Research On IoT Security Based On Backscatter Propagation Signatures

Internet of Things(IoT)is gaining widespread attention in many fields,while security issues are becoming increasingly serious.In response to the threats of impersonating legitimate identities,forging attack commands,and tampering with abnormal data,security mechanisms based on cryptographic computing,feature extraction,and sensor assistance face the challenges of limited computing resources,insufficient hardware configuration,and weak environmental adaptability.With the wide application of backscatter communication in IoT and its reliance on wireless links for symbiotic transmission,its unique physical layer characteristics provide new opportunities for security design of IoT.To this end,this paper deeply explores the diffraction signatures,spatial directivity,and nonlinear effects of backscatter,and constructs a low-complexity security authentication mechanism,a lightweight attack detection mechanism,and a non-intrusive anomaly monitoring mechanism,which effectively guarantee the authenticity of the transmitter,the validity of the transmitted commands,and the integrity of the received data.The main contributions are concluded as follows.First,due to limited computational resources,it is difficult to achieve wearable tag authentication by complex encryption,and we design a low-complexity security authentication mechanism based on the diffraction signature.Specifically,we first make an in-depth exploration on the differences between the diffraction signature of wearable tags on the body surface and the propagation mode of off-body devices in theory.At the same time,combining the effects of body fluctuation and device movement on electromagnetic waves,we propose a gesture movement-based authentication scheme to strengthen the correlation between diffraction and body movement,and realize the low-complexity physical layer security.Further,based on the matching relationship between movement patterns and electromagnetic features,we design scale change-based and variance fluctuation-based mechanisms to eliminate interference,effectively improving the authentication performance of the system.Second,due to the insufficient hardware configuration,it is difficult to extract finegrained features to detect the forgery command with antenna arrays,and we design a lightweight attack detection mechanism based on the directivity of backscattering array.Specifically,we first attach backscatter tags around the gateway to construct a distributed array,from which the spatial directivity of reflected multipath can be used to determine the uniqueness of the device.To capture reliable propagation signatures,we design a feature spatial description scheme for multi-dimensional feature extraction,and develop a feature segmentation and modification approach to distinguish the differences between the legitimate and illegitimate features.In addition,Besides,to defend against a powerful attacker that can estimate the channel state,we also provide a tag-random and a multi-AP coordination schemes to enhance the system attack detection capability.Finally,to address the challenge that monitoring system faces data tampering,and the sensor-assisted approaches require intrusive modification of the robot,and have weak environmental adaptability,we design a non-intrusive anomaly monitoring mechanism based on the nonlinear effect.Specifically,we first attach backscatter tags on the robot’s arms and characterize the features of trajectory by leveraging the effect of the motions on backscatter signals.Further,the motion signatures in the signals would be fused and captured by the nonlinear effect as well as the corresponding transceiver signal design.Meanwhile,we also develop a multi-stage feature extraction scheme to capture the time-frequency features based on the robot operation differences at each stage.We further provide a deep neural network to filter and reconstruct the features to eliminate the interference caused by the robot reflections.In this paper,we design a low-complexity authentication mechanism,a lightweight attack detection mechanism,and a non-intrusive anomaly monitoring mechanism for the security challenges faced by the IoT system,and validate the effectiveness of the schemes,which provides strong technical support for IoT security.