Estimation and optimization of design parameters in diffusive molecular nanonetworks

Abstract

Advancements in the eld of nanotechnology require an e cient and reliable commu- nication between arti cial or genetically engineered bio-nanomachines having limited operational capabilities. The size, power consumption and computational requirement of transceiver and constraint on antenna size make the electromagnetic (EM) wire- less communication inappropriate at nanoscale. Moreover, macroscale communication inside tunnels, pipelines, or saline water environment can be ine cient due to the in- creased conductivity resulting in large attenuation of EM signals. In view of the limita- tions of the existing EM wave-based technologies, communications among nanomachines require di erent paradigms. This thesis focuses on a new energy-e cient communication paradigm at nanoscale, known as molecular communication (MC), which uses chemicals or molecules as the information carrier. Owing to its biocompatibility, MC has found application in nano- medicine for the targeted drug delivery and the cutting-edge in vivo biomedical appli- cations such as diagnosis, therapy and the monitoring of diseases. However, MC su ers from residual noise, reduced communication range, high latency and lower molecular throughput. As such, we investigate the aforesaid problems for the di usive molecular communication (DMC) systems. To this end, rst we try to mitigate residual noise by adjusting design parame- ters based on signal-to-noise ratio (SNR) estimate at the receiver nanomachine (RxN). Considering molecular concentration-encoded signal with residual noise, we present a SNR estimation using maximum likelihood (ML) principle with sampled observations. Moreover, we derive the Cramer-Rao lower bound and assess the performance of our proposed estimator in terms of its mean square error. Then, we extend our analysis towards improving the communication range, molec- ular throughput and latency of DMC systems by proposing a new relaying strategy, known as estimate-and-forward (EF). The proposed relaying scheme forwards an es- timate of the transmitted number of molecules, which is derived using ML principle. Based on this estimate, we assess the performance of EF relaying in terms of molecular throughput and end-to-end error probability. Further, we calculate energy consumptionin synthesizing the information molecules at the relay nanomachine (RN). Our results demonstrate that the proposed EF scheme performs better than the existing amplify- and-forward scheme and also better than the competitive decode-and-forward scheme when the molecular gap is marginal or when RN is positioned near the RxN. Moreover, it is found that the EF relaying boosts molecular throughput with the same molecular budget as allocated to the baseline direct transmission. Thereafter, we improve the performance of DMC system by optimizing the detection threshold. We came up with a new approach that yields optimal solution for the convex optimization problem in DMC. For this, we make use of logarithmic barrier followed by modi ed Karush-Kuhn-Tucker conditions, Newton Raphson method, and nally round- ing the solution to the nearest integer value. Our proposed optimization technique is more e ective than the existing approaches which provided either suboptimal threshold or optimal one whose convergence is sluggish for the same value of tolerance. Finally, we address the joint optimization of allocation of molecules and RN location for the given detection thresholds. We solve the joint optimization problem by using the block coordinate descent algorithm. Our results reveal the improvement in error performance when molecules distribution and RN placement are in accordance with their joint optimal value. Above all, the theoretical ndings in this thesis provide useful insights and analytical frameworks for the design of reliable DMC systems.