Page 72 - ITU Journal Future and evolving technologies Volume 2 (2021), Issue 3 – Internet of Bio-Nano Things for health applications
P. 72

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




                                         Table 6 – BER performance of different detection schemes
           Reference   Modulation   Detection                      Lowest BER      SNR/Number      Distance
                                                                                   of  transmitted  between
                                                                                   molecules       transmitter
                                                                                                   and receiver
           [33]        MoSK         i) MAP detection without ISI   i) 5 × 10 −3    4 dB/10 3       100 µm
                                    ii) Viterbi detection with ISI  ii) ≈ 10 −5
           [34]        MoSK         MAP detection with noise whitening  7 × 10 −6  10 dB/2.5 × 10 4  600 µm
                                     ilter
           [40]        Rectangular  Adaptive receiver based on steep‑  i) 0        16 dB/4 × 10 9  0.05 µm
                       pulse  based  est descent algorithm for i) Maxi‑  ii) 3 × 10 −5
                       OOK          mum likelihood sequence detection  iii) 10 −6
                                    ii) MMSE equalizer iii) DFE with de‑  iv) 3 × 10 −6
                                    cision metric iv) DFE with quantizer
           [45]        MTSK         i) MMSE equalizer              i) ≈ 10 −4      500             5 µm
                                    ii) DFF                        ii) 3 × 10 −4
           [52]        Rectangular  Non linear adaptive threshold detec‑  2 × 10 −4  8 dB          9 nm
                       pulsed based  tion based on local geometry and en‑
                       OOK          ergy difference of received signal
           [60]        Rectangular  Two layered detection for reduc‑  10 −5        9 dB/10 4       3 µm
                       pulsed based  ing the number of sequences to be
                       OOK          searched
           [29]        OOK          ANN detection with LM optimizer  ≈ 10 −4       31 dB           0.5 µm
           [105]       OOK          MF detector                    2 × 10 −4       4 × 10 4        0.5 µm
           [30]        OOK          Parzen‑PNN based detection     5 × 10 −4       10 dB           2 µm
           [111]       CSK          Fuzzy C‑means clustering       ≈ 10 −3         11 dB/3 × 10 4  0.5 µm
           [135]       OOK          Local convexity detection      2 × 10 −4       31 dB           0.065 µm
           [31]        Binary and M‑  Recurrent neural network detection  5 × 10 −4  ‑             ‑
                       ary amplitude  schemes
                       modulation
                                                                                         4
           [137]       OOK          Single sample detector with initial  4 × 10 −3  3 × 10 for pilot  1 µm
                                    distance estimation based on pilot             signal and 4 × 10 4
                                    signal                                         for information


          3.2.3  Distribution detection‑based mobile MC        Authors  demonstrated  that  for  a  higher  probability  of
                 systems                                       false alarm, OR rule performed better than AND rule and
                                                               vice‑versa.  This is because in case of high chances of er‑
          In [157], 1‑D MMC with drift was considered, where mul‑
                                                               ror,  all  the  nano‑machines  may  not  be  able  to  make  the
          tiple cooperative nano‑machines were used between the
                                                               correct decision, and hence performance of the AND rule
          source  and  the  destination.  Channel  capacity  and  the
                                                               degrades.  Further,  to reduce the error probability,  a co‑
          probability  of  bit  error  expressions  were  derived  at  the
                                                               operative  abnormality  detection  scheme  was  proposed
          destination using the probabilities of detection and false
                                                               in  [158].  In  this  scheme,  a  sensor  could  be  activated  if
          alarm.   Authors’  analysis  demonstrated  that  the  opti‑
                                                               it  detects  an  abnormality  itself  or  it  receives  signaling
          mal decision threshold at the destination depends on the
                                                               molecules from other sensors that detected abnormality.
          probabilities of detection and false alarm of the last coop‑   Finally, an FC collected responses from all the sensors and
          erative nano‑machine.
                                                               checked the activation  lag of all the sensors to decide the
                                                               presence or absence of abnormality. Optimal threshold at
          A cooperative detection strategy was presented in [151],
                                                               the FC was derived by minimizing the error probability.
          where several cooperative nano‑machines sent their de‑
          cisions to an FC about the presence or absence of an ab‑
          normality inside the blood vessel.  In this work,  each of   3.3  Performance  and  complexity  comparison
          the nano‑machines and the FC were assumed to be mo‑        of  different  detection  techniques  for
          bile under the in luence of both diffusion and drift.  For   mobile MC
          performance analysis, the PDF of the  irst hitting time of
          molecule at the FC was also described, where the concept   Computational  complexity  of  the  non‑coherent  detector
          of effective diffusion coef icient was used as described in   proposed in [145] is O(N s ) that is much less than the de‑
          [117]. OR (if one of the nano‑machines sent a positive re‑   tection  schemes  such  as  coherent  MAP  method  and
          port)  and  AND  (if  all  the  nano‑machines  sent  a  positive   MMSE [40], [98].
          report) rules were used at the FC for making the global
          decision about the presence or absence of abnormality.



          60                                © International Telecommunication Union, 2021
   67   68   69   70   71   72   73   74   75   76   77