Maximum likelihood carrier frequency deviation estimation method of compression reconstruction

1. A compressed reconstructed maximum likelihood carrier frequency offset estimating method, used for frequency synchronization of the carrier of the communication system, characterized in that comprises:

A) on the basis of the received transmitting end of the training sequence, N × 1 structure x the received signal sequence, and is read from the designated storage position of the sensing matrix

B) according to the receiving signal sequence x, using the sensing matrix Compression ml frequency offset estimation metric is the compressed domain metric y_CS;

C) collecting the compressed domain metric y_CS reconstructs y ml frequency offset estimation metric set;

D) ml frequency offset estimation metric set according to the maximum value of index position y, injection ml frequency offset estimated value mapping

2. Method according to Claim 1, characterized in that the received signal from the received sequence is x for N+L the length of the sample in the training sequence, extracting the length of the training sequence N+L N of the sampling point, in order to prevent the data portion of the signal form the training sequence portion of a multi-path interference.

3. Method according to Claim 1, characterized in that step a) the sensing matrix Preconfigures, structure the steps of:

A1) configuration search length Z is:

Wherein symbol Said rounding operation the upward x, f_max to the maximum possible carrier frequency offset value, the Δf is set according to the estimated precision of the search step;

A2) to the measurement of the structure according to the type of the matrix Φ is the length of the observed quantity M;

A3) the measuring matrix M×Z structure:

A4) structure to a frequency offset of N×N Z matrix

According to Z= 1, 2, the [...] , Z, to calculate the frequency offset value estimated frequency offset

Attempts using the frequency offset value Diagonal matrix structure That is, the

A5) B projection matrix structure;

Using the known training sequence a= [a_-L+1, a_-L+2, …, a₀, a₁, …, a_N-1]^T, A the training matrix structure

According to the structure of the training matrix A, the projection matrix structure for B N×N B=A (A^H A)^-1 A^H;

A6) structure sensing matrix Sub-matrix M= 1, 2, the [...] , M, in other words:

Θm=[!phi!]m1Γ(v~1)BΓH(v~1)+[!phi!]m2Γ(v~2)BΓH(v~2)+...+[!phi!]mZΓ(v~Z)BΓH(v~Z)

A7) forming sensing matrix To:

Θ=Θ1Θ2...ΘM,The sensing matrix Stored in the designated memory space.

4. Method according to Claim 1, characterized in that step b) the compressed domain metric set y_CS x by the receiving signal sequence and perceive the matrix Sub-matrix (M= 1, 2, the [...] , M) structure, in other words:

yCS={xHΘ1x,xHΘ2x,...,xHΘMx}.

5. Method according to Claim 1, characterized in that step c) the ml frequency offset estimation metric set of reconstructed y, reconstruction algorithm is to utilize compression perception, according to domain metric set y_CS M elements of Z element reconstructed out of a frequency offset estimation metric set ml y= {y₁, y₂, …, y_Z}.

6. Method according to Claim 5, characterized in that the utilization of the compressed domain metric y_CS reconstructs y ml frequency offset estimation metric reconstruction algorithm comprises a set of: matching tracing algorithm, orthogonal matching pursuit algorithm, compressing and sampling a matching pursuit algorithm, based tracing algorithm, sub-space tracking algorithm.

7. Method according to Claim 1, characterized in that step d) of the estimated value of the frequency deviation ml Mapping is: set of frequency deviation estimation metric search ml y= {y₁, y₂, …, y_Z} z in the position of maximum, z∈ {1, 2, the [...] , Z}; then z according to the maximum value position, injection ml frequency offset estimated value mapping v^ML=v~z=-fmax+(z-1)×Δf,Thus the frequency offset estimated value of ml