Research On Non-cooperative Signal Link Layer Analysis Technology

Non-partner link layer analysis is a reverse engineering problem in the field of communication countermeasures.That is,on the premise that the type of synchronization parameters and the type and parameters of the channel encoding are unknown,the inverse solution method is used to identify the synchronization parameters and the encoding algorithm and encoding parameters used for the channel encoding according to the received data stream.In a non-cooperative environment,the receiving end accurately and quickly recognizes the frame synchronization parameters,channel coding type,and parameters of the transmitting end as necessary conditions for information interpretation and analysis.This study of non-cooperative partner link layer analysis technology is of great significance.The purpose of this paper is to improve the performance of the system?including equal-length frame synchronization identification and synchronization parameter extraction,non-equal-length frame synchronization identification and synchronization parameter extraction,and encoding type identification and encoding parameter extraction?,including recognition accuracy,algorithm commonality,and time-consuming.The specific research is as follows:In order to improve the performance of non-equal-length frame data step recognition and parameter extraction module,this paper proposes a synchronization recognition and parameter extraction algorithm based on frame structure traversal.This algorithm first knows only the frame length range and the synchronization word length range.Through the traversal principle of at least one synchronization word between two adjacent maximum frames,a search window is set and traversed to obtain candidate synchronization words.Then,the candidate synchronization words are calculated by calculating the distance of the same candidate synchronization words in the intercepted data,and finally the frame is determined by decision Length and synchronization word length,to realize the synchronization identification and parameter extraction of non-equal-length frames.In terms of recognition accuracy,it is experimentally verified that the recognition accuracy of the algorithm is better than that of the classical synchronous recognition algorithm based on the storage matrix.When the transition probability is 1e^-3,the network recognition accuracy can reach100%;in terms of algorithm time,due to the algorithm Candidate synchronization words are only selected within two frames,and appropriate critical conditions are set when traversing after obtaining candidate synchronization words.Therefore,compared with the traditional blind synchronization algorithm for frame synchronization,the statistics based on frame structure traversal proposed in this paper The algorithm reduces the amount of calculation in the recognition process.In terms of algorithm commonality,the algorithm recognizes based on the inter-frame structure of the captured frames instead of the characteristics of the specific encoding method.Therefore,the algorithm has strong versatility and can be used for non-cooperation of unknown encoding structures.Non-equal-length frame synchronization identification and parameter extraction at the link layer.In order to improve the performance of isochronous frame data synchronization recognition and synchronization parameter extraction module while ensuring the robustness to noise,this paper proposes a synchronization recognition and parameter extraction algorithm based on deep neural network self-learning.On the premise of obtaining certain labeled training data,the algorithm realizes long-frame synchronization identification and synchronization parameter extraction through data preprocessing,network construction,and parameter adjustment.In terms of recognition accuracy,it is experimentally verified that the recognition accuracy of the algorithm is better than the classic memory matrix recognition algorithm.When the transition probability is 1e^-3,the network recognition accuracy can reach 92.9%,and the generalization performance of the neural network improves the network Robustness:in terms of algorithm time,although the offline training part has a large amount of training data and a long training time,after the training is completed,the test part can achieve the effect of approximate real-time testing,and the average time for a single test set is up to 120us;In terms of algorithm commonality,because the training features of the network are obtained through iterative learning,and are identified based on the training features obtained under data-driven rather than the characteristics of a specific encoding method,the algorithm is more versatile and can be used for unknown encoding Isochronous frame synchronization identification and parameter extraction of the non-cooperative link layer of the structure.In order to improve the performance of the coding type recognition and parameter extraction module,this paper proposes a neural network based coding type recognition and parameter extraction algorithm.This algorithm,after obtaining certain labeled training data,undergoes data preprocessing,network construction,and parameter adjustment.The realization step realizes coding type identification and parameter extraction.By constructing two networks,the first implementation of the network includes link-16,GPRS,3G,and Wi Fi coding standards.The channel coding part identifies the coding type,and the network 2 realizes the Ldpc coding parameter recognition under the Wi Fi system.This algorithm can more accurately realize the identification type encoding type and Ldpc encoding parameter identification.When the network 1 has a transition probability of 1e-3,the network recognition accuracy can reach 89%;when the network 2 has a transition probability of1e-3,the network The recognition accuracy can reach 95.72%;in terms of time consuming,the near real-time characteristics of the test time of the neural network are used,so the time is short,and the average time of a single test set is 160us;in terms of algorithm commonality,due to the characteristics of the training part The extraction is based on data-driven automatic encoding,and the network generation is independent of specific encoding types and parameters.Therefore,the algorithm is highly versatile and can be used for non-cooperative link layer encoding type identification and parameter extraction of unknown encoding structures.