Research On Channel Model And Signal Detection Techniques For Diffusion Based Molecular Communications

One of the ultimate goals of human scientific research is exploring the mystery of life and understanding the operational mechanism of life activities.The study of biological phenomena is helpful to understand and reveal the underlying mechanism,and is of great significance to the regulation of biological processes,disease prevention and precision medicine.In recent years,with the rapid development of nano-technology and synthetic biology,the understanding of biological systems has become increasingly profound.The mode of information exchanging between organisms in a biological system is different from that of the traditional wireless communication based on electromagnetic wave,yet it has withstood the test of natural selection.Therefore,inspired by nature,molecular communication as a communication mode with great biological adaptability comes into being,among which the diffusion based molecular communication mechanism is the most concerned.Diffusion based molecular communication utilizes messenger molecules as information carrier and information exchanging is realized through fluid medium channel.This paper focuses on the channel model and signal detection techniques in diffusion based molecular communication and begins with the biological channel in which the operation mechanism of molecular communication in organism is deeply studied.On this basis,the research extends to the field of nano-machine and the signal detection techniques are investigated according to the characteristics of the environment of molecular channel and molecular signal.The main contributions of this paper are as follows:Firstly,this paper studies the biological channel model for diffusion based molecular communications.An end-to-end biological channel model is theoretically established and the expression of channel response is derived.The channel model focuses on the degradation reaction of messenger molecules occurred during the transmission and the reversible reaction between messenger molecules and receptors at the receiver.The molecular signal is decomposed into effective signal,inter-symbol-interference and noise to be quantified,and the probability distribution of each component is derived.Subsequently the modulation techniques based on molecular concentrations and molecular types are introduced and the detection probabilities of signal decision along with channel capacity are derived.The simulation algorithm based on particle random movements is utilized to verify the accuracy of the channel response,and the channel capacity under different chemical reaction rates is contrasted.Secondly,to solve the difficulty of obtaining the molecular channel response and the inapplicability of conventional coherent signal detection algorithms with high complexity in molecular communication,a non-coherent signal detection scheme based on feature extraction with low complexity for molecular communication is proposed.The common characteristics of molecular channel response and molecular signal wave are qualitatively analyzed.Three molecular signal features including local wave crest of a single bit,valley of successive bits and energy difference of successive bits,are extracted.Then the features extracted above are quantified as corresponding measurements and a non-coherent signal detection algorithm based on equal ratio combination of measurements is proposed.With the consideration of the discrepancy between measurements,a weight assignment scheme for measurements is designed and then a non-coherent signal detection algorithm based on optimal weight combination of measurements is proposed.The simulation results show that the proposed scheme could achieve reliable transmission of molecular signals without channel information.Thirdly,to further improve the performance of signal detection,a non-coherent signal detection scheme based on machine learning is proposed.This scheme transforms the signal detection problem into a classification problem on the basis of feature extraction,and an unsupervised clustering algorithm from machine learning is combined to realize signal decision.The simulation results show that this scheme could reduce the BER significantly compared with the schemes based on linear combination of measurements,and the computational complexity along with data storage consumed are not increased enormously.Furthermore,to reduce the leakage and the control pressure of release accuracy of messenger molecules,a reactive release model is utilized.The simulation results show that this release model can effectively approach the optimal instantaneous release model when the reaction rate is large.Finally,to solve the problem that the communication performance declines when the nano-transceivers are relatively distant,a multi-relay assisted molecular signal transmission scheme is proposed.Firstly,the influence of receiver models of relay nodes and types of messenger molecules released in different transmission stage on transmission performance is discussed.Then,four relay modes according to the forwarding modes and transmission mechanisms are proposed.The applicability of non-coherent signal detection algorithms based on feature extraction is qualitatively evaluated and proved mathematically,and the theoretical BER under each relay mode is derived.The simulation results show that the introduction of relays could effectively improve the reliability of molecular signal transmission with long distance and the actual performance is consistent with the theoretical BER.Moreover,the relay mode should be selected according to the specific noise condition to obtain the optimal transmission performance.