Research On Secure Communication Technology Based On Voice Subliminal Channel

With the rapid development of communication technology,different communication technologies are being applied in daily life.Since the bell laboratory started voice communication which changed the way of communication,the rapid development of voice communication technology has drawn people's attention and pursuit of voice communication security.Sound subliminal channel is a kind of hidden channel,which can realize the transmission of important confidential sound signal under unknown circumstances.It ensures the security of transmitting of the sound signals to a certain some extent.However,because of the high concealment of the hidden channel,the acoustic communication may also be used as a means of attack,so there are some security risks in some situations.At present,related research work has pointed out that the existence of sound potential channels about the working characteristics of user devices such as headphones?keyboards and microphones,and the existence of sound potential channels about the natural characteristics has not been fully explored and studied.Combining with some existing theoretical knowledge,this thesis proposes a kind of secure communication technology of sound potential channel,which aims to realize the sound directional transmission and the sound subliminal transmission from space direction.The main tasks accomplished in this thesis are as follows:(1)The theoretical basis of the sound latent channel communication is analyzed,sound modulation,beamforming and the sound generation theory of the loudspeaker array are studied,and the corresponding modeling analysis is carried out.(2)Simulation analysis of the performance of acoustic submarine channel communication is carried out,combining beam assignment methods for three different cases: the conventional case,a given direction,and an undefined direction but with access to channel information,and using software tools to investigate the factors affecting the performance of submarine channel communication.(3)With the aid of readily available experimental equipment,the hardware experimental platform for the conventional case and the given direction of the acoustic subliminal channel is built to verify the feasibility of the proposed acoustic subliminal channel communication.(4)The possible risks of this voice latent channel in the field of intelligent voice assistants are considered,and a defense method based on acoustic spectrum features,time domain waveform features,and MFCC features is proposed.(5)The extraction method of MFCC features is improved based on HHT,and machine learning algorithms are used to experiment with the experimentally collected data,and experiments are carried out to prove its effectiveness based on the improved MFCC features.In summary,thesis takes sound communication technology as the background,modelling the communication process of the sound subliminal channel based on the relevant theory of communication and conducting software simulation experiments and hardware physical experiments to verify the feasibility of the sound subliminal channel model proposed in thesis.In addition,machine learning algorithms and improved method for MFCC feature extraction based on HHT are combined to propose defensive measures for detecting subliminal channel speech,the results show that the defence effect is better that after improving the features of the subliminal channel speech than before.Thesis has certain reference significance in the area of secure voice communication and speech signal feature processing.