Page 16 - ITU KALEIDOSCOPE, ATLANTA 2019
P. 16
The physical domain of PANACEA will comprise all the bio-nanosensors and actuators (e.g. drug
delivery devices, pacemakers, etc.) embodied by the RIMOR (explorer in Latin) subsystem, which
consists of 3 parts: bio-nanosensor, sensor interface chip, and a coil/inductor for wireless
communication to wearable hub outside of the body. The bio-nanosensor can be diversified by
sensing quorum sensing signals directly or via a reporter bacteria. Moreover, the signals generated
by bacteria can be sensed by utilizing electro-chemical or fluorescence methods. The bio-nanosensor
of RIMOR, has two parts, namely, the bacterial sensor and the physical sensor. The bacterial sensor
senses molecular communication signals generated by the bacteria in the body, and produces light
detected by the physical sensor which converts light to electrical current. This way, MC signals are
transduced to electrical signals to be further relayed to the wearable hub. Interactions between
physical and cyber domains are established by heterogeneous wireless communication modules that
utilize radio-frequency (RF), ultrasonic and molecular communications through RIMOR and
wearable devices.
The cyber part of the PANACEA is in charge of collecting sensing data and performing complex data
processing and learning procedures for the early detection of diseases and infections. The access to
PANACEA is made possible by the Human-Machine Interface (HMI), which provides an easy and
intuitive Data Visualization Interface (DVI) enabling the visualization of relevant information of each
patient and provides alert message management to notify both caregivers and patients when an
infection occurs. The DVI allows human-in-the-loop control thus making it possible for caregivers to
dynamically and actively interact with the system and to regulate drug delivery through ad-hoc
control primitives and APIs exposed by actuator devices. PANACEA not only facilitates interactions
with humans, but it also enables advanced automated drug delivery systems that rely on supervised
machine learning. The learning block is fed with both data collected by the physical system and
supervised input data generated by caregivers. Such an approach makes it possible to train
PANACEA with patient-dependent data so that individual medical treatments can be achieved for
each patient.
Even though applications such as PANACEA are very promising since they are based on the better
defined and more studied MC technique of bacterial communication, a plethora of biomedical
applications can be enabled by the rest of the MC techniques such as calcium signaling, nervous
networks, endocrine network, and molecular motors. The standardization efforts in molecular
communication started in 2014 with the IEEE P1906.1.1 - Standard Data Model for Nanoscale
Communication Systems and they have released IEEE 1906.1-2015 - IEEE Recommended Practice
for Nanoscale and Molecular Communication Framework. Although this standard is a step towards
developing MC as an implementable technology, it only covers the basic diffusion-based molecular
communication and it also includes THz band communication under the nano-communication
umbrella which overlooks underlying challenges arising from the biological nature of MC. Despite
the prior work in the field on the channel characterization, estimation, and capacity calculations of
these aforementioned techniques, a unifying information-theoretic framework that captures the
peculiarities of an MC channel over classical communication systems for all the various MC
techniques, is currently missing.
We aim at filling the aforementioned research gap by providing a mathematical framework rooted in
chemical kinetics and statistical mechanics to define the main functional blocks of MC, to abstract
any MC system and determine or estimate the information capacity of their communication channels.
By using the general formulation of the Langevin equation of a moving nanoscale particle subject to
unavoidable thermally driven Brownian forces as a unifying modeling tool for molecule propagation,
we build a general mathematical abstraction of an MC system. Then, we derive a methodology to
determine (or estimate, whenever closed-form analytical solutions are intractable) the MC channel
capacity based on the decomposition of the Langevin equation into two contributions, namely,
propagation according to the Fokker–Planck equation followed by a Poisson process.
We classify diverse implementations of MC based on their underlying physical and chemical
processes and their representation in terms of the Langevin equation. MC systems based on random
– xiv –