relay nodes, they are randomly distributed in a vol‑ 7
ume. However, sensor nodes and the sink are larger, so
they are not necessarily distributed randomly. Fig. 1 de‑ Fig. 1 – An example of an organism, with sink shown as red and sensors
picts the organism. We made the following assumptions nodes shown with yellow rings around them. The other nodes are relay
regarding the operation of the nodes: nodes.

112 © International Telecommunication Union, 2021