Page 56 - 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

the amplitude of the signal reduces quickly. However, ligand‑receptor type communication. The irst technique
higher order derivatives may amplify the noise. Hence considered the likelihood of observing bound receptor
the optimal derivative order is required for a trade‑off be‑ molecules given the current transmitted bit and the esti‑
tween the noise ampli ication and ISI mitigation. For de‑ mate of ISI bits. The other one considered the likelihood
coding, maximum likelihood sequence detection with re‑ of observing the unbound time of receptors for detection.
duced channel memory was used in [62]. The latter technique experienced better BER performance
because the bound state of receptors can provide little in‑
In [63], a variant of OOK was proposed in which two dif‑ formation when the receiver saturates due to the ISI.
ferent types of molecules were considered for transmis‑
sion of bit‑1, and no molecules were released for bit‑0. Further, three different detection techniques based on the
For bit‑1, the type‑A molecule was transmitted at the be‑ number of bound receptors, unbound time (around the
ginning of a bit‑interval, and the type‑B molecule was sampling time) of the receptors, and bound time of the
transmitted at the peak time of the type‑A molecule. Fur‑ receptors were proposed in [67]. The detection based
ther, in [63], another modulation scheme was the Order on the bound time of the receptors was better in perfor‑
Shift Keying (OSK) in which type‑A molecule was trans‑ mance than the other two schemes. In this work, an esti‑
mitted at the beginning of a bit‑interval, and the type‑ matorbasedonthelog‑likelihoodoftheratioofligandand
B molecule was transmitted at the peak time of type‑A total concentration was also proposed for the detection.
molecule for bit‑1 and the order of molecules transmis‑ In [68], chemical reaction networks for detecting binary
sion was reversed for sending bit‑0. These procedures MoSK signals were proposed. In this work, sampling was
reduced the ISI at the receiving nano‑machine as type‑ based on transcription networks and the demodulation
A and type‑B molecules were assumed to react together. process used an integral feedback controller that senses
The schemes proposed are shown in Fig. 15. In addition whether a chemical signal is above the threshold or not.
to these schemes, a suboptimal maximum likelihood de‑ The demodulator output is used for detecting symbols us‑
tection scheme and ISI neglecting detection scheme were ing the biological XOR gate.
also proposed.
In [69], OOK with rectangular pulse shaping was used
The authors in [64] proposed an optimal detec‑ at the transmitter. The transmitted pulse was convolved
tion scheme based on Accelerated‑Particle Swarm with the CIR to obtain the received signal. Further, at the
Optimization (A‑PSO). This scheme used an ac‑ receiver, the detection process was carried out in three
celeration factor to ind the optimal weights steps: (i) stochastic resonance‑based nonlinear iltering
], such that the er‑
w w w = [w jN s , w jN s +1 , · · · , w (j+1)N s (processing based on local transient features), (ii) calcula‑
ror probability is minimum as described below. tion of the non‑coherent metric, and (iii) threshold‑based
detection. The decision metric was constructed by adding
w w w = argmin P e . (4)
w w w the metrics of the convex property of the iltered signal,
transient shape among the symbols, and the energy dif‑
Finally, using these optimal weights, a weighted sum de‑ ference between the successive symbols.
tector was employed at the receiver i.e., each sample
within a bit‑interval was multiplied by a weight and all The derivation of an event (e.g. presence of a cancer cell)
the weighted samples were added to compare against the detection probability was presented in [70], where the
threshold θ for detection, event was de ined as the hitting of a single molecule at any
{
∑ (j+1)N s one of the multiple receivers within a certain period. The
1 if w k y k ≥ θ;
ˆ b j = k=jN s (5) detection probabilities were derived for both degradable
0 otherwise, and non‑degradable molecules. In this work, the centers
of the receivers were assumed to be distributed as a Pois‑
where w k ≥ 0. This detector was shown to perform better
than the MF detector. son point process. Also, the number of molecules to be
transmitted for a speci ied probability of event detection
Further, a suboptimal detection scheme was presented in was derived.
[65], where the CSI was modeled as a Gamma distributed
random variable. A non‑coherent decision feedback de‑ Rectangular pulse‑based OOK was used as the modulation
tection was also proposed therein, which utilizes statisti‑ in [71]. Further, a weighted sum method with optimal
cal CSI. The proposed decision feedback detector consid‑ weights was proposed for detection and ISI mitigation at
ered a ixed detection window of size K where a constant the receiver. Optimal weights were obtained by maximiz‑
CSI was assumed. Further, a blind CSI estimation‑based ing the Signal‑to‑Interference‑plus‑Noise Ratio (SINR) i.e.,
detection was also proposed in which CSI estimates were differentiating SINR with respect to weights and setting it
found by averaging over the expected positions of bit‑1 equal to zero. Also, a block‑wise data detection‑based it‑
and bit‑0. erative method for ISI mitigation was proposed where the
expected ISI was subtracted from the total received signal.
In [66], two different detection schemes based on the Simulation results showed that the iterative method out‑
maximum likelihood criterion were presented for a performed the weighted sum method for ISI mitigation.

44 © International Telecommunication Union, 2021