Research On Source Localization Of Molecular Communication Via Diffusion

The rapid development of nanotechnology has made it possible for its application in the fields of disease prediction,drug delivery and environmental monitoring.Nanomachines are connected to each other to form a nanonetwork,which realizes communication and information sharing within the network,and they can cooperate to complete complex functions.Traditional communication methods are difficult to use in the formation of nanonetworks due to issues such as device size and energy consumption.Inspired by nature,the use of biochemical molecules for information coding,transmission and reception of molecular communication is a better choice for building nanonetworks due to its advantages of good biocompatibility and high energy efficiency.The molecular communication via diffusion(MCv D)system has become a widely studied architecture at present due to its simple structure and no energy consumption.Source localization is a key parameter in the MCv D system.Determining the location of the source helps to optimize system parameters such as the number of released molecules,receiving time,and receiving threshold;it can also optimize the movement path of the nanomachine and adjust the layout of the nanomachines;It also plays an important role in determining environmental pollution sources and targeted drug delivery applications.Therefore,the research focus of this article is source localization in the MCv D system.The main work consists of the following aspects:1)For the source localization problem of the single-input multi-output diffusion molecular communication(SIMO-MCv D)system,this paper proposes a single source localization scheme based on distance estimation.Firstly,construct the expression of the cumulative receiving molecular number function for the multi-receiver scenario;secondly,use the Levenberg-Marquardt method to fit the cumulative receiving molecular number function to obtain the distance from the transmitter to each receiver;then,use multi-point positioning method to get a rough position of the source,and then take it as the initial value to optimize the localization result using the steepest descent method;finally,the simulation analysis of the localization effect of the method and the influence of different parameters on the localization performance are carried out.The simulation results show that the localization method has good accuracy in short and medium distances,and the localization effect can be improved by increasing the number of emitted molecules and the diffusion coefficient.In addition,the algorithm can also be used in advection environments to improve localization accuracy by controlling the direction of fluid movement in the environment.2)For the multi-source localization problem of the multi-input multi-output diffusion molecular communication(MIMO-MCv D)system,this paper designs a multi-source localization scheme based on fingerprint matching.First,a single source location scheme based on fingerprint matching is proposed,which is divided into two phases,offline and online phase.In the offline phase,the received signal of of the area is measured and recorded in terms of cumulative molecule numbers to establish a fingerprint database;in the online phase,calculate the summation of the normalized Euclidean distance between the signal and the fingerprint data,and find the fingerprint with the smallest sum of the normalized Euclidean distance so that the location of the source is roughly determined,and then use the cross search method and spatial interpolation to further optimize the location of the source.Then,based on the single-source localization scheme,a multi-source localization scheme was designed,combining the greedy algorithm and interior point method to initially determine the number and location of the sources.The multi-target cross search method was used to iteratively optimize the location of the sources.Finally,the simulation analysis of the localization effect of the method and the cause of the error are provided.The simulation results show that this scheme can estimate the coordinates of most sources in the area,and has excellent localization performance.For single source,almost100% correct-estimation rate are obtained;Regarding multiple sources,for most cases,the estimation errors are within 5 microns.