Page 74 - 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

quired as there are three unknown variables to be deter‑
160
mined. By solving three different equations similar to (6)
r=3.14 m r=3.52 m
r=2.76 m
r=2.38 m
, and
Number of received molecules, y(r(t),t) 100 as the distance r changes at each sampling time, we can‑
r=2 m
140
with three samples of the received signal N rx 1
, N rx 2
, three unknown parameters i.e., N tx , D, r can be es‑
N rx 3
timated. However, for a mobile transmitter and receiver,
120
not estimate ive unknown variables i.e., N tx , D, r 1 , r 2 , r 3
, and
80
using three samples of the received signal N rx 1
, N rx 2
. Hence, non‑coherent detectors [119] and ANN de‑
N rx 3
60
tectors [31] are more suitable in this case.
40
Unknown channel model
3.4.3
20
If channel model is not known then the adaptive
0 threshold‑based detection techniques used in [118] and
0 2 4 6 8 10 [136] will not yield good performance because (6) can not
Time (s)
be used for distance estimation. Therefore, to deal with
Fig. 20 – Received signal including noise and ISI for the transmitted se‑ unknown channel model, non‑coherent detection tech‑
quence [1 1 0 1 0]. The maximum signal is below the threshold in 4th bit niques proposed in [145], [146], [119] can be considered.
duration for bit‑1 leading to incorrect detection.
In [146], amplitude difference was used as the decision
The detection based on the convexity metric is also metric, whereas energy difference was used as the de‑
suitable for the system experiencing only ISI, where cision metric in [145]. Also, the technique proposed in
noise can be iltered by using a moving average ilter as [145] is suitable for the scenario when the system expe‑
described in [135]. Further, to improve the perfor‑ riences strong ISI. Further, to improve the detection per‑
mance, the ISI mitigation technique described in [118] can formance in noisy and unknown channel scenarios, ANN
be considered under mobile scenarios. In this work, dy‑ detectors [31] can also be considered as one of the possi‑
namic distance is estimated using (6) if the number of re‑ ble solutions.
ceived molecules N rx (r, t) is known. This estimated dis‑
tance is further used to reconstruct the received signal 3.4.4 Synchronization
(within a bit‑interval) and subtract it from the total re‑
For a static MC scenario, few synchronization techniques
ceived signal for ISI mitigation. Note that the distance es‑
such as blind synchronization [161], [162], synchroniza‑
timation using (6) in [118] is not correct if bit‑0 is trans‑
mitted as N tx is zero for bit‑0. tion using peak observation time and threshold triggering
[163], and reference broadcast synchronization consider‑
ing molecule synthesis time [164] have been proposed.
Hence, in [118], the adaptive threshold for detection in the
Note that these synchronization schemes proposed for
current bit‑interval was calculated using the distance esti‑
static MC cannot be used in MMC scenarios. Moreover,
mated in the bit‑interval when the last bit‑1 was detected.
synchronization is more challenging in mobile MC due
If the coherence time of the channel is small then the
to the time‑varying distance between the communicat‑
channel can change considerably between the current bit‑
ing nano‑machines. It is also worth noting that joint de‑
interval and the bit‑interval in which the last bit‑1 was de‑
tection and synchronization have to be done to realize
tected. Hence, the detection threshold can be wrong. This
practical mobile MC systems. Recently, two synchroniza‑
problem can be solved by transmitting some molecules for
tion schemes based on the least‑square method and peak
bit‑0, as well [136]. Considering the non‑zero number of
time have been proposed for mobile MC in [165]. How‑
molecules for bit‑0, the adaptive threshold in the current
ever, these techniques rely on a known channel model and
bit‑interval can be obtained using the distance estimated
will not work if extended for unknown channel models
in the previous bit‑interval. Also, for MMC in a drift‑based
e.g., channel model with multiple fully‑absorbing receiver
environment, threshold determination has been shown in
[152]. nano‑machines. Considering limited computational re‑
sources for nano‑machines, investigation of novel asyn‑
chronous detection techniques for mobile MC is required.
3.4.2 Multiple parameters estimation
If the number of transmitted molecules N tx and the diffu‑
sion coef icient D change along with the distance r then
the estimate of these parameters over time is dif icult to
obtain using (6). In [161], a method for estimating multi‑
ple parameters has been shown. If we consider that N tx ,
D, and r are constant within a bit‑interval then three sam‑
ples of the received signal within a bit‑interval will be re‑
62 © International Telecommunication Union, 2021