JP2008219571A - Multipath suppression device and multipath suppression method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To alleviate difficulty of ranging based on transmission time of radio wave under environment that mutlipaths superimpose. <P>SOLUTION: The multipaths are suppressed by assuming a plurality of models superimposed by the multipaths, evaluating reliability of the models, selecting a model with high reliability, obtaining the transmission time of direct wave based on the selected model and obtaining the distance between a signal source and a receiver. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a technique for measuring the distance or position between a transmission source and a receiver using a signal (radio wave) transmitted from the transmission source.

In GPS (Global Positioning System) and radar systems, the effect of multipath is one of the major error factors in ranging and positioning. Until now, in order to reduce multipath errors, MMT (Multipath Mitigation Technology) (Patent Document 1) and Narrow Correlator (Non-Patent Document 1, page 120) have been developed.
MMT is known to be able to reduce the influence of multipath with a short delay distance, which has been difficult in the past, by estimating the parameters of the signal model using the maximum likelihood estimation method in the signal model of one direct wave and one multipath signal. It has been.
Non-Patent Document 2 describes information criterion, and Non-Patent Document 3 describes model selection.

U.S. Patent No. 6,370,207 Global Positioning Systems, Inertial Navigation, and Integration Mohinder S. Grewal, Lawrence R. Weill, Angus P. Andrews John Wiley & Sons, Inc., 2001 Konishi, Kitagawa: “Information Criteria”, Asakura Shoten (2004) Shimohira, Ito, Kubogawa, Takeuchi: "Model selection", Iwanami Shoten (2004)

  Since the conventional multipath suppression device presupposes a model composed of a total of two waves of a direct wave and one multipath wave, when there is no multipath or when two or more multipaths are superimposed Has a problem that the error becomes large. An object of the present invention is to suppress the difficulty of ranging caused by multipath.

  The multipath suppression apparatus of the present invention assumes signal acquisition means for acquiring a signal transmitted from a transmission source at a receiver, and a plurality of models in which a multipath is superimposed on the signal, and the reliability among the plurality of models. Analyzing the propagation time of the direct wave based on a high-level model and calculating the distance between the transmission source and the receiver using the propagation time, and the output that outputs the result obtained by the signal analysis means Means.

  In addition, the multipath suppression method of the present invention assumes a signal acquisition step of acquiring a signal transmitted from a transmission source by a receiver, and a plurality of models in which a multipath is superimposed on the signal. Analyzes the propagation time of the direct wave based on a reliable model, calculates the distance between the transmission source and the receiver using the propagation time, and outputs the results obtained in the signal analysis process And an output process.

  The multipath suppression apparatus of the present invention assumes signal acquisition means for acquiring a signal transmitted from a transmission source at a receiver, and a plurality of models in which a multipath is superimposed on the signal, and the reliability among the plurality of models. Analyzing the propagation time of the direct wave based on a high-level model and calculating the distance between the transmission source and the receiver using the propagation time, and the output that outputs the result obtained by the signal analysis means The multipath can be suppressed with respect to the received signal on which the multipath having an undetermined wave number is superimposed.

  In addition, the multipath suppression method of the present invention assumes a signal acquisition step of acquiring a signal transmitted from a transmission source by a receiver, and a plurality of models in which a multipath is superimposed on the signal. Analyzes the propagation time of the direct wave based on a reliable model, calculates the distance between the transmission source and the receiver using the propagation time, and outputs the results obtained in the signal analysis process The output process is provided, so that the multipath can be suppressed with respect to the reception signal on which the multipath having an undetermined wave number is superimposed.

Embodiment 1 FIG.
Radio waves go straight when there are no obstacles, and their propagation speed is a constant speed called the speed of light. If the transmission time from the radio wave emitted at the point A going straight to another point B is measured and multiplied by the speed of light, the distance between the two points AB is measured. Many systems measure the transmission time of radio waves and use it for distance measurement. For example, the GPS identifies the distance between the satellite and the terminal by measuring the transmission time until the signal reaches the terminal from the GPS satellite. The radar also measures the delay time until the transmitted radio wave is reflected by the target and returns, and identifies the distance between the radar and the target.

  For example, when a radio wave traveling straight from point A is reflected at another point C and the reflected wave travels straight and reaches point B, the transmission time of the radio wave passing through this route is equal to the distance between two points AC and two points. It is equal to the sum of the distance between CBs divided by the speed of light. The phenomenon in which radio waves are transmitted through a plurality of paths is called multipath. When multipath is occurring, radio waves emitted from the point A reach the point B with different transmission times, so that it is difficult to uniquely determine the transmission time. Therefore, it is difficult to uniquely determine the distance between the two points AB. This difficulty due to multipath is often a serious problem in many applications such as GPS positioning and radar target identification.

It is possible for the receiver to know in advance the signal information transmitted by the GPS satellite or radar transmitter. For example, GPS satellites are spread by a C / A code belonging to what is known as a 1023 chip long gold code, the contents of which have been published in advance. Using this code m (t), the transmission signal s (t) at time t is expressed by the following equation (1).

  Here, A represents the complex amplitude of the transmission signal, i represents the imaginary unit, and ω represents the angular frequency. The description is given by taking the GPS signal as an example, but the present invention can be applied to a system that uses a signal spread by using the code m (t) expressed by Expression (1).

ここで、モデルとは、物理的な対象である信号に対応する、数学的な表現を指す。ａ、bを実数として、a＋ｉｂは受信信号の複素振幅、τは信号源から受信機までの信号伝達時間、ｎ（ｔ）は時刻ｔにおける雑音をそれぞれ表す。 The receiver that has received the transmitted signal obtains a baseband signal from which the carrier wave component has been removed. The received signal r (t) at time t is modeled by equation (2).

Here, the model refers to a mathematical expression corresponding to a signal that is a physical object. With a and b as real numbers, a + ib represents the complex amplitude of the received signal, τ represents the signal transmission time from the signal source to the receiver, and n (t) represents the noise at time t.

ここで、ａ_p＋ｉｂ_pとτ_pとは、第ｐ番目の信号の複素振幅と伝達時間とをそれぞれ表す。（３）式が表すＰ波モデルは、これらＰ波のベースバンド信号に雑音が重畳した信号を表している。もしも、観測標本から、直接波の伝達時間の値が得られると、その値を利用して送信機と受信機との間の距離を得ることが可能になる。 When the receiver receives a signal in which a direct wave and a multipath wave are combined and superimposed on a P wave, the received signal is represented by a P wave model according to Equation (3).

Here, a _p + ib _p and τ _p represent the complex amplitude and transmission time of the p-th signal, respectively. The P wave model represented by the expression (3) represents a signal in which noise is superimposed on the baseband signal of these P waves. If the value of the propagation time of the direct wave is obtained from the observation sample, it is possible to obtain the distance between the transmitter and the receiver using the value.

Now, it is assumed that the sample {r (t _n )} _n of the baseband signal of the receiver is sampled at time t _n (n = 1, 2, 3,..., N) having a sufficiently short time period. To do. Here, N is the number of samples. At this time, the sampling sample {r (t _n ) of the probability density function based on the P-wave model when the parameter values θ _P = {{a _p } _p , {b _p } _p , {τ _p } _p } are fixed } The value for _{n is} expressed as f _P ({r (t _n )} _n | θ _P ). This value is regarded as a function of θ _P and is called a likelihood function, and its natural logarithm is called a log likelihood function. The method of selecting the parameter value θ _{P, ML} that maximizes the likelihood function value as the estimated value of the parameter is the maximum likelihood method.

しかし、このＰ波モデルの分布関数に、最尤法を適用すると、波数Ｐが発散するという困難が発生する。 Here, the symbol arg max _θ g (θ) represents the value of the variable θ that maximizes the function g (θ). An error factor remaining in the receiver that suppresses systematic errors other than multipath waves is thermal noise. Thermal noise is an electromagnetic wave generated by thermal energy, and its distribution function is described by a Gaussian distribution with an average 0 standard deviation σ. From the thermal noise probability density distribution function, the probability distribution function of the observed signal in the P-wave model of equation (3) is expressed by equation (5).

However, when the maximum likelihood method is applied to the distribution function of the P wave model, the difficulty that the wave number P diverges occurs.

  If the pth complex amplitude is selected as 0 (a + ib = 0) among the parameters of the P wave model, it matches the (P-1) wave model. Including. Therefore, the model that maximizes the likelihood among the (P-1) wave models is included in the P wave model. As a result, the maximum likelihood value in the (P-1) wave model is included in the likelihood value of the P wave model. Therefore, the parameter value that maximizes the likelihood is included in a model having a larger number of waves, and the wave number P of the model selected in the maximum likelihood estimation diverges. Therefore, in an environment where the number of multipath waves cannot be determined in advance, it is impossible to estimate the propagation time of a direct wave by a simple maximum likelihood method.

  FIG. 1 is a block diagram of a multipath suppression apparatus according to the present invention. In order to overcome the difficulty that the model wave number diverges, the present invention utilizes the means shown in FIG. The signal acquisition unit 1 receives a physical signal by the antenna 11, converts the frequency by the frequency conversion unit 12, and samples it by the sampling unit 13. In the signal acquisition means 1, the signal transmitted from the transmission source is acquired by the receiver. The signal processing unit 2 converts the sample obtained by the signal acquisition unit 1 into a baseband signal by the baseband signal conversion unit 21 and stores it in the storage unit 22. You may comprise these using a well-known technique. The signal analysis unit 3 optimizes the likelihood function with respect to the baseband signal stored by the signal processing unit 2 by the likelihood function optimization unit 31 in the parameter space of the model in which the wave number is sequentially increased. . This likelihood function optimizing means 31 may be configured by a known technique such as Newton's method. The model evaluation selection means 32 compares and selects models based on the reliability of the models. Here, the reliability is a quantification of the probability that a sample is collected from the model. The output means 4 outputs a parameter value including a direct wave transmission time obtained based on the selected model. The signal analysis means 3 assumes a plurality of models in which a multipath is superimposed on the signal, analyzes the propagation time of the direct wave based on a model having high reliability among the plurality of models, and transmits using the propagation time. The distance between the source and the receiver will be calculated.

FIG. 2 is a block diagram of the multipath suppression apparatus. For example, a GPS signal transmitted from a GPS satellite serving as a transmission source is received by the antenna 11 of the receiver. The received signal is frequency-converted to an intermediate frequency signal by an RF (Radio Frequency) module 52 and sampled as a digital signal at a predetermined period by an A / D converter 53. The sampled signal is converted into a baseband signal (corresponding to {r (t _n )} _n above) by the baseband signal conversion means 21. The baseband signal is stored in the RAM 57 that is the storage unit 22. The signal processing analysis means 3 is stored as a program in the ROM 56 and executed by the CPU 55.

  FIG. 3 is a flowchart of the multipath suppression method of the present invention. The specific model selection steps are as shown in FIG. The wave number parameter is initialized to P = 1 (ST10). The maximum likelihood method using the collected sample is applied to the P wave model parameter, the likelihood function is optimized, and the result is stored (ST11). The maximum likelihood method is applied to the (P + 1) wave model parameter with the collected sample, and the result is stored (ST12). The reliability of the P wave model and the reliability of the (P + 1) wave model are each evaluated (ST13). If the reliability of the P wave model is equal to or higher than the reliability of the (P + 1) wave model, the P wave model is selected (ST13). In other cases, the value of the wave number parameter P is increased by 1 (ST14). For the new P value, the maximum likelihood method is applied to the P + 1 wave model parameter with the sampled sample, and the result is stored (ST12). Repeat until the end. From the definition of reliability, it stops at a finite value of the wave number parameter P. Finally, the parameter maximum likelihood values of the selected model and the distance measurement or positioning result using them are output and the process ends (ST15).

Model comparison means 32 in the present embodiment, when the baseband signal samples have been taken, the probability that the signal sample is generated from each P-wave model M _P assumed, that the posterior probability _{p (M P | {r (} t _n )} The approximate values of _n ) are compared in each P-wave model, and a model with a large value is selected as having high reliability. Posterior probability p is collected signal {r (t _{n)} n} is a probability that is generated from the P-wave model M _P, probability sampling data is generated by the P-wave model _{p P ({r (t n} ) } _n ) and the conditional probability defined by the equation (6) derived from the probability P (M _P ) generated by the P-wave model.

で近似しても良い。近似すると式（８）となる。

Here, assuming that the probability P (M _P ) generated by the P-wave model is constant, and further assuming that the number of samples N is large, this posterior probability is expressed by the Bayesian information criterion shown in Equation (7).

May be approximated by When approximated, Equation (8) is obtained.

  The validity of this approximation is described by the equation (6.19) on page 156 of Non-Patent Document 2. In this embodiment, approximation of the posterior probability value based on the Bayesian information criterion is described. However, the present invention is not limited to this approximation method, and the reliability of the model can be evaluated by the posterior probability value. That's fine. By using multiple signal generation models including multipaths, evaluating their reliability, and selecting a model with high reliability, multipath can also be applied to received signals on which multipaths with undecided wave numbers are superimposed. Can be suppressed. In addition, the magnitude of the posterior probability can be accurately approximated only from the maximum likelihood value and the number of parameters, the reliability of the model can be evaluated with high accuracy, and the accuracy of multipath suppression can be increased. it can. In addition, the calculation amount can be reduced.

  In addition, by incorporating the multipath suppression device of the present embodiment into a distance measuring device or a positioning device, a distance measuring device or a positioning device with high multipath suppression accuracy can be obtained.

  FIG. 4 is a diagram of transmission times of the first incoming wave and the second incoming wave with respect to actually measured data. This is a result of applying to a GPS signal actually collected outdoors. Here, the horizontal axis represents the elapsed time from the start of measurement, and the vertical axis represents the transmission time. Further, the first incoming wave in the selected wave number model is indicated by a point marker, and the second incoming wave when two or more models are selected is indicated by a cross marker. Note that the primary component in the code delay estimation amount of the first incoming wave is eliminated, and the average code delay amount of the first incoming wave is set to zero. From the figure, it can be seen that a multipath wave delayed by approximately 50 m can be identified, but this is a reasonable value from the measurement conditions.

Embodiment 2.
In this embodiment, the model comparison unit 32 replaces the posterior probability used in the first embodiment, and in this embodiment, the probability density function f ({r (t _n )} _n | θ _ML ) and the risk distribution RISK _P of the sample distribution function are compared with each other, and a model having a small value is selected as having high reliability. The risk RISK _P represents the degree of risk of deviating from the true distribution when a new sample value is collected using the maximum likelihood estimated value, and refers to an integral value defined by Equation (9).

Here, the expression r _k = r (t _k ) was used. Further, q (r _k ) (k = 1, 2,..., N) represents a sample distribution function. Since the number of samples N is large, these risks are represented by the Akaike information criterion (Equation 10).

(Equation 11).

  The validity of this approximation is described on pages 39 to 45 of Non-Patent Document 3. In this embodiment, the approximation of the risk value based on the Akaike information criterion is described, but the present invention is not limited to this approximation method, and any risk value that approximates the distribution of the signal sample can be evaluated. That's fine. When a sample baseband signal sample is collected, the posterior probability generated from the model assumed by the signal is evaluated, and the one with a high posterior probability value is selected as having high reliability.

  By using multiple signal generation models including multipaths, evaluating their reliability, and selecting a model with high reliability, multipath can also be applied to received signals on which multipaths with undecided wave numbers are superimposed. Can be suppressed. By calculating the Akaike information criterion and approximating the risk value, it is possible to accurately approximate the magnitude of the posterior probability from a simple calculation, and it becomes possible to evaluate the reliability of the model with high accuracy. The accuracy of multipath suppression is increased.

  In addition, by incorporating the multipath suppression device of the present embodiment into a distance measuring device or a positioning device, a distance measuring device or a positioning device with high multipath suppression accuracy can be obtained.

Embodiment 3 FIG.
In Embodiments 1 and 2, the method of positioning the distance between a receiver and one transmission source by suppressing multipath has been described. However, the propagation delay times of signals transmitted from a plurality of transmission sources are used. The position of the receiver can be calculated. In the case of positioning, at least three artificial satellites as transmission sources are required. When signals transmitted from four transmission sources are received, the time offset of the internal clock provided in the receiver can be adjusted, and accurate positioning is possible. On the other hand, when signals transmitted from three transmission sources are received, accurate positioning is possible, for example, by separately storing ground surface data on the receiver side. In this way, the signal acquisition unit 1 acquires signals from at least three transmission sources, and the signal analysis unit 3 measures the distance between the transmission source and the receiver, and uses the result to determine the position of the receiver. Positioning can be performed.

It is a block diagram of the multipath suppression device in Embodiment 1 of the present invention. It is a block diagram of the multipath suppression device in Embodiment 1 of the present invention. It is a figure of the flowchart of the multipath suppression method in Embodiment 1 of this invention. It is a figure of the transmission time of the 1st arrival wave and the 2nd arrival wave with respect to measurement data in Embodiment 1 of the present invention.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 Signal acquisition means, 2 Signal processing means, 3 Signal analysis means, 4 Output means, 11 Antenna, 12 Frequency conversion means, 13 Sampling means, 21 Baseband signal conversion means, 22 Storage means, 31 Likelihood function optimization means 32 model evaluation selection means, 52 RF module, 53 A / D converter, 55 CPU, 56 ROM, 57 RAM.

Claims (7)

Signal acquisition means for acquiring a signal transmitted from a transmission source by a receiver;
Assuming a plurality of models in which a multipath is superimposed on the signal, the transmission time of the direct wave is analyzed based on a model having high reliability among the plurality of models, and the transmission source is analyzed using the transmission time. Signal analysis means for calculating the distance to the receiver;
A multipath suppression device comprising: output means for outputting a result obtained by the signal analysis means. The signal analysis means evaluates the posterior probability generated from the signal with a plurality of models, and selects the model with the high posterior probability value as the model with high reliability. Multipath suppression device. The multipath suppression device according to claim 2, wherein the posterior probability value is approximated by calculating a Bayesian information criterion. The signal analysis means evaluates a risk value approximating a distribution of a sample of a signal with a plurality of models, and selects a model with the low risk value as the model with high reliability. Multipath suppression device. The multipath suppression apparatus according to claim 4, wherein the risk value is approximated by calculating an Akaike information criterion. The signal acquisition means acquires signals from at least three transmission sources,
6. The multipath suppression device according to claim 1, wherein the signal analysis means measures the position of the receiver using the calculated distance. 7. A signal acquisition step of acquiring a signal transmitted from a transmission source by a receiver;
Assuming a plurality of models in which a multipath is superimposed on the signal, the transmission time of the direct wave is analyzed based on a model having high reliability among the plurality of models, and the transmission source is analyzed using the transmission time. A signal analysis step of calculating a distance to the receiver;
An output step of outputting a result obtained in the signal analysis step.

Images