Automatic Modulation Classification For OFDM With Index Modulation

Orthogonal frequency division multiplexing with index modulation(IM)is emerging as a novel multi-carrier modulation technology.Higher bit error rate performance and spectral efficiency are achieved by transmitting additional information bits through index of idle subcarriers.As an efficient communication method,it is important to study automatic modulation classification(AMC)in non-cooperative communication environment.In OFDM-IM,the parameters to be identified include not only the signal constellation but also the number of activated subcarriers,which is different from the modulation classification of conventional OFDM that only needs to identify the modulation scheme.Therefore,AMC of OFDM-IM is a challenging problem.To solve this problem,the classification scheme based on the likelihood probability method is first studied in this paper.In the scenario with known channel state information(CSI),an optimal average likelihood ratio test(ALRT)classifier and a hybrid likelihood ratio test(HLRT)classifier are derived.From a Bayesian perspective,ALRT achieves the largest probability of correct classification(PCC)and is therefore optimal,while HLRT is an implementation that reduces complexity.Furthermore,in the HLRT classifier,it is proposed to identify the activated subcarrier by the energy detector and the log likelihood ratio(LLR)detector,which eliminates the unknown activated subcarrier set variable,thereby the complexity is reduced.Concretely,the energy detector estimates the activated subcarrier set utilizing the characteristics of the transmitted symbols of the OFDM-IM system,and the LLR detector estimates the activated subcarrier set according to the posterior probability ratio.Then,in the scenario with unknown CSI,a blind AMC based on HLRT is proposed to estimate the channel fading coefficient and the noise variance using the EM algorithm.The simulation results prove the effectiveness of above scheme,and 100% PCC can be achieved as the signal-to-noise ratio increasing.Finally,the identification results of OFDM-IM system and OFDM system are compared.In order to reduce the complexity of the classifier,a feature-based classification scheme is then studied in this paper.Firstly,the fourth-order cumulants of different combined modulation parameters are calculated,and an identification method based on the fourthorder cumulants is proposed.The PCC performance of known channel fading coefficients and unknown channel fading coefficients is compared,and the simulation proves the feasibility of the scheme.In order to further improve PCC,the identification scheme based on higher-order moment is studied.The scheme extracts the higher-order moments as the features from the received signals.Since the extracted moments obey the normal distribution,the likelihood function of these moments are obtained under different assumptions,thereby obtaining the modulation parameters to be identified.Through the analysis of complexity and simulation results,it can be concluded that the identification scheme based on higherorder moments has both lower complexity and higher classification performance,so it is an effective identification method.