36 results about "Channel complex" patented technology

A method of extracting data from a received signal including multi-path interference in a rake receiver. The method includes sampling and filtering the received signal; estimating a time delay τl between paths for the filtered samples ψ(τ); and estimating channel complex coefficient cl for the filtered samples ψ(τ). Transmitted data x(τl) is extracted from the filtered samples ψ(τ) for each path l by solutions of simultaneous equations of the following filtered samples ψ(τ) equationψ⁡(τ)⁢Rff-1⁡(τk-τ^0)⁢ΛssH⁡(τ^k)=∑l=0Np+1⁢⁢ci⁡(τl)×(τl)⁢Λss⁡(τl)⁢Rff⁡(τl-τ^0)⁢Rff-1⁡(τk-τ^0)⁢ΛssH⁡(τ^k)+⁢(τ)wherein k is a particular path, Np is the number of visible paths, Rƒƒ(τl–τo) is a double convolution matrix of the filtering process and Rƒƒ−1(τk−–o) is the pseudo inverse, Λss(τl) is the product of spreading and scrambling matrices and ΛssH(τk) is the inverse, and (τ) is noise.

The invention discloses a binaural speech enhancement method based on deep learning, comprising the following steps: respectively processing a left / right channel noisy speech signal containing a target speech signal to be enhanced so as to obtain a left / right frequency domain signal, and combining amplitudes thereof to obtain single-channel complex features; using the frequency domain signal of the left / right channel and corresponding target frequency domain signal theoretical value to calculate corresponding target speech ideal complex masking respectively; combining to form a target speech single-channel complex masking theoretical value, combining the single-channel complex features to train a complex feedforward neural network so as to obtain a binaural speech enhancement model; usingthe target speech single-channel complex masking estimated value outputted by the model to respectively process the left / right channel noisy speech signal so as to obtain a left / right channel frequency domain signal; and finally obtaining a corresponding target speech time-domain signal. By the method, noise interference can be suppressed and spatial information of a target sound source can be maintained. By making full use of the generalization ability of deep neural networks, the enhancement of binaural speech is achieved.

The invention discloses a channel estimation method based on a complex exponential-basis expansion model. The method comprises the following steps of: (1) estimating gradient of channel impulse response in intervals according to the data processed by a wireless receiver; (2) based on the CE-BEM (Complex Exponential-Basis Expansion Model), calculating baseline compensation quantity of an observed value, and a new signal model; (3) estimating the channel impulse response after compensation; and (4) calculating the channel impulse response before compensation. The invention further discloses a channel estimator. With the method and the estimator provided by the invention, the error of the CE-BEM is reduced, the accuracy of channel estimation is improved, and the implementation complexity is low.

A WCDMA receiver in which the dependencies of different functional modules are arranged to allow proper tailoring of the Channel Estimation (CE) module bandwidth. A PreCE provides rough estimation results of the channel complex gain. The rough estimation results are passed to the Automatic Frequency Control (AFC). The AFC outputs a signal transmitted to two mixers. The first mixer mixes the output of the PreCe with the output of the AFC and outputs the result to a Velocity estimator (VE) and a PostCE to generate the compensating signals. The output of the VE is also transmitted to the PostCE. The output of the PostCE is sent to a Maximum Ratio Combining module, whose output is mixed by the second mixer with the output of the AFC to generate a final signal.

The invention provides a method for discovering dirty data propagation paths based on complex networks, which can translate binary programs without source codes, process the obtained results, and mine and generate useful information. Step 1: Decompile the binary file and get the intermediate code of C language. After testing a simple C language program, get the intermediate language code; Step 2: Capture the function call path and resolve the function address into a function Name, process and simplify, and generate a matrix format; finally generate a function call relationship graph; the third step: analyze the function call relationship graph, obtain node, edge, weight information, calculate node degree, and establish a complex network with key nodes Figure; the fourth step: According to the non-uniform characteristics of the power law distribution of the complex network graph, find out the points related to the construction of dirty data and the points with high calling frequency.

The present invention provides a scrambling code cancellation method aimed at combining channel codes and scrambling codes, comprising the following steps: performing frame division processing on demodulated data; counting the number of frames under each frame length; Two frames are extracted from the set with the number of frames greater than 1, and the bearer data of the two frames are extracted and put into the sequence S1 and the sequence S2 respectively; XOR the sequence S1 and the sequence S2 to obtain the sequence with the scrambling code removed; based on removing The sequence of the scrambling code is identified by the channel code identification method. The invention utilizes the properties of the same frame scrambling position and the same scrambling sequence to cancel the scrambling code, and proposes a scrambling code canceling method in which the channel code and the scrambling code are combined. Then, according to the property that the codewords obtained by XOR between codewords of the channel codes still belong to the codewords in the original codeword space, the existing channel code recognition method can be used to identify the channel code parameters.

The embodiment of the invention discloses a method and a device for channelizing a real signal. The method comprises: after receiving a certain multichannel real signal, carrying out shunt delay on the multichannel real signal to obtain 2D channel data; respectively carrying out 2D fold extraction on the 2D channel data, and alternately changing the symbol of each extracted channel data; filtering the 2D channel data processed by symbol change; synthesizing the filtered 2D channel data into D channel complex signal, and carrying out multi-index modulation on each channel complex signal; and carrying out D point FFT on the modulated each channel complex signal, and outputting a D channel baseband signal. The invention carries out extraction treatment 2 folds higher than the prior art on a multiphase treatment front end, thus reducing the complexity of post signal treatment; and the invention carries out optimization on the multi-index modulation part, thus lowering the complexity of the channelized structure of the existing filter set and solving the problem of over high consumption of system resource.

The invention discloses a speech enhancement method, device and electronic equipment based on spatial features. The method includes: performing Fourier transform on a dual-channel noisy speech to obtain a dual-channel complex spectrum; obtaining the first dual-channel complex spectrum based on beamforming A single-channel complex spectrum and a second single-channel complex spectrum; calculating the logarithmic power spectrum of the first single-channel complex spectrum; calculating a direction energy ratio based on the energy of the first single-channel complex spectrum and the energy of the second single-channel complex spectrum, and Take the logarithm to obtain the energy ratio in the logarithmic direction; use the logarithmic power spectrum and the energy ratio in the logarithmic direction as feature inputs to the pre-trained speech enhancement neural network to obtain the masking value; add the masking value to the first single-channel complex spectrum, and The first single-channel complex spectrum after the masking process is inversely Fourier-transformed to obtain the enhanced speech. The solution provided by the embodiment of the present invention can effectively reduce noise and better avoid voice distortion.