JP6982288B2 - Signal processing system, reception method and program - Google Patents

Description

The present invention relates to a signal processing system, a receiving method and a program, and in particular, a signal in which the receiving unit compensates for the phase distortion caused by the two-dimensional phase modulation performed by the transmitting unit in addition to the phase distortion caused by the propagation path. Regarding processing systems, etc.

As described in Patent Document 1, the inventors have studied and developed two-dimensional phase modulation using a time diffusion code and a frequency diffusion code for continuous time.

Patent Document 2 describes a frequency diffusion type radar device that detects an object by using a spectrum-spreading detection radio wave.

Further, in technical fields such as wireless communication, non-overlapping multiplexing (Multiple Access) (Multiple Access) is performed by evenly dividing the time domain (TD) and frequency domain (FD). FDMA, TDMA) is being performed. For example, CDMA (Code Division Multiple Access) uses a spread spectrum code (SS code). Also, OFDM (Orthogonal Frequency Division Multiple) emphasizes the orthogonal relationship of pulse waveforms. CDMA has been widely used in 3rd generation mobile communication systems. After that, OFDM was adopted and it is called the 4th generation mobile communication system.

Japanese Unexamined Patent Publication No. 2016-189500 Japanese Patent No. 5291267

In the technical field of wireless communication, discussions are generally made on the premise of OFDM and the like with an emphasis on strict orthogonality and the like. However, although it is known that phase distortion related to the propagation path (delay time and Doppler shift) occurs in the receiving device, it is hardly known that other phase distortions occur. Therefore, the phase distortion due to the propagation path is compensated, but the phase distortion due to other factors is not compensated. OFDM is currently widely used, but if higher performance is required in the future, orthogonality will be disrupted due to time fluctuations and frequency fluctuations, and synchronization will become difficult. Therefore, it is expected that it will be difficult to realize wireless communication with overwhelmingly high performance even if OFDM and the like are simply expanded.

Further, Patent Document 2 is generally called a (one-dimensional) "compression radar". The band is widened to increase the time resolution, and the delay time estimation accuracy is improved, but it is vulnerable to Doppler shift.

In this way, even if we try to use the time diffusion code or frequency diffusion code to improve the performance of wireless communication and radar equipment, it seems vague that there is a limit to the performance improvement. However, it was difficult to grasp the problem accurately, let alone find a clue to solve the problem.

Therefore, the present invention realizes wireless communication with better performance than simply performing compensation by phase distortion due to the propagation path when "two-dimensional compression processing" is realized by using a time diffusion code and a frequency diffusion code. It is an object of the present invention to provide a signal processing system or the like capable of the above.

The first aspect of the present invention is a signal processing system using two-dimensional phase modulation, which includes a transmitting unit and a receiving unit, and the transmitting unit uses a time diffusion code and a frequency diffusion code to provide a two-dimensional phase. A transmission signal generation means for generating a transmission signal by modulation and a transmission signal transmission means for transmitting the transmission signal are provided, and the receiving unit includes a reception signal receiving means for receiving the transmission signal and obtaining a reception signal, and the reception signal receiving means. A compensating means for compensating for phase distortion in a received signal is provided, and the compensating means compensates for phase distortion due to the two-dimensional phase modulation using the time diffusion code and the frequency diffusion code.

The second aspect of the present invention is the signal processing system of the first aspect, and in the transmission unit, the transmission signal generation means performs data multiplexing and / or chip multiplexing by the two-dimensional phase modulation. In the receiving unit, the compensating means compensates for the phase distortion due to the data multiplexing and / or the chip multiplexing.

The third aspect of the present invention is a signal processing system using two-dimensional phase modulation, which includes a transmitting unit and a receiving unit, and the transmitting unit has a time shift and a frequency that satisfy symmetry between time and frequency. The transmission signal generation means for generating a transmission signal by preliminarily offsetting using a shift operator for shifting and performing two-dimensional phase modulation using a time diffusion code and a frequency diffusion code, and the transmission to the reception unit. The receiving unit includes a transmitting signal transmitting means for transmitting a signal, and the receiving unit compensates for a receiving signal receiving means for receiving the transmitted signal through a propagation path to obtain a received signal, and compensation for phase distortion due to the two-dimensional phase modulation. comprises an estimation means for estimating a delay time t _d and the Doppler shift f _D in the propagation path, the estimating means compensates the phase distortion due to the two-dimensional phase modulation to obtain the maximum likelihood estimate of the Doppler shift f _D TD and estimating means, the two-dimensional phase modulation compensates the phase distortion by comprising a FD estimating means for obtaining the maximum likelihood estimate of the delay time t _d, the estimating means, by the FD estimating means and the TD estimating means, by updating alternately whereas the maximum likelihood estimates on with the other, is to estimate the delay time t _d and the Doppler shift f _D.

A fourth aspect of the present invention is a receiving method for receiving a transmission signal generated by two-dimensional phase modulation using a time diffusion code and a frequency diffusion code, wherein the compensating means receives the transmission signal. The obtained received signal includes a compensation step for compensating for phase distortion due to the two-dimensional phase modulation using the time diffusion code and the frequency diffusion code.

A fifth aspect of the present invention is obtained by receiving a computer in a receiving device that receives a transmission signal generated by two-dimensional phase modulation using a time diffusion code and a frequency diffusion code. This is a program for functioning as a compensating means for compensating for phase distortion due to the two-dimensional phase modulation using the time diffusion code and the frequency diffusion code in a received signal.

The invention of the present application may be regarded as a computer-readable recording medium that constantly stores the program of the fifth aspect.

If a diffusion code is introduced, multiplexing and the like can be easily realized. However, the inventors have the disadvantage that the continuous time result is changed to the discrete time type, and phase distortion occurs with the introduction of the diffusion code, and this is due to the group operation of the noncommutative operator. It was found that accurate compensation is required in the receiving device because it propagates to the receiving device.

Conventionally, in particular in wireless communication technology, phase distortion due to a propagation path is known, but phase distortion caused by other than that is hardly known. Further, conventionally, a part of such phase distortion appears as "multiplication of a power of -1", and by taking an absolute value, it becomes "multiplication of 1", which does not affect the processing. Therefore, this phase distortion is caused. It is probable that it was not compensated.

In the future, in order to improve the performance, it will be necessary to pay attention to the phase distortion other than the phase distortion caused by the propagation path. However, in the past, since there was no need for compensation, neither the existence of the phase distortion nor the need for compensation was recognized, and suddenly, the constraint of performance improvement of unknown cause was realized, causing confusion.

According to each aspect of the present invention, by compensating for the phase distortion caused by the two-dimensional phase modulation using the diffusion code in the receiving unit, for example, in the field of wireless communication, more precise synchronization can be realized, or radar technology. In the field, it is possible to realize high-performance signal processing such as irradiating an object with a transmission signal to obtain a received signal from the object and estimating a position with high accuracy. In particular, as in the second aspect of the present invention, it is possible to compensate for the phase distortion due to data multiplexing and / or chip multiplexing by the diffusion code, and realize high-performance signal processing. Historically, this is a situation similar to the one in which the limits of Newtonian mechanics were elucidated by quantum mechanics and further developed.

Further, according to the third aspect of the present invention, even when synchronization is difficult by a simple extension such as OFDM, synchronization can be realized by appropriately compensating for phase distortion. Further, synchronization can be easily realized by using the symmetry between time and frequency so that the receiving device alternately updates and realizes, for example, by using a filter bank. For example, it is possible to estimate unknown parameters (delay time and Doppler shift) of the propagation path and realize wireless communication having a receiving unit having both a synchronization function and a decoding function, in which data decoding is realized at the estimation end stage. can.

The invention of the present application is unique in that it has found a new method of utilizing the two-dimensional diffusion code. That is, we have found 1) an information embedding method that utilizes the generation of phase distortion due to multiplexing, and 2) a parameter estimation method that compensates for the phase distortion of the embedded various signals. _{This made it possible to estimate t d} and f _D without the need for other information (ie, without information) (allowing incoherent communication).

The problem of estimating the delay time of the radar problem and the unknown parameter of the Doppler shift is reduced to the identification of where it exists in the two-dimensional parameter space. For example, 4 × 4 time width T and bandwidth F data in FIG. 1 (c) are subdivided into 4 × 4 chip time width T _c = T / N and chip bandwidth F _c = F / N ′. It is to specify the two-dimensional address of the small rectangular area of the chip level of. First, a two-dimensional phase-modulated chip waveform with a two-dimensional address is assigned to each rectangular region, and these are linearly coupled with a two-dimensional diffusion code without superposition, and a broadband waveform (called a signature) is used as a radar emission signal. .. Further, for example, by using a Gaussian pulse, it is possible to perform operations separately.

The present invention has focused on the fact that the phase distortion due to data multiplexing and / or chip multiplexing by the diffusion code is address-specific. This is quite different from the well-known role of diffusion codes.

Also, the discussion is usually limited to the time domain (TD). However, the frequency domain (FD) is also treated symmetrically so that the time-frequency symmetrical (TFS) is satisfied, and the discussion of TD is extended to FD as a natural consequence of TFS and discussed equally. In the present invention, TD and FD are discussed in parallel in the equation development and FIGS. 2 to 5. In this sense, variable functions of time and frequency appear in the form of twin in both functions and devices, and the filter bank of the implementation example is a twin filter bank.

Further, although it seems that the specification of the two-dimensional address of n rows and n'columns usually requires the labor of ordering NN', the signature waveform design method according to the present invention reduces the labor of ordering N + N'. ing. The linear combination (non-superimposition sum) of the TD template waveform for row identification by the diffusion code is the TD signature of equation (23) (its frequency dual: the linear combination of the FD template waveform for column identification (non-superimposition sum)). It is an FD signature of Eq. (25)), and both are in a Fourier transform relationship with each other. Pattern matching for N and N'row and column identification (specifically, TD for Doppler shift estimation and FD for delay time estimation) is performed for each of TD and FD, and parameter estimation value information is obtained between the two. By exchanging, it is possible to estimate with high efficiency and high accuracy.

The pattern matching by the proposed method correlator (TD ambiguity function type cross-correlation, equation (50) (its frequency dual: FD cross-correlation, equation (51)) is N in the parameter estimation theory of statistics. It is the same as the matched filter type likelihood function for seed signal detection. In the present invention, since pattern matching is also performed in FD, a matched filter type likelihood function for N'type signal detection in FD is given. The split-correlation method, which reduces two unknown problems to two one-unknown variable problems, is von Neumann's cross-correlation theorem (a parameter estimate that gives the maximum value of the most-likelihood of one of the likelihood functions). Is updated to the parameter estimation value of the other likelihood function), and the parameter estimation accuracy is improved with high efficiency.

In the receiving unit, parameters can be estimated by designing a correlator that compensates for the phase distortion generated by non-superimposition coupling. Increasing the diffusion code length improves the parameter estimation accuracy, but increases the required frequency bandwidth and lengthens the observation time. This is a trade-off relationship with the parameter estimation accuracy.

The present invention is applicable to a wireless communication synchronization method in which it is necessary to constantly grasp the location and moving speed of a mobile phone user. Further, for example, the radar device can be realized by the transmitting device irradiating the measurement target with a signal and the receiving signal receiving the signal from the measurement target. At this time, the transmitting unit and the receiving unit may be realized by one device.

(A) TD, (b) FD, (c) Gabor division multiplexing, and (d) symmetric time-frequency shift operator Τ _{τ, ν} x (t) are shown. The SFB that generates the TD signature v [k] and the FD signature V [l] using the SS code of TD and FD is shown. The SFB that generates (a) TD transmission signal and (b) FD transmission signal for searching for _{(t d} , f _D ) in which PP'information data is embedded or can exist in a wide range of PP'locations is shown. .. An AFB consisting of (a) TD correlator array and (b) FD correlator array is shown as a receiver for decoding PP'information data or _{measuring distance at (t d} , f _{D) at PP'.} Information exchange and parameter update procedure for alternate estimation of parameters based on (a) TD, (c) FD correlator array AFB and (b) Neumann APT at a specific location ^{→ p = (p, p'), (} d) The state of Neumann's alternating update projection theorem is shown. An example of the configuration and operation of the signal processing system according to the embodiment of the present invention is shown. Another example of the configuration and operation of the signal processing system according to the embodiment of the present invention is shown.

Hereinafter, examples of the present invention will be described with reference to the drawings. The embodiment of the present invention is not limited to the following examples.

First, the theoretical background of the present invention will be described. FIG. 1 shows three types of multiplexing: (a) time domain (TD), (b) frequency domain (FD), and (c) Gabor division (TD-FD). FIG. 1 (d) shows a symmetric time-frequency shift operator Τ _{τ, ν} x (d) with non-commutable characteristics Τ _{τ, 0} Τ _{0, ν} (t) = e ^{-j π} τ ν Τ _{o, ν Τ τ, 0.} t) is shown.

For example, in the radar problem, the delay time and Doppler shift are unknown, and accurate distance measurement is required. _{By introducing TD and FD diffusion codes and introducing microcells of T c} = T / N and F _c = F / N'in chip units in which T and F are evenly divided into N and N', the delay time and The Doppler shift can be determined with the accuracy of the chip units T _c and F _c. Therefore, the radar emission signals are wideband signals v (t) and V (f).

It seems that synchronization simply requires time and effort NN'. However, for easy synchronization, N'and N template signals as clues are embedded in the TD and FD signals v (t) and V (f), respectively. Therefore, the labor is reduced to N + N'.

N, N'-TD, FD correlator arrays perform pattern matching between the received signal and the template. Each correlator is a parameter subspace obtained _{by direct sumizing the parameter spaces Θ τ and ν} , and performs template matching with the templates assigned to the respective N and N'subspaces. That is, in order to characterize the address by evenly dividing the parameter space into rectangular areas specified by the address (nth row (1 ≦ n ≦ N), n'th column (1 ≦ n'≦ N')), n Design a template for the row (its frequency dual: template for the n'column). Furthermore, template matching is performed between these templates and the received signal (its frequency dual: its Fourier transform). At that time, a diffusion code is introduced to ensure independence between the template signals. This is because the likelihood function of the N-type signal detection theory (resp.N'type signal detection theory) in the parameter estimation theory of statistics is realized by the correlator of the receiving device. The present invention employs a two-dimensional phase-modulated signal for the parameter estimation problem, and provides a new method of utilizing the conventional diffusion code.

The introduction of the diffusion code has advantages such as multiplexing. The biggest drawback is that phase distortion due to the code occurs, and phase distortion compensation is required for this. That is, _{phase distortion exp (jπTF) and exp (± jπT c} F _c ) are generated due to the multiplexing (data level, chip level) that follows the phase distortion exp (jπt _d f _{D ) of t d} and f _D. On the receiving side, compensation for phase distortion is required. In the field of wireless communication, such phase distortion compensation is often ignored.

Hereinafter, a specific description will be given.

The inventor has conducted research and presentations on continuous time (see Patent Document 1). The continuous-time variable and the discrete-time variable are associated with each other by Eq. (1), and the continuous-time result is changed to the discrete-time type. Further, Eq. (2) is for the orthogonal relationship of the chip pulses. Equation (3) is an L-point twiddle factor. In the following, in order to confirm the symmetry between TD and FD, both will be discussed in parallel.

Equation (4) and (5), respectively, the delay time t _d of the waveform z (t), the echo signal and its Fourier transform of the Doppler shift f _D (FT). There is no symmetry with respect to (t _d , f _{D) between them.}

With a little ingenuity, Eq. (6) can be obtained. _{The functions in parentheses of the Fourier transform of the first equation (6) and the third equation are (t d} , f _D ) for the TD function z (t) and the FD function Z (f) of the equation (7), respectively. ) Is given the definition of the symmetric time / frequency shift calculation formula, and the formula (8) is given (see FIG. 1 (d)).

For the time being, for the sake of simplicity, we will focus on a _{single pair (t d} , f _{D). The shoulder terms −t d} / 2 and −f _D / 2 of the exponential function in Eq. (7) play an essential role in the non-overlapping superposition method and the like. _{First, Τ td, fD z (t} ), Τ f fD, is t _d, f _D in symmetry is established (TD signal (t _d, f _D) between the _-td Z (f), Fully corresponds to _{f D} , -t _d in the FD signal). Next, in the _{solution problem of two unknowns (t d} , f _D ), it is natural to prepare two functions z (t) and Z (f). _{When estimating t d} (resp.f _D ) for loose functions z (t) and Z (f) from the shoulder of the exponential function of equation (7), Z (f) (resp.z) is easy to detect. It is understood that (t)) should be processed in parallel and considered.

Next, by introducing Eq. (9), signal theory and quantum mechanics are made to correspond.

Equation (10) is a Heisenberg's Canonical Commutation Relation (CCR) for the Hilbert space Η self-accompaniment operators Q and P. However, h / (2π) is Planck's constant. Stone-von Neumann theorem proved that the operators Q and P satisfying equation (10) are the only Schrodinger representations in equation (11). According to this theory, the Baker-Cambell-Hausdorff formula gives the Weyl form of Heisenberg's CCR in formula (13) by the one parameter unitary operator for operators Q and P in formula (12). Also, the Weyl form of Eq. (15) is obtained by von Neumann's two parameter operator in Eq. (14).

The correspondence between the quantum mechanics of Eq. (16) and the time / frequency shift operator means that the symmetric time / frequency shift operator Τ _{τ, ν} has the same form as the Neumann operator S (a, b). do. Furthermore, the phase term in Eq. (15) is expressed as a 2 × 2 determinant as in Eq. (17) in order to clarify the chain rule of group operations. From this, the distortion-free condition τν∈Z corresponds to the classical limit of Dirac's basic quantum condition in Eq. (18): h / (2π) → 0. However, the two operators [P, Q] = PQ-QP are Heisenberg's brackets. In Eq. (17), the important relational expression of Eq. (19) was used. This is because (i) both operators are non ^-commutable _{: Τ td, 0} Τ _{0, fD} = e -j2πtdfD Τ _{0, fD} Τ _{td, 0} and a phase term occurs, (ii) addition of both shift amounts. It means that a phase term is generated with the sex. In particular, when the pulse train p (t), which is often used in radar, is used as the emission signal, it is strongly suggested that a correlation function that compensates for the phase distortion of the received signal via the propagation path having _{(t d} , f _{D) is necessary.} is doing.

In order to enable high-precision ranging, design the transmission signal of the 4-level hierarchical structure of Eq. (20).

First, for the realization of the causal discrete-time prototype filter of length _{L g (resp.L G), g} [k] (resp. G [l]) the interval _{[-L g Δt / 2, L} g Δt / 2] (resp. [-L _G Δf / 2, L _G Δf / 2]) time (resp. Band) limit g (t) (resp. G (f)), and DΔt / 2 receives the delay _{D = L g -1 (resp.D'Δf /} 2, D '= L G -1), is defined by the normalized function of equation (21). Hereinafter, the discrete-time TD and FD functions are functions obtained by subjecting the continuous-time TD and FD functions to a time delay D / 2 and a frequency shift D'/2, respectively.

The signature v (t; Χ) of the TD and the V (f; Χ) of the FD are represented by the equations (22) to (25). The TD template u ^TD _s' (t; X) and the FD template U ^FD _s (f; X') are embedded, respectively. However, the TD signal of the template is the equation (26) and the equation (27). The FD signals of the template are the equations (28) and (29). After all, equations (30) to (33) hold.

_{When X m} and X'm of the period N and _{N'are applied to the infinite length m} ∈ Z and m'∈ Z, the above equation is given by equation (35) using the modulated filter (MF) property of equation (34). It becomes. Symbol lcm [M, N '] = M 0 N' = MN '0, lcm [M', N] = M ' introduces _{0 N = M'N 0, M 0} N of the formula (36)', M ' ₀ By introducing N polyphase components, each polyphase filter (Type 1 polyphase of Vaidyanathan) in Eq. (37) is defined. Therefore, the SFBs of the two-dimensional time-frequency spread code-modulated signatures v [k] and V [l] of FIGS. 2 (a) and 2 (b) can be obtained. 2 (a) and 2 (b) show SFBs having TD and FD SS codes for TD signature v [k] and FD signature V [l], respectively. 2 (a) is formula (23) and TD signature v TD for [k] of the (35), FD of SS symbols X _m, X _'m' and m 'th TD template u ^TD _m' [k] It is SFD of (0 ≦ m'≦ N'-1). 2 (b) is formula (25) and TD for FD signature V [l] of the (35), FD of SS symbols X _m, X _'m' and m 'th FD template U ^FD _m [l] ( 0 ≦ m ≦ N-1) SFB.

The transmit signal in which the information signal is embedded is Eq. (38) and Eq. (39). The FD transmission signals of the FT are equations (40) and (41). Equation (42) is an SFB of the transmission signal of FIGS. 3 (a) and 3 (b). However, when the modulated filter (MF) characteristics of the data transmission of the equations (43) and (44) are used, the polyphase filter (Type 1 polyphase of Vaidyanathan) is defined by the equation (45), respectively. FIG. 3A is an SFB of _{data {d p, p'} } ^P, _{P'p, p'= 1} for the TD transmission signal s [k] defined by the equations (39) and (42). FIG. 3B is an SFB of _{data {d p, p'} } ^P, _{P'p, p'= 1} for the FD transmission signal S [l] defined by the equations (41) and (42). In the normal SFB, since there is no signature generation process, FIGS. 2 (a) and 2 (b) are unnecessary. In FIGS. 3A and 3B, when N = N'= 1, it corresponds to a normal SFB.

The baseband complex signal r (t) on the output side of the heterodyne receiver and its FT R (f) = F [r (t) via a channel having t _d ≈ k _d Δt, f _D ≈ l _{D Δf.} ] (F) are equations (46) to (49), respectively. However, η (t) and E (f) are the interference component and its FT, ξ (t) and Ξ (f) are the external noise and its FT, and α and φ are the channel attenuation and phase characteristics. ..

Type-3 TD correlator with control parameters, estimation parameters (μ, ^ t _d ) ≈ (l _μ Δf, ^ k _d Δt) (its frequency dual: (σ; ^ f _D ) ≈ (k _σ Δt,) A type-4 FD correlator with ^ l _D Δf) and each estimation template are given. On the receiving side, assuming incoherent communication, TD, spreading codes of the FD is _'a chip address, data address each (n, n Y = {Y n}, Y' = {Y 'n}'), → Let q = (q, q'). The correlation functions of type-3 and type-4 are Eq. (50) and Eq. (51), respectively. However, the even function of the pulse waveform: g [k] = g [Dk], G [l] = G [D'-l] was used.

Modulated filter coefficients AFB of filter characteristics and _{P '0 MN, P 0 M'N} ' (Formula (52)) is used number of polyphase components (formula (53)), respectively, N ', N number of polyphase filter (Equation (54)) (Type 2 polyphase of Vaidyanathan) is defined, and the AFBs of FIGS. 4 (a) and 4 (b) are obtained. FIG. 4A shows an AFB with a TD correlator array for the reception of _{{d p, p'} } ^P _{p = 1, 1 ≦ p'≦ P'.} 4 (b) shows an AFB with FD correlator array for the reception of ^{_{{d p, p '} P}} ' p '= 1, 1 ≦ p ≦ P. N = N '= 1, N 0 = N' when ₀ = 1 is a normal AFB.

FIG. 5 shows (a) TD, (c) Analysis Filter Bank of FD correlator array and (b) Neumann's APT. 5 (a) and 5 (c) show P × P'information data {d _{→ p} } ^P, _{P'p, p'= 1} ^→ N, N'in units of p = (p, p'), respectively. Extract the case of a single data of specific p-Band, p'-Duration in the correlator array on the final output side consisting of pieces, and extract only multiple array type correlators for the parameter estimation method of the maximum likelihood method. ing. Synchronization can be achieved by using a variable filter.

That is, the TD and FD type AFBs in FIGS. 5 (a) and 5 (c) have arg max in order to detect signals without information from received signals including distortion due to _{t d} and f _{D, as shown in FIG. 5 (b).} obtained in the calculation, adopting each maximum likelihood estimate l ^* _mu, the k ^* _sigma other, FD, ^ l _D of TD, to ^ k _d, von Neumann's APT alternating as shown in FIG. 5 (d) Update. In addition, in the case of a single data, FIG. 4 requires only the correlator array of the central block as shown in FIGS. 5 (a) and 5 (c). After all, in order to detect PP'information data, P'and P array type correlators of N and N'are arranged in parallel in TD and FD, respectively, and non-superimposed transmission of data of time width T and bandwidth F is performed. _{High-precision, high-speed detection of t d} and f _D , which are evenly present in the time / frequency domain of PT × P'F required for the above, and high-speed detection of information data are executed in parallel at the same time.

Incidentally, the conventional OFDM etc. is a multi-carrier method, t _d, if a full synchronization at f _D -free environment, communication without the synchronization shift occurs for normal operation. The basis for this is a design method that satisfies the commutative condition (distortion-free condition): TF ∈ Z in the case of transmission of a signal having a time width T and a bandwidth F. However, Eq. (57) shows the phase distortion due to _{the deviation ε T} and ε _{F << 1 of T and F.} Fluctuations lead to a serious situation because it is directly linked to the dramatic decrease in the real part of the correlator from equation (57). However, the "absolute value manipulation" often used in demodulators only obscures this.

Since the diffusion codes are independent, the cross-correlation <ψ _{η, s', → p} [k], u ^TD _{s', → p} [k; Y]> _{d, k} is assumed to be relatively small and approximated. The cross-correlation between the input CE ψ _w [k; X] of the element Ae ^{iκ to the receiver and the impulse response of the filter matching the (s', →} p) th TD template is defined as Eq. (58). Here, d _{→ p} = 1, and the code Y is used instead of X.

Using the inner product of the discrete-time TD function in space l ₂ (Z), Eq. (59) is obtained. However, θ _xy (・, ・) and Θ _XY (・, ・) are ambiguity functions defined by Eq. (60). The ambiguity function is closely related to the Wigner function in quantum mechanics. In addition, the inner product between two time- and frequency-shifted functions is simply expressed as equations (61) to (63).

From equation (59), the phase term has five phase distortion elements: 1) channel distortion + multiplexing of the symbol level of the transmission signal, 2) multiplexing of the chip level of the transmission signal, and 3) of the symbol level of the receiver. It consists of channel distortion estimation by multiplexing, 4) multiplexing of the chip level of the received signal, and 5) phase of the ambiguity function type correlator. In the past, we were not aware of distortion at the data level and chip level, so let's reduce _{^ k d} -k _d and l _μ −l _{D without considering the data level deviation and chip level deviation.} It was supposed to be.

Unfortunately, in general, θ _gg (τ, ν) and Θ _GG (ν, −τ) have many side lobes. However, Gaussian pulse g (t) provides a drastic solution to the estimation problem. This is because the Gaussian pulse has a separable and exponentially decaying ambiguity function as shown in Eq. (64).

A correlator output with a large value of N, N'>> 1 means that ^{the first and second arguments of the ambiguity function θ d} _gg are close to 0 (ie ^→ p = ^→ q). ^{Therefore, all the terms of →} p ≠ ^→ q in Eq. (59) can be ignored.

When ν ₀ [l _μ ] = l _μ −l _D and τ ₀ [^ k _d ] = ^ k _{d −} k _d are introduced, the sum of the three exponents of Twiddle factor W = e ^{-j2π / L (Equation (65)} )) Is given. Calculate the sum of the three diffusion codes. If the receiver is address ^→ p = ^→ q, n = m, and Y = X, Y'= X', equation (66) holds. The sum is expressed by the Dirichlet fucntion form as in Eq. (67).

Since the diffusion codes are independent, the cross-correlation <Es,, _{→ p} [l], U ^FD _{s, → p} [l; X]> _{d, l} is assumed to be relatively small and approximated to the receiver. ^{The cross-correlation} between the FD input CE Ψ _W [l; X] of the element Ae ^{iκ and the FD impulse response of the filter matching the (s, →} p) th FD template is defined as Eq. (68). Here, d _{→ p} = 1, and the code Y'is replaced with X'. This can be rewritten as Eq. (69).

Similarly, a correlator output with a large value of N, N'>> 1 means that ^{the first and second arguments of the ambiguity function Θ d} _GG are close to 0 (ie ^→ p = ^→ q). ^{Therefore, all the terms of →} p ≠ ^→ q in Eq. (69) can be ignored.

When ν ₀ [l _μ ] = l _μ −l _D and τ ₀ [^ k _d ] = ^ k _{d −} k _d are introduced, the sum of the three exponents of Twiddle factor W = e ^{-j2π / L (Equation (70)} )) Is given. Calculate the sum of the three diffusion codes. If the receiver is an address ^→ p = ^→ q, n = m, and Y = X, Y'= X', equation (71) holds.

The two correlators are perfectly symmetric with respect to _{t d} and f _D (or k _d , l _{D ∈ Z).} By exchanging the TD SS code and FD code of the pair of c _{→ p,} ^{s'TD, d} (l _μ ; k _d ) and C _{→ p, s} ^{TD, d} (k _σ ; ^ l _{D), the equation (} 72) and other pairs of correlators in Eq. (73) are obtained.

Ae ^jκ and (t _d , f _D ) estimates are separable. Therefore, for example, for two sets of correlators (so-called complementary pairs) defined by the equation (74), an integer pair (^ k _{d, s} , ^ l _{D, s} ) is used by using the equation (75). Is updated as _{^ l D, s} = l ^* _μ when s is an even _{number, and ^ k d, s} = k ^* _σ when s is an odd number. The initial value (^ k _{d, 0} , ^ l _{D, 0} ) can be arbitrarily selected. (^ k _{d, 0} , ^ l _{D, 0} ) = (0,0) may be used. Select l ^* _μ and k ^* _σ as candidates for ^ f _{D, s + 1} and ^ t _{d, s + 1} , respectively. If _{| ^ t d, s + 1} - ^ t d, s | <T C / 2 and _{| ^ f D, s + 1} - ^ f D, s | <F C / 2 then the ends of the estimation procedure ..

Since the 2d-SS code involved in the calculation of the four types of correlator outputs and the corresponding series sum can be organized in the DFT and IDFT formats, they are abbreviated _{as F ○} and F _○ ^{-1 respectively.} The output value can be organized by the triple series sum. However, ◯ means n, n'of the running variable.

The control parameter that maximizes the real part of both correlation values and its correlator number are defined by equations (77) and (78), and the fixed parameters ^ t _{d, r} , ^ f _{D, r} of each other's r-step are updated alternately. do. This is the application of von Neumann's APT (Alternating Projection Theorem ), (μ *, s *) corresponds to M = N by M or TD signal detection method likelihood function of Maximum Likelihood (ML), n ^* Is the signal type detection number and μ ^* is the optimum estimation value of _{f D.} ^{That is, the variable parameter μ * of the} TD function that gives the maximum value in the argmax operation becomes the ML estimate of f _{D , becomes the f D} estimate of the FD function, and the variable parameter σ ^* of the dual FD function becomes the ML estimate of t _d. becomes, the t _d estimated value of the TD function. Give each other an estimate of the other. The operation of argmax performs the _{optimum estimation of f D in} the TD function and the optimum estimation of t _{d in} the FD function. Parameter estimation is diagonal. This appears in equation (82). The lowercase correlation function performs integration in the time domain, and the uppercase one performs integration (inner product) in the frequency domain. On the other hand, (σ ^* , s ^* ) corresponds to M = N'in the M-type FD signal detection method using the likelihood function of ML (the variable is a frequency variable), and s ^* is the signal type detection number and σ ^* is. This is the optimum estimate of _{t d.}

The original pair in equation (77) operates at a high SNR, the complementary pair in equation (78) operates at a low SNR, the former has a large number of steps to convergence, and the latter has a small number. The trade-off relationship between SNR and the number of steps has been confirmed by numerical simulation. The results using the pulse train by SS code are equations (79) and (80). However, Δt and Δf are sampling time intervals and frequency intervals of Gaussian waves z (t) = e ^{-πt ^ 2 / st ^ 2} and Z (f) = e ^{-π (fs_t) ^ 2, respectively.} Dispersion of the Gaussian waves, each _{^{s t 2 = Σ k k 2}} g [k], s f 2 = Σ l l 2 G [l], a _{(s t · s f = 1} / 2π). d ^'2 is spectral amplitude N ^-1 by SS modulation, N' Considering ^{-1, d '2 = d 2} N 2, d' f 2 = N ' than ² d ^2, the right side are each 1 / (4π ² It becomes d' ² T _c ² ) and 1 / (d _f ' ² F _c ^2). However, NT _c = T and N'F _c = F.

To express in the form of AFB, the calculation of both correlators is the received signal r (t) and the template u ^TD _{n ′} (t; Y) (its frequency dual: R (f) and the template U ^FD _n (f;). It was organized in the form of a template matching with Y')).

As other precautions, first, the type-1 and type-2 templates will be described. By exchanging the roles of the TD and FD codes, another template decomposition is possible.

TD of the signature signal v (t; X) and (Equation (81)), the FD: V (f; X) (X = (X; X ') ( Equation (82)), respectively, TD template u _n ^{FD (t; X '),} FD template U _n' ^TD.; in (f X), is decomposed formula (83) and (84) i.e., signature v a (t) (resp.V (f) ) As an emission signal, the received signal passes through the propagation path of _{(t d} , f _{D ) and becomes noise and interference with T t_d, f_D} v (t; X) (resp. T ^f _{f_D, -t_d} V (f; X)). However, the TD and FD signals of each template are the equations (85) and (86).

The proposed method showed that the TD-FD representation (Gabor expansion) of the signal can be applied to the division of the unknown parameter space (τ, ν). Diffuse code modulation is important because it is desirable that the M types of signals s _j (t) and 1 ≦ j ≦ M are independent of each other. In order to perform high-precision parameter estimation, N × N'division by the two-dimensional diffusion code divides TD and FD by N', respectively, and divides the parameter search space into N and N', so that the above M types Corresponds to hypothesis creation.

The time and effort O (NN') for parameter estimation calculation has been reduced to O (N + N'). That is, N and N'hypotheses were created for TD and FD, respectively. As a result of the parameter search space division, new phase distortions occur, so N and N'correlators with phase distortion compensation are required. That is, there is a trade-off relationship between improving the accuracy of parameter estimation and increasing various phase distortions. These are necessary costs because they assume Binary incoherent channel and Incoherent M-ary Channel. Also, with coarse parameter estimates, the likelihood ratio does not reach the desired likelihood ratio level, which is expected to result _{in low False Alarm probabilities Q 0} and detection probabilities Q _d. Since the various statistical methods of Helstrom described above have been mainly discussed in TD, it is necessary to pay attention to TFS (Time-Frequency Symmetry) and define the likelihood function and the likelihood ratio in FD.

Based on the above theoretical background, an example of the configuration and operation of the signal processing system according to the embodiment of the present invention will be described with reference to FIGS. 6 and 7.

FIG. 6 shows an example of (a) configuration and (b) operation of the signal processing system 1 according to the embodiment of the present invention.

With reference to FIG. 6A, the signal processing system 1 includes a transmitting device 3 and a receiving device 5. The transmission device 3 includes a diffusion code storage unit 7, a transmission signal generation unit 9, and a transmission signal transmission unit 11. The receiving device 7 includes a receiving signal receiving unit 13, a compensating unit 15, and a diffusion code storage unit 16. The spreading code storage unit 7 of the transmitting device 3 stores the spreading code, and the spreading code storage unit 16 of the receiving device 5 stores the corresponding spreading code.

An example of the operation of the signal processing system 1 of FIG. 6A will be described with reference to FIG. 6B. The diffusion code storage unit 7 stores the time diffusion code and the frequency diffusion code. The transmission signal generation unit 9 generates a transmission signal by two-dimensional phase modulation using the time diffusion code and the frequency diffusion code (see equations (43) and (45)). ^{Here, according to Eqs} . (43) and (45), (-1) qq'NN'is included. This value is -1 when q, q', N, and N'are all odd numbers, and 1 in other cases. Therefore, phase distortion occurs when q, q', N, and N'are all odd numbers. This is the phase distortion due to the two-dimensional phase modulation. This phase distortion was clarified by changing from a continuous time variable to a discrete time variable. The transmission signal transmission unit 11 transmits a transmission signal to the reception device 5 (step ST2). The reception signal receiving unit 13 receives the transmission signal and obtains the reception signal (step ST3). The compensation unit 15 compensates for the phase distortion of the received signal due to the two-dimensional phase modulation using the time diffusion code and the frequency diffusion code (step ST4). These processes can be realized by using a communication device such as an ordinary antenna or a computer.

In step ST1, the transmission signal generation unit 9 may perform data multiplexing and / or chip multiplexing by two-dimensional phase modulation. At this time, in step ST4, the compensation unit 15 compensates for the phase distortion due to data multiplexing and / or chip multiplexing.

FIG. 7 shows an example of (a) configuration and (b) operation of the signal processing system 21 according to the embodiment of the present invention.

With reference to FIG. 7A, the signal processing system 21 includes a transmitting device 23 and a receiving device 25. The transmission device 23 includes a diffusion code storage unit 27, a transmission signal generation unit 29, and a transmission signal transmission unit 31. The receiving device 27 includes a receiving signal receiving unit 33, an estimating unit 35, and a diffusion code storage unit 36. The estimation unit 35 includes a TD estimation unit 37 and an FD estimation unit 39. The spreading code storage unit 27 of the transmitting device 23 stores the spreading code, and the spreading code storage unit 36 of the receiving device 25 stores the corresponding spreading code.

An example of the operation of the signal processing system 21 of FIG. 7A will be described with reference to FIG. 7B. The diffusion code storage unit 27 stores the time diffusion code and the frequency diffusion code. The transmission signal generation unit 29 generates a transmission signal by two-dimensional phase modulation using a time diffusion code and a frequency diffusion code using a shift operator that performs a time shift and a frequency shift that satisfy the symmetry between time and frequency. (Step STE1). The transmission signal transmission unit 31 transmits a transmission signal to the reception device 25 (step STE2). The reception signal receiving unit 33 receives the transmission signal that has passed through the propagation path and obtains the reception signal (step STE3). Estimation unit 35, for example, formula (56) and as described for formula (82) and (83), estimate the delay time t _d and the Doppler shift f _D in a propagation path to compensate for the phase distortion by the two-dimensional phase modulation (Step STE4). These processes can be realized by using a communication device such as an ordinary antenna or a computer. Further, the TD estimation unit 37, the FD estimation unit 39, and the like may use dedicated hardware.

That is, the TD estimation unit 37 compensates for the phase distortion due to the two-dimensional phase modulation and obtains the maximum likelihood estimation value of the _{Doppler shift f D.} FD estimator 39 obtains the maximum likelihood estimate of compensating for the phase distortion by the two-dimensional phase modulation delay time t _d. In step STE4, estimating unit 35, by the TD estimator 37 and the FD estimator 39, while the by the other the maximum likelihood estimate is updated alternately by using an estimated delay time t _d and the Doppler shift f _D (See equation (56), etc.).

The present invention can be used in fields such as radar and wireless communication.

1,21 signal processing system, 3,23 transmitter, 5,25 receiver, 7,27 diffused code storage, 9,29 transmit signal generator, 11,31 transmit signal transmitter, 13,33 receive signal receiver , 15 Compensation section, 35 estimation section, 37 TD estimation section, 39 FD estimation section

Claims (4)

A signal processing system that uses two-dimensional phase modulation.
Equipped with a transmitter and a receiver
The transmitter is
A transmission signal generation means that generates a transmission signal by two-dimensional phase modulation using a time diffusion code and a frequency diffusion code.
A transmission signal transmission means for transmitting the transmission signal is provided.
The receiver is
A reception signal receiving means that receives the transmission signal and obtains the reception signal,
A compensation means for compensating for the phase distortion in the received signal is provided.
The compensating means compensates for the phase distortion due to the two-dimensional phase modulation using the time diffusion code and the frequency diffusion code.
The time domain and the frequency domain are divided into rectangular regions by the number of divisions N and N', respectively.
If the transmission signal generation means generates the transmission signal by the signature of the time domain by the linear coupling by the diffusion code of the N'time domain template, the phase distortion compensated by the compensation means is m (0 ≦ m). For ≤N-1) and m'(0≤m'≤N'-1), m and for the m + 1st Gaussian pulse to generate a template for the m'+ 1st time domain in the linear coupling. Includes those that occur when both m'are odd and do not occur when either m or m'is even.
If the transmission signal generation means generates the transmission signal by the signature of the frequency domain by the linear coupling by the diffusion code of the template of N frequency domains, the phase distortion compensated by the compensation means is m (0 ≦ m ≦). For N-1) and m'(0≤m'≤N'-1), m and m for the m'+ 1st Gaussian pulse to generate a template for the m + 1th frequency domain in the linear coupling. A signal processing system that occurs when both'are odd and does not occur when either m or m'is even. A signal processing system that uses two-dimensional phase modulation.
Equipped with a transmitter and a receiver
The transmitter is
A transmission signal generation means that generates a transmission signal by two-dimensional phase modulation using a time diffusion code and a frequency diffusion code.
A transmission signal transmission means for transmitting the transmission signal is provided.
The receiver is
A reception signal receiving means that receives the transmission signal and obtains the reception signal,
A compensation means for compensating for the phase distortion in the received signal is provided.
The compensating means compensates for the phase distortion due to the two-dimensional phase modulation using the time diffusion code and the frequency diffusion code.
The time domain and the frequency domain are divided into rectangular regions by the number of divisions N and N', respectively.
A data address (q, q') is specified and a two-dimensional phase-modulated chip waveform is assigned to each divided rectangular area.
The transmission signal generation means uses a signature obtained by linearly combining chip waveforms assigned to each rectangular region with a two-dimensional diffusion code as a transmission signal.
The phase distortion compensated by the compensating means includes those that occur when q, q', N and N'are all odd numbers and do not occur when any one of q, q', N and N'is even. Signal processing system. A receiving method in which a transmission signal generation means provided in a transmission device receives a transmission signal generated by two-dimensional phase modulation using a time diffusion code and a frequency diffusion code.
The compensating means includes a compensating step of compensating for the phase distortion due to the two-dimensional phase modulation using the time spreading code and the frequency spreading code in the received signal obtained by receiving the transmitted signal.
The time domain and the frequency domain are divided into rectangular regions by the number of divisions N and N', respectively.
If the transmission signal generation means generates the transmission signal by the signature of the time domain by the linear coupling by the diffusion code of the N'time domain template, the phase distortion compensated by the compensation means is m (0 ≦ m). For ≤N-1) and m'(0≤m'≤N'-1), m and for the m + 1st Gaussian pulse to generate a template for the m'+ 1st time domain in the linear coupling. Includes those that occur when both m'are odd and do not occur when either m or m'is even.
If the transmission signal generation means generates the transmission signal by the signature of the frequency domain by the linear coupling by the diffusion code of the template of N frequency domains, the phase distortion compensated by the compensation means is m (0 ≦ m ≦). For N-1) and m'(0≤m'≤N'-1), m and m for the m'+ 1st Gaussian pulse to generate a template for the m + 1th frequency domain in the linear coupling. A receiving method including those that occur when both'are odd and do not occur when either m or m'is even. A receiving method in which a transmission signal generation means provided in a transmission device receives a transmission signal generated by two-dimensional phase modulation using a time diffusion code and a frequency diffusion code.
The compensating means includes a compensating step of compensating for the phase distortion due to the two-dimensional phase modulation using the time spreading code and the frequency spreading code in the received signal obtained by receiving the transmitted signal.
The time domain and the frequency domain are divided into rectangular regions by the number of divisions N and N', respectively.
A data address (q, q') is specified and a two-dimensional phase-modulated chip waveform is assigned to each divided rectangular area.
The transmission signal generation means uses a signature obtained by linearly combining chip waveforms assigned to each rectangular region with a two-dimensional diffusion code as the transmission signal.
The phase distortion compensated by the compensating means includes those that occur when q, q', N and N'are all odd numbers and do not occur when any one of q, q', N and N'is even. Receiving method.

Images