Research On Interference Detection And Cancellation Algorithm Based On Machine Learning

The mobile communication users'transmission signals are high-power interference to IoT devices,so the IoT devices interference use cancellation solutions to solve the interfenrence.Traditional pilot based interference cancellation solutions require IoT devices to cooperate with mobile communication systems,which increases the complexity and communication loss of IoT devices.Based on the concurrent spectrum access model in the cognitive radio system,this dissertation designs a non-cooperative machine learning based iterative receiver for the IoT devices.This iterative receiver uses a clustering algorithm to estimate the interference,and can estimate the interference of the mobile communication system when the channel state information from the interference source to the receiver of the Internet of Things device is unknown,which reduces the complexity and energy consumption of IoT devices.This dissertation is divided into three chapters to discuss the iterative receiver design schemes in the single interference,multiple interference and asynchronous interference scenarios.The main contributions of this dissertation are:1.Aiming at the single interference cancellation problem in the concurrent spectrum access model,this dissertation designs a non-cooperative iterative receiver that can effectively recover the transmission signal under the strong power interference of the primary user for the secondary user.This iterative receiver use unsupervised machine learning algorithms to eliminate strong power interference generated by the primary user system and recover the transmission signal without cooperating with the primary user system.In addition,since the iterative receiver does not need to use pilots to estimate channel state information,the machine learning based iterative receiver can reduce the energy consumption and complexity of secondary user.2.This dissertation improves the Gaussian mixture model based expectation maximization clustering algorithm,it makes full use of the modulation mode information of the interference signal of the primary user,and proposes a modulation constraints based expectation maximization clustering algorithm.In addition,this dissertation also applies external information computing technology to the clustering module,and further proposes a external information and modulation constraints based expectation maximization algorithm.The simulation results show that when the external information and modulation constraints based expectation maximization algorithm is used to implement the clustering module,the iterative receiver only loses a little performance,but can greatly reduce the system complexity and communication overhead.3.Aiming at the problem of multi-interference elimination in the concurrent spectrum access model,this dissertation designs an iterative receiver based on the affinity propagation clustering algorithm to help secondary user eliminate the mixed interference of multiple interference sources and recover the transmission signal.Subsequently,this dissertation optimizes the affinity propagation algorithm by using the interference modulation mode information to constrain the number of groups obtained.Experimental simulation results show that the clustering module using the affinity propagation algorithm can help the iterative receiver to effectively recover the required signals under multi-user interference.In addition,the improved affinity propagation algorithm can bring significant performance gains to iterative receivers.4.Aiming at the problem of asynchronous interference cancellation in the concurrent spectrum access model,this dissertation proposes an iterative receiver based on the affinity propagation algorithm to help the secondary user eliminate asynchronous interference and recover the transmission signal.This dissertation uses the hidden Markov model to model the interference signal and proposes an hidden Markov model based affinity propagation clustering algorithm.Experimental simulation results show that the iterative receiver using the hidden Markov model based affinity propagation clustering algorithm can recover the transmission signal under asynchronous interference,and the hidden Markov model based affinity propagation algorithm can significantly improve the performance over the original affinity propagation algorithm.