Joint blind estimation of multiple frequency offsets and multiple channels is difficult in distributed Multiple Input Multiple Output Orthogonal Frequency Division Multiplexing (MIMO-OFDM) system under the multipath fading channel. In order to solve the problem, an effective algorithm was proposed. The proposed algorithm made use of blind deconvolution separation method to receive signal and got the multiple channels embedded with frequency offsets meanwhile. After estimating frequency offsets of the separated signals, the real channels estimation could be obtained by removing channel ambiguity and compensating the whole channels. The simulation results show that, the proposed algorithm is able to get 1e-6 average Mean Square Error (MSE) of frequency offsets estimation at 5 dB and 1e-2 average MSE of channels estimation at 15 dB compared with existing frequency offset channel estimation method based on pilot, the joint blind estimation of multiple frequency offsets and multiple channels for distributed MIMO-OFDM signal is realized

The traditional blind equalization with single receiver is significantly influenced by fading channel, and has high Bit Err Ratio (BER). In order to improve the BER performance, a Distributed Particle Filter (DPF) algorithm with low complexity for cooperative blind equalization was proposed in cooperative receiver networks. In the proposed algorithm, multiple receivers composed distributed network with no fusion center, estimated the transmitted sequences cooperatively by using the distributed particle filter. In order to reduce the complexity of particle sampling, the prior probability was employed as importance function. Then the minimum consensus algorithm was used to evaluate the approximation of the global likelihood function across the receiver network, therefore, all nodes achieved the same set of particles and weights. The theoretical analysis and simulation results show that the proposed algorithm does not centralize data at a fusion center and reduces the computational complexity. The fully distributed cooperative scheme achieves spatial diversity gain and improves the BER performance.

In order to improve the convergence rate of distributed consensus algorithms under directed communication topologies, a high-order distributed consensus algorithm was proposed. Under directed topologies, the previous state values of two-hop adjacency nodes were utilized to improve the convergence rate based on single-hop communication. The performance and convergence rate of the high-order distributed consensus algorithm were analyzed under directed networks. The simulation results were provided to verify these analytical results. The results show that an average consensus can be reached under certain conditions, the convergence rate of the high-order algorithm is superior to the other algorithms utilizing the information of two-hop adjacency nodes, but the high-order algorithm can tolerate smaller communication time-delays than the other algorithms utilizing the information of two-hop adjacency nodes.

The Physical-layer Network Coding (PNC) is difficult to be wiretapped. In order to wiretap and recover the signals from the PNC system, a blind signal processing algorithm based on decoding was proposed to recover the information sequences from the senders in the paper. Firstly, the algorithm analyzed the mapping signal from the second slot in PNC scheme in order to get the parameters of modulation and encoding. Secondly, the algorithm processed the mixed signal from the first slot in the PNC scheme, and clustered the sampling data to judge the component of bit information. At last, the paper used the decoding algorithm to correct the information and took the mixed signal apart successfully. Extensive simulation studies show that the new algorithm based on constellation analysis and decoding can recover the signals blindly, and the Bit Error Rate (BER) reaches 10-3 when the Signal-to-Noise Ratio (SNR) is 11 dB.

For the modulation recognition of seven kinds of Quadrature Amplitude Modulation (QAM) in non-cooperative communication, a new blind identification algorithm was proposed based on combined features. Based on the discussion and analysis of the cyclostationarity and the instantaneous amplitude distribution of the QAM signals, the algorithm used the combined features which were cyclostationary detection feature, fourth-order zero-conjugate cyclic accumulation feature and instantaneous envelope feature. The algorithm used the binary tree support vector machine as classifier to classify the seven Intermediate Frequency (IF) QAM signals. The simulation results show that the correct recognition rate of the algorithm reaches over 90% when the number of symbols is 1000 and the Signal-to-Noise Ratio (SNR) is more than 6dB.