Research On Fusion And Security Communication Of Heterogeneous Mobile Internet Of Things

With the continuous evolution of mobile communication technology,the fifth-generation mobile network(5G)has been officially commercialized.This brings wide coverage,large connectivity and low-latency network access services to the Internet of Things(IoT),and the era of Internet of Everything has arrived.In the face of heterogeneous network access technologies,mobile IoT data presents characteristics such as massiveness,heterogeneity and dynamics.The message service system needs to support the access and management of large-scale messages,and provide efficient and reliable signaling control capabilities and scheduling capabilities.In addition,the computing and storage capabilities of IoT edge devices are limited.Converged communication between heterogeneous mobile IoT systems has become more and more difficult,and information security issues have become more prominent.The traditional intrusion detection system(IDS)lacks dynamic learning and update capabilities,and the model training is expensive,unable to effectively detect heterogeneous and unknown anomalies,and has the problem of cold start.For the key issues of convergence and secure communication such as heterogeneous convergent networking,heterogeneous messaging services,heterogeneous data feature fusion and identification,and heterogeneous network security protection and upgrades of the mobile IoT.This article mainly carried out the following research work:Inspired by the immune technology of biological organ transplantation,the mechanism of using immune tolerance to induce anti-rejection reaction is proposed to solve the fusion problem between heterogeneous mobile communication systems.With transplantation immunity as the technical mechanism,a 5G Non-Standalone(NSA)networking fusion architecture based on the immune tolerance mechanism is constructed.The deep learning model is used to identify the type of donor protocol,and the donor signaling is decoded based on the "exclusive OR" operation and re-encoded based on the locus field.Simulation results show that this scheme can effectively identify donor signaling,improve the affinity of heterogeneous signaling and coding and decoding efficiency,and achieve the complementarity of immune tolerance mechanisms and algorithms.In order to improve the ability of dispatching and distributing messages in mobile communication networks,it is proposed to apply artificial immune theory to the message service of mobile communication systems.Using the simulated immune response mechanism,a semi-distributed immune dynamic adaptive network architecture is proposed to construct the detector dynamic learning mechanism and immune memory mechanism.The concept of an immune message distribution system is proposed,the message header is classified and cloned using the clone selection algorithm,and the message body is subjected to high-frequency mutation in combination with the positive selection algorithm.On the premise of ensuring the diversity of antibodies,the space consumption problem of the hash mapping algorithm is solved.The simulation results show that the ability of message recognition and message distribution has been improved.Aiming at the problem of limited computing resources of IoT security protection equipment and the difficulty of upgrading and updating,with 5G Narrowband IoT(NB-IoT)as the technical application background,a NB-IoT IDS architecture based on immune dynamic adaptive mechanism is proposed.Solve the problem of collaborative update of the abnormal feature database of each network element of the NB-IoT.Design an immune-based incremental data extraction method,and then propose a model weight update training method based on incremental data.In order to reduce the computing resources of edge devices,an IDS model based on a simple structure of multi-layer perceptrons,long and short-term memory and convolutional neural networks was constructed,and its static detection efficiency was verified.Evaluate the incremental learning performance of different models in multiple scenarios,and discuss the suitability of different models in the NB-IoT.The simulation results show that the proposed scheme can meet the needs of small data packets and large access volume of NB-IoT,and the training index changes more smoothly.Make up for the limitation of static models that cannot be adaptively updated,reduce the risk of data integrity being damaged,shorten the model update cycle,and save computing resources and storage resources.Faced with the challenges of data identification and fusion faced by heterogeneous mobile IoT intrusion detection,an IDS based on word embedding deep transfer learning is proposed.A simple domain alignment method is used to maintain the consistency of the source domain tensor and the target domain tensor to complete the sample transfer.Using the feature correlation between heterogeneous networks,word embedding is used to map the mathematical and logical features of the physical network to feature space vectors to complete feature transfer.Use different deep learning algorithms to complete model transfer.And verify the effectiveness of the proposed scheme in multiple heterogeneous data sets and multiple scenarios.The simulation results show that this solution can complete the feature extraction of the IDS neighborhood data of the heterogeneous IoT,save the data preprocessing time and training time of the heterogeneous IoT IDS model,and solve the cold start problem of the heterogeneous IoT IDS.