Research On Multi-target Localization In Molecular Communication Via Diffusion Systems

Molecular communication is a new communication method that uses tiny molecules to transmit information.It has the advantages of good biocompatibility and small size.Nanonetworks can be constructed based on molecular communication,which has important application prospects.Although the theory of molecular communication has been widely studied and related experiments have made progress,in fact,there are still many problems to be solved in molecular communication via diffusion systems(MCv D).In the MCv D systems,the release of a large number of molecules makes it an important problem to locate or track the signal source.The localization and tracking of signal sources in the MCv D systems is of great significance for the application of targeted drug delivery and environmental control.In this thesis,the localization of multiple signal sources in MCv D systems is studied,including the following two aspects.First,the multi-target localization method of molecular communication based on Sparse Bayesian Learnin(SBL)is studied.Considering that multiple signal sources release molecules in space at the same time,multiple receivers receive and count the cumulative number of received molecules synchronously,and estimate the positions of signal sources based on this,so as to be used for targeted drug delivery or source locations.Firstly,the off-line fingerprint database with the number of receiving molecules was established by deploying referenced points,and the MCv D multi-source localization problem model based on fingerprint database was constructed;Then,based on SBL method,sparse modeling is needed for off-line fingerprint database,sparse vector,noise interference and observation data;Finally,the sparse vector is obtained by SBL algorithm,and the corresponding position coordinates of the sparse vector are obtained.The simulation results show that the SBL algorithm can be used to locate the signal sources quickly,and the location effect is good.Second,the multi-target localization method of molecular communication based on Karhunen-Loeve(KL)transform is studied.In order to improve the low positioning accuracy in MCv D systems,firstly,KL transform method is used to reduce the correlation of data in fingerprint database;Then,on the basis of the KL transform on the number of receiving molecules,the positioning problem was modeled as QP(Quadratic Programming(QP)problem,and the approximate position of the signal source was obtained;Finally,by further refining the search near the approximate location,the final estimated position coordinates are obtained.Simulation results show that the proposed KL-QP algorithm can significantly improve the positioning accuracy.When the signal source is located on the grid points,the positioning accuracy of single source is close to 100%,the positioning accuracy of two sources is close to 90%,and the positioning accuracy of three sources is close to 40%.