Page 79 - 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

based detection can be robust detectors [31] as they do In this context, one of the possible research areas is neural
not require CSI and are also capable of working in the high network detection considering the channel model which
noise environment. is not known or dif icult to obtain. Further, the per‑
formance of different irst order and second order algo‑
rithms [198] used for optimizing the neural network de‑
5.3 Future research directions tectors should be investigated.

Within IoBNT, multiple transmitters and receivers have On the other hand, most of the detection schemes as‑
to work together to perform complex tasks, including sume perfect synchronization between the transmitter
sensing and actuation. However, exact analytical chan‑ and the receiver. However, for practical MC systems, joint
nel models considering multiple fully absorbing receivers synchronization and detection or asynchronous detection
in the medium are not available in the existing literature has to be performed, as shown in [163] and [50], re‑
due to mathematical intractability of corresponding diffu‑ spectively. In this context, novel low‑complexity schemes
sion equations. In this context, the work in [196] consid‑ should be devised to perform asynchronous detection or
ered two fully absorbing receivers in a 3‑D medium and detection with synchronization at the fully absorbing re‑
derived an approximate irst hitting time distribution to ceiver, especially for a low‑induced mobile MC where
demonstrate the mutual dependency between receivers. each of the nano‑machines are considered to be mobile in
However, this approximation is only valid when r 1 > 3a, diffusive channel along with drift. Moreover, the state of
r 2 > 3a, and R > 3a, where r 1 , r 2 , R, and a denote the the art detectors [45], [52], [118], [146] considered per‑
distance between the transmitter and irst receiver, dis‑ fect synchronization while analyzing the system perfor‑
tance between the transmitter and second receiver, dis‑ mance. Thus, the performance evaluation of these detec‑
tance between the irst and second receiver, and the ra‑ tors considering synchronization error is still lacking in
dius of the receiver, respectively. In this derivation, the the current literature.
radius of each receiver is assumed to be identical.
6. CONCLUSION
The analysis was further extended for an underlay‑based For the IoBNT applications such as drug delivery, in‑body
cognitive paradigm in [197] where both primary and sec‑ health monitoring, etc, the nanoscale and microscale de‑
ondary link performances were evaluated by employing vices are expected to perform collaborative tasks using
a simple molecule control mechanism at the secondary MC. However, the MC system performance in these ap‑
transmitter. In this work, the radius of each receiver plications signi icantly depends on the transmission and
is assumed to be different and the impact of molecule detection schemes employed at the transmitter and re‑
degradation over time was also considered while analyz‑ ceiver nano‑machines (or bio‑nano‑machines), respec‑
ing the system performance. However, the analyses in tively. This survey, therefore, presented the transmis‑
[196] and [197] is restricted only for two fully absorbing sion and detection techniques existing in the current lit‑
receivers and cannot be easily extended for more than two
erature for static and mobile nano‑machines under pure‑
receivers. Thus, development of exact analytical channel diffusive and low‑induced diffusive channels. In each
models involving multiple fully absorbing receivers in a category, different types of MC system such as SISO,
3‑D medium is still an open problem.
MIMO, relay‑assisted, and FC‑based cooperative detection
scheme have been discussed to support several health
Further, many MC systems [33], [40], [45], [118], [148], applications within IoBNT. Various coherent and non‑
[136], [144] that are proposed in the literature are co‑ coherent detection schemes are presented under each
herent. These coherent MC systems require CSI and suf‑ category. The detectors have also been classi ied based
fers from the drawback of complex channel estimation. on symbol‑by‑symbol detection and sequence detection.
Note that the channel in these systems can be very un‑ Also, theperformanceandthecomplexitiesofsomedetec‑
predictable with very short coherence times. Hence the tion techniques are discussed. Further, several challenges
MC systems that use pilot signal [137], [148], [187] for in detection have been described under various scenarios.
estimating the CSI, are not very suitable. Also, the re‑ Experimental works related to MC are also presented. At
ceivers that can detect the information for fast‑varying the end of this survey, some major challenges related to
channels (i.e., short coherence time) are required to build the practical design of the transmitter and receiver along
practical MC systems. Machine/deep learning‑based MC with future research directions have been added.
systems that do not need CSI and perform well in fast‑
varying channels [31] are suitable in such complex sce‑
narios. Also, machine/deep learning‑based algorithms do
not rely on accurate channel models and can be classi ied
as non‑coherent schemes. Hence, the possibility of imple‑
menting these algorithms should be explored to address
IoBNT applications with static [30], [186], [189] and mo‑
bile MC [106], [31].

© International Telecommunication Union, 2021 67