Research On Signal Detection Algorithm Of GSM-MIMO System

With the rapid development of wireless communication network technology in the decades,the number of devices and users in the network has increased dramatically,and the mutual interference between user signals has also presented a more complex situation.In practice,due to the existence of various complex noise and interference in the physical layer communication system of mobile network,the signal received by the receiver is more distorted than that transmitted by the transmitter.For example,when the symbol interval between interfering users is multiple of the expected number of users,the interfering signals will show some dynamic correlation in the time domain.For orthogonal frequency division multiplexing(OFDM)systems,especially when synchronization and channel estimation are not perfect,short-term residual interference produces correlated dynamic interference in the frequency domain.In addition,in specific application scenarios such as satellite communication,the system is subject to more dynamic and complex electromagnetic interference,which is often difficult to analyze on its model.More importantly,in the modern information confrontation,more stringent requirements have been put forward for the interference/anti-interference capability of communication equipment.Large-scale multiple-input multiple-output(MIMO)technology has become one of the most important technologies supporting 5G communication because of its high transmission reliability and high spectral efficiency.However,the large-scale MIMO technology also has a huge disadvantage: the complexity of the MIMO system increases exponentially with the increase of the transmission antenna or the growth of the modulation constellation set,which is unacceptable in practice.To solve the problem that the system complexity in large-scale MIMO becomes too complex as the number of transmit antennas increases,we study the optimal signal detection problem in MIMO system using the generalized spatial modulation(GSM)technique,and propose an optimal decision tree search(ODTS)scheme.First,we need to build a GSM decision tree.When searching a GSM decision tree,an effective pruning strategy is needed to identify invalid signals.However,the existing optimal algorithms have exponential complexity.In order to solve this problem,we propose a memory-efficient pruning strategy by leveraging the combinatorial nature of the GSM signal structure.Under this strategy,the required memory size is squared to the number of transmit antennas.Combining the proposed pruning strategy of GSM decision tree with the best-first search algorithm,we further propose an efficient maximum likelihood search algorithm with finite memory(FM-MLS).The theoretical and simulation results show that the search algorithm proposed by us can obtain the optimal performance of signal detection bit error rate(BER)with limited memory space.In addition,with the increase of signal-to-noise ratio and system freedom,the expected time complexity of signal detection decreases exponentially,and tends to converge to the square of time in practical application scenarios.In addition,to solve the problem of signal estimation in related noise environment,we design a new receiver signal detection framework which combines the detection scheme mentioned above with convolution neural network(CNN).In the new signal detection framework designed by us,an effective CNN model is obtained by training the CNN,which can effectively capture the correlation characteristics between different symbols of noise,and thus obtain a more realistic noise output.The theoretical and simulation results show that the proposed detection scheme which combines the signal detector with CNN can effectively reduce the BER of receiver in related noise environment,and detecting signal repeatedly,the performance of BER can be further improved.