Estimation device and estimation method

Abstract

This estimation device comprises: an input unit that inputs channel state information (CSI) acquired by using a plurality of reception antennas and a plurality of transmission antennas; a first offset removal unit that divides the CSI among the reception antennas and that generates a first signal; a second offset removal unit that divides the CSI among the transmission antennas and that generates a second signal; a first phase difference correction unit that corrects the phase of a dynamic component extracted from the first signal by using a static component extracted from the second signal and that generates a third signal; a second phase difference correction unit that corrects the phase of a dynamic component extracted from the second signal by using a static component extracted from the first signal and that generates a fourth signal; an AoD estimation unit that estimates the AoD, which is the transmission direction of the radio wave, by using the third signal; and an AoA estimation unit that estimates the AoA, which is the arrival direction of the radio wave, by using the fourth signal.

Description

Estimation device and estimation method

[0001] The present invention relates to an estimation device and an estimation method.

[0002] There is a known technique for estimating a person's position by estimating information about the propagation path of radio waves reflected from a person using CSI (Channel State Information), which represents changes in amplitude and phase of Wi-Fi (registered trademark) radio waves due to propagation (see, for example, Non-Patent Documents 1 to 4).

[0003] Qian, Kun, et al., "Widar2. 0: Passive human tracking with a single Wi-Fi link.". Proceedings of the 16th annual international conference on mobile systems, applications, and services. 2018.Jin, Yue, et al., "WiSen: Zero-knowledge passive human tracking using a single Wi-Fi link.", IEEE Transactions on Instrumentation and Measurement 71 (2022): 1-15.Wu, Dan, et al., "WiTraj: Robust indoor motion tracking with WiFi signals.", IEEE Transactions on Mobile Computing 22.5 (2021): 3062-3078.Li, Xiang, et al., "IndoTrack: Device-free indoor human tracking with commodity Wi-Fi.", Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies 1.3 (2017): 1-22.

[0004] However, the CSI acquired by actual devices contains phase offsets, and in particular, carrier frequency offsets (CFOs) present a problem in that the accuracy of estimating the propagation path of radio waves using the CSI decreases.

[0005] The embodiments of the present invention have been made in view of the above-mentioned problems, and aim to improve the estimation accuracy of an estimation device that estimates a propagation path of a radio wave using CSI.

[0006] In order to solve the above problem, an estimation device according to an embodiment of the present invention includes an input unit that inputs CSI acquired using a plurality of receiving antennas and a plurality of transmitting antennas, a first offset removal unit that divides the CSI between the receiving antennas to generate a first signal, a second offset removal unit that divides the CSI between the transmitting antennas to generate a second signal, a first phase difference correction unit that corrects the phase of a dynamic component extracted from the first signal using a static component extracted from the second signal to generate a third signal, a second phase difference correction unit that corrects the phase of the dynamic component extracted from the second signal using the static component extracted from the first signal to generate a fourth signal, an AoD estimation unit that estimates AoD, which is the direction of transmission of radio waves, using the third signal, and an AoA estimation unit that estimates AoA, which is the direction of arrival of radio waves, using the fourth signal.

[0007] According to an embodiment of the present invention, it is possible to improve the estimation accuracy of an estimation device that estimates a propagation path of a radio wave using CSI.

[0008] It is a diagram showing an example of the configuration of an estimation device according to the present embodiment.It is a flowchart showing an example of estimation processing according to the present embodiment.It is a diagram showing an example of the hardware configuration of a computer.

[0009] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. The embodiment described below is merely an example, and the embodiment to which the present invention is applied is not limited to the following embodiment.

[0010] (Regarding the Estimation Device) The estimation device according to this embodiment is a device that estimates propagation path information of radio waves reflected from a person and estimates the position of the person by using CSI, which represents changes in amplitude and phase of Wi-Fi radio waves due to propagation. Here, before describing the estimation device according to this embodiment, an overview of conventional estimation techniques will be described.

[0011] Recent advances in Wi-Fi radio wave communication technology have made it possible for commercially available Wi-Fi devices to utilize CSI (Channel State Information), which indicates changes in amplitude and phase due to propagation. Current Wi-Fi communications employ communication technologies such as MIMO (Multiple-input Multiple-output), which uses multiple transmitting and receiving antennas, and OFDM (Orthogonal Frequency Division Multiplexing), which transmits data on multiple subcarriers with different frequencies. These communication technologies allow CSI to be acquired for each Wi-Fi packet between each antenna and for each subcarrier.

[0012] Using this CSI, the following four types of information regarding the propagation path can be estimated. For example, the Doppler Velocity (DV), which is the rate of change of the propagation path length, can be estimated from the phase difference of the CSI between packets. The Angle of Departure (AoD), which is the direction of radio wave transmission, can be estimated from the phase difference between transmitting antennas. The Angle of Arrival (AoA), which is the direction of radio wave arrival, can be estimated from the phase difference between receiving antennas. The Time of Flight (ToF), which is the time of arrival of radio waves, can be estimated from the phase difference between subcarriers.

[0013] Furthermore, by estimating information about these propagation paths reflected from a person from the CSI, the person's position can be estimated. In order to extract the propagation paths reflected from a person, it is assumed that the person is moving, and a high-pass filter or the like is used to extract the dynamic components of the CSI that change over time. This makes it possible to separate static components, which are direct waves and radio waves reflected from furniture, walls, etc., from dynamic components, which are radio waves reflected from a person.

[0014] In the conventional methods of Non-Patent Document 1 and Non-Patent Document 2, AoD, AoA, DV, ToF, etc. are estimated from CSI, and human position estimation is realized using one transmitter and one receiver.

[0015] (Problems) Position estimation methods that use propagation path information have two problems. The first problem is that the CSI acquired by actual devices contains phase offsets, particularly CFO (Carrier Frequency Offset), making it difficult to extract information about each propagation path and information about propagation paths reflected by a person. The second problem is that when a person is present between the transmitter and receiver, the propagation path information becomes the same at all positions between the devices. Furthermore, when a person is present between the devices, their body blocks direct waves, reducing the estimation accuracy of the propagation path information and making position estimation difficult.

[0016] Here, CFO refers to a phase offset that occurs due to the oscillators in the transmitter and receiver being inaccurate and not synchronized between the transmitter and receiver. Because of this CFO, a random value is added to the CSI phase information for each packet, making it impossible to perform processing between packets, such as high-pass filtering, on the CSI phase.

[0017] To address this CFO problem, Non-Patent Document 1 and Non-Patent Document 2 remove the offset by performing complex conjugate multiplication on the CSI between the receiving antennas. Non-Patent Document 3 removes the offset by performing division by the CSI between the receiving antennas.

[0018] However, these conventional methods have a problem in that when complex conjugate multiplication or division of CSI is performed between receiving antennas, the estimation accuracy of the AoA estimated from the phase difference between the receiving antennas may decrease.

[0019] Regarding the second problem of difficulty in estimating a position between a transmitter and a receiver, Non-Patent Documents 1 and 2 perform position estimation using one transmitter and one receiver, but do not consider the case where a person is present between the transmitter and receiver. Non-Patent Documents 3 and 4 increase the number of links between the transmitter and receiver, for example, by using multiple receivers. As such, conventional technologies have not been able to solve the problem of difficulty in estimating a person's position using one transmitter and one receiver when a person is present between the transmitter and receiver.

[0020] In order to solve such a problem, the estimation device according to this embodiment has, for example, a configuration as shown in FIG.

[0021] <Configuration of Estimation Apparatus> Fig. 1 is a diagram showing an example of the configuration of an estimation apparatus according to this embodiment. As shown in Fig. 1, the estimation apparatus 100 includes, for example, an input unit 110, a dynamic component extraction unit 120, a propagation path information estimation unit 130, a position tracking unit 140, and an output unit 150.

[0022] The input unit 110 is an interface for inputting CSI acquired using multiple transmitting antennas and multiple receiving antennas. For example, the input unit 110 receives CSI acquired by a Wi-Fi transceiver installed in the environment to be estimated.

[0023] (Dynamic Component Extraction Unit 120) The dynamic component extraction unit 120 includes, for example, a first offset removal unit 121, a second offset removal unit 122, high-pass filters 123 and 124, low-pass filters 125 and 126, a first phase difference correction unit 127, and a second phase difference correction unit 128.

[0024] The first offset removal unit 121 performs a first offset removal process of dividing the CSI between the receiving antennas and generating a first signal.

[0025] The second offset removal unit 122 performs a second offset removal process to divide the CSI between the transmitting antennas and generate a second signal. Since the CFO (offset) is a common value for each packet, the CFOs of the CSI of a certain packet are canceled out by dividing the CSI between the antennas, and the CFO can be removed.

[0026] The high-pass filter 123 extracts the dynamic component from the first signal generated by the first offset removal unit 121. The high-pass filter 124 extracts the dynamic component from the second signal generated by the second offset removal unit 122.

[0027] The low-pass filter 125 extracts static components from the first signal generated by the first offset removal unit 121. The low-pass filter 126 extracts static components from the second signal generated by the second offset removal unit 122.

[0028] The first phase difference correction unit 127 performs a first phase difference correction process in which the phase of the dynamic component extracted from the first signal by the high-pass filter 123 is corrected using the static component extracted from the second signal by the low-pass filter 126, thereby generating a third signal.

[0029] The second phase difference correction unit 128 performs a second phase difference correction process in which the phase of the dynamic component extracted from the second signal by the high-pass filter 124 is corrected using the static component extracted from the first signal by the low-pass filter 125, thereby generating a fourth signal.

[0030] With the above configuration, the dynamic component extraction unit 120 removes the CFO by dividing the CSI between the antennas, extracts only the dynamic component using a high-pass filter, and corrects the phase of the dynamic component using the phase difference of the static component. At this time, division is performed between the receiving antennas to estimate AoD, and division is performed between the transmitting antennas to estimate AoA. Therefore, in this embodiment, the Wi-Fi transceiver requires two or more (multiple) antennas.

[0031] The dynamic component extraction process executed by the dynamic component extraction unit 120 will be described in detail later.

[0032] (Propagation Path Information Estimation Unit) The propagation path information estimation unit 130 includes, for example, an AoD estimation unit 131, an AoA estimation unit 132, and a DV estimation unit 133.

[0033] The AoD estimation unit 131 estimates the AoD using a third signal obtained by correcting the phase of the dynamic component extracted from the first signal by the first phase difference correction unit 127 using the static component extracted from the second signal. For example, the third signal obtained by correcting the phase of the first signal divided between the receiving antennas has a size of the number of packets multiplied by the number of transmitting antennas. The AoD estimation unit 131 estimates the AoD from the phase difference between the transmitting antennas of the third signal.

[0034] The AoA estimation unit 132 estimates the AoA using a fourth signal obtained by correcting the phase of the dynamic component extracted from the second signal by the second phase difference correction unit 128 using the static component extracted from the first signal. For example, the fourth signal obtained by correcting the phase of the second signal divided between the transmitting antennas has a size of the number of packets multiplied by the number of receiving antennas. The AoA estimation unit 132 estimates the AoA from the phase difference between the receiving antennas of the fourth signal.

[0035] The DV estimator 133 performs a DV estimation process to estimate the DV, which is the rate of change of the propagation path length, using the third signal generated by the first phase difference corrector 127 and / or the fourth signal generated by the second phase difference corrector 128. The DV can be estimated from either the dynamic component due to division between the transmitting antennas or the dynamic component due to division between the receiving antennas. The DV estimator 133 estimates the DV, for example, using the average value of the third signal and the fourth signal.

[0036] To estimate AoD, AoA, and DV, for example, the well-known MUSIC (Multiple Signal Classification) method or SAGE (Sparse Recovery, Space-alternating Generalized Expectation-maximization) method can be used, which can estimate parameters even when the number of antennas is small or the packet acquisition interval is long.

[0037] With the above configuration, the propagation path information estimating unit 130 estimates the AoD, AoA, and DV of the propagation path reflected from a moving person from the dynamic components of the CSI extracted by the dynamic component extracting unit 120.

[0038] (Position Tracking Unit) The position tracking unit 140 includes, for example, a weighting unit 141, a resampling unit 142, a position estimation unit 143, a particle movement unit 144, and the like.

[0039] The weighting unit 141 weights each particle using the AoD, AoA, and DV estimated by the propagation path information estimating unit 130. The resampling unit 142 performs resampling according to the weight of each particle. After resampling, the position estimating unit 143 outputs the average value of the coordinates of each particle at time t-F as an estimated position. The particle moving unit 144 moves the coordinates of each particle according to a normal distribution.

[0040] With the above configuration, the weighting unit 141 generates multiple particles and repeatedly weights the particles, resampling, estimates their positions, and moves the particles to track the position of a person using a particle filter. In this embodiment, the weighting unit 141 increases the variance of the distribution used to weight particles that are close to devices that have acquired CSI, and smooths the movement trajectory using fixed delay smoothing.

[0041] Specifically, the position tracking unit 140 first generates and initializes a large number of particles. In this embodiment, since the position is estimated on a two-dimensional plane, the particle state is represented by two-dimensional coordinates that represent the position of each particle.

[0042] Next, the weighting unit 141 weights each particle using the estimated AoD, AoA, and DV. For example, the weighting unit 141 calculates the AoD and AoA of the propagation path when reflected by the particle, and the propagation path length, based on the coordinates of each particle and the positions and orientations of the transmitting antenna array and the receiving antenna array. The weighting unit 141 also calculates the DV of the particle by calculating the difference from the propagation path length of the particle at the previous time. Furthermore, the weighting unit 141 calculates the likelihood of the AoD, AoA, and DV calculated from the position of each particle using a normal distribution with the AoD, AoA, and DV estimated from the CSI as the mean, and uses the sum of these as the weight.

[0043] If a human body is present within the first Fresnel zone formed by the transmitter and receiver, the parameter estimation accuracy is thought to decrease. Also, the AoD and AoA are the same at all positions between the transmitter and receiver. Therefore, the weighting unit 141 increases the variance of the normal distribution used for weighting the AoD and AoA when weighting particles near the devices, using the distance obtained by adding the width of the human body to the first Fresnel zone as the boundary.

[0044] Next, the resampling unit 142 performs resampling according to the weight of each particle. For example, the resampling unit 142 randomly selects particles by probabilistic selection such that particles with large weights are more likely to be selected. At this time, the resampling unit 142 resamples particles at time t-F using the weight at time t through fixed delay smoothing. Particles from time t-F to t are retained, and the history of these particles is resampled using the weight at time t. As a result, even when position estimation between a transmitter and a receiver is difficult, the movement trajectory is smoothed, and improvement in estimation accuracy can be expected.

[0045] After resampling, the position estimation unit 143 outputs the average value of the coordinates of each particle at time t−F as the estimated position. Thereafter, the position tracking unit 140 moves the coordinates of each particle according to a normal distribution and performs position tracking by repeating the weighting.

[0046] The configuration of the estimation device 100 shown in Fig. 1 is an example. For example, the functional components of the estimation device 100 shown in Fig. 1 may be distributed across multiple devices. Furthermore, the estimation device 100 may not include the position tracking unit 140 and may be a device that estimates propagation path information.

[0047] <About CSI> CSI is the transmission path response of radio wave propagation used in the physical layer of Wi-Fi communication. CSI expresses the change in amplitude and phase due to radio wave propagation as the absolute value and argument of a complex number, respectively. For example, in a multipath environment where multiple propagation paths exist, such as radio waves reflected by walls, the composite wave of each propagation path is observed as CSI. The path length of each propagation path is expressed as l ｉ Then, CSIh is expressed by the following equation (1).

[0048] where L is the number of propagation paths, a ｉ is the amplitude of each propagation path, f is the frequency of the radio wave, and c is the speed of light.

[0049] If the AoD of a certain propagation path is φ, the difference in propagation path length between transmitting antennas is d ｔ Similarly, if AoA is θ, the difference in propagation path length between receiving antennas can be expressed as d ｒ It can be expressed as sin θ, where d ｔ is the distance between the transmitting antennas, and d ｒ is the distance between the receiving antennas.

[0050] Therefore, the phase difference between the transmitting antennas is −2jπfd ｔ sinφ / c, the phase difference between the receiving antennas is -2jπfd ｒ It is sin θ / c.

[0051] Furthermore, if DV is v, the difference in propagation path length from time Δt later is vΔt, so the phase difference between them is -2jπfvΔt / c. If ToF is τ, the propagation path length can be expressed as τc. If the frequency difference between subcarriers is fδ, the phase difference between subcarriers is -2jπfδτ.

[0052] The AoD, AoA, DV, and ToF can be estimated from the phase differences between transmitting antennas, receiving antennas, packets, and subcarriers, for example, by the MUSIC method.

[0053] In real equipment, oscillators are not perfectly accurate and are not synchronized between transmitters and receivers, resulting in a phase difference between the transmitter and receiver. This phase offset is called CFO, and a different phase is added to the CSI for each packet. This CFO must be removed to extract the dynamic component.

[0054] <About Dynamic Component Extraction Processing> The dynamic component extraction unit 120 removes the CFO by dividing the CSI between antennas. At this time, the CSI is divided between the transmitting antennas and between the receiving antennas, and the division is used for AoA estimation and AoD estimation, respectively. Because the CFO is a common value for each packet, by dividing the CSI between antennas in the CSI of a certain packet, the CFOs cancel each other out, and the CFO can be removed.

[0055] If the direct wave and static components reflected by walls, furniture, etc. of the CSI are denoted as s, and the dynamic component reflected by a moving person is denoted as d, then the CSI can be expressed as h = s + d, and the reciprocal of h, 1 / h, can be transformed into the following equation (2).

[0056] Furthermore, the Maclaurin expansion of f(x)=1 / (1−x) is given by the following equation (3).

[0057] Therefore, if x=-(d / s), the reciprocal of h, 1 / h, can be expressed by the following equation (4).

[0058]

[0059] Here, since the direct wave has a greater intensity than the reflected wave and |d / s|<<1, ignoring k=2 and above, the reciprocal 1 / h of h can be approximated by the following equation (5).

[0060]

[0061] Here, the CSI obtained for packet p, transmit antenna m, receive antenna n, and subcarrier q is expressed as h ｐ，ｍ，ｎ，ｑ Then, the division of CSI between transmit antennas m0 and m1 for AoA estimation can be expressed by the following equation (6):

[0062]

[0063] Here, the first term in equation (6) is composed only of static components, and can be removed by removing the static components using a high-pass filter, etc. Furthermore, the fourth term in equation (7) is a multiplication of dynamic components, and is smaller in value than the second and third terms, so if it is ignored, the CSI from which only the dynamic components are extracted using a high-pass filter can be approximated as shown in the following equation (7).

[0064]

[0065] Here, HPF is a static component removal operation performed by applying a high-pass filter to packets (time). In conventional methods for removing CFO between receiving antennas in AoA estimation, the AoA estimation accuracy varies depending on the power ratio of static components between the selected antennas. However, in this embodiment, any antennas can be selected as transmitting antennas m0 and m1, regardless of the ratio of static components.

[0066] The AoA of the propagation path reflected from a person can be estimated from the phase difference between the receiving antennas of the dynamic component of the CSI. However, the right side of equation (7) is in the form of dividing the dynamic component by the static component, and a phase shift occurs by the amount of the static component from the phase of the target dynamic component. Therefore, by applying a low-pass filter to the phase difference between the receiving antennas of the CSI, the phase difference between the receiving antennas of the static component is extracted, and this is used to correct the phase on the left side of equation (7). Specifically, the difference Δ between the receiving antennas of the static component ｎｓｐ，ｍ１，ｑ The average value of the phase difference between adjacent receiving antennas is used as shown in the following equation (8).

[0067]

[0068] Here, N represents the number of receiving antennas, and LPF represents the static component extraction operation by applying a low-pass filter to the packet (time). The division of CSI in equation (8) can use the result of division between receiving antennas for AoD estimation. Alternatively, since the phase difference between receiving antennas for the static component is required, the division of CSI in equation (8) can be complex conjugate multiplication.

[0069] The dynamic component d′ input to the propagation path information estimation unit 130 ｐ，ｍ１,ｎ，ｑcan be expressed by the following equation (9).

[0070] Here, ∠ represents the argument of a complex number.

[0071] Similarly, in CFO removal by division between receiving antennas for AoD estimation, after the dynamic components are extracted using a high-pass filter, the phase is corrected using the static components of the phase difference between the transmitting antennas and the result is input to the propagation path information estimation unit 130.

[0072] <Processing Flow> Next, the processing flow of the estimation method according to this embodiment will be described.

[0073] 2 is a flowchart showing the flow of the estimation process according to this embodiment. This process shows an example of the estimation process executed by the estimation device 100 described with reference to FIG.

[0074] In step S201, when CSI acquired by a Wi-Fi transceiver installed in the environment to be estimated is input to the input unit 110, the estimation device 100 executes, for example, the processes from step S202 onwards.

[0075] In step S202, the first offset removal unit 121 divides the CSI between the receiving antennas to generate a first signal.

[0076] In step S203, the second offset removal unit 122 divides the CSI between the transmitting antennas to generate a second signal. Note that the process of step S202 and the process of step S203 may be executed in parallel, for example.

[0077] In step S204, the first phase difference correction unit 127 corrects the phase of the dynamic component extracted from the first signal by the high-pass filter 123 using the static component extracted from the second signal by the low-pass filter 126, thereby generating a third signal.

[0078] In step S205, the second phase difference correction unit 128 corrects the phase of the dynamic component extracted from the second signal by the high-pass filter 124 using the static component extracted from the first signal by the low-pass filter 125, thereby generating a fourth signal. Note that the processing of step S205 is executed in parallel with the processing of step S204, for example.

[0079] In step S206 , the AoD estimation unit 131 estimates the AoD using the third signal generated by the first phase difference correction unit 127 .

[0080] In step S207, the DV estimation unit 133 estimates the DV using the third signal and the fourth signal. For example, the DV estimation unit 133 estimates the DV using an average value of the third signal and the fourth signal.

[0081] In step S208, the AoA estimation unit 132 estimates the AoA using the fourth signal generated by the second phase difference correction unit 128. Note that the processes of steps S206, S207, and S208 are executed in parallel, for example.

[0082] By the above processing, according to this embodiment, it is possible to improve the estimation accuracy of the estimation device 100 that estimates the propagation path of radio waves using CSI.

[0083] In step S209, the position tracking unit 140 tracks the position of the person using a particle filter, using the AoD, AoA, and DV estimated by the propagation path information estimating unit 130. Preferably, the position tracking unit 140 increases the variance of the distribution used to weight particles close to the transmitter / receiver from which the CSI was acquired, and smooths the movement trajectory using fixed delay smoothing. As a result, according to this embodiment, when a person is present between the transmitter / receiver, one transmitter / receiver is used (e.g., one transmitter / receiver and one receiver / receiver), and the person's position can be easily estimated even when there is a person between the transmitter / receiver and the receiver.

[0084] In step S210, the output unit 150 outputs the position (estimated position) of the person tracked by the position tracking unit 140.

[0085] The processing in FIG. 2 can improve the estimation accuracy of the estimation device 100 that estimates the propagation path of radio waves using CSI.

[0086] <Hardware Configuration> The estimation device 100 according to this embodiment has, for example, the hardware configuration of a computer 300 as shown in Fig. 3, and realizes each functional configuration as shown in Fig. 1 by executing a predetermined program on the computer 300. Alternatively, the estimation device 100 in Fig. 1 may be realized by, for example, a digital signal processor (DSP), a programmable logic device (PLD), or a field programmable gate array (FPGA).

[0087] 3 is a diagram showing an example of the hardware configuration of a computer 300. In the example of Fig. 3, the computer 300 includes a processor 1001, a memory 1002, a storage device 1003, a communication device 1004, an input device 1005, an output device 1006, a bus B, etc.

[0088] The processor 1001 is, for example, an arithmetic device such as a CPU (Central Processing Unit) that realizes various functions by executing a predetermined program. The memory 1002 is a storage medium readable by the computer 300 and includes, for example, a RAM (Random Access Memory) and a ROM (Read Only Memory). The storage device 1003 is a computer-readable storage medium and may include, for example, a HDD (Hard Disk Drive), an SSD (Solid State Drive), various optical disks, and magneto-optical disks.

[0089] The communication device 1004 includes one or more pieces of hardware (transmitting / receiving devices) for communicating with other devices via a wireless or wired network. The input device 1005 is an input device (e.g., a keyboard, a mouse, a microphone, a switch, a button, a sensor, etc.) that accepts input from the outside. The output device 1006 is an output device (e.g., a display, a speaker, an LED lamp, etc.) that outputs to the outside.

[0090] The bus B is commonly connected to the above components and transmits, for example, address signals, data signals, and various control signals. The processor 1001 may include, in addition to (or instead of) a CPU, for example, a DSP (Digital Signal Processor), a PLD (Programmable Logic Device), or an FPGA (Field Programmable Gate Array).

[0091] The above program may be for realizing some of the above-described functions, or may be capable of realizing the above-described functions in combination with a program already recorded in the computer 300. Furthermore, some or all of the functional components of the estimation device 100 may be realized using hardware such as a programmable logic device (PLD) or a field programmable gate array (FPGA).

[0092] <Effects of the embodiment> According to the present embodiment, it is possible to improve the estimation accuracy of the estimation device 100 that estimates the propagation path of radio waves using CSI. Furthermore, by estimating the AoA, AoD, and DV from the CSI and performing position tracking using a particle filter, it is possible to realize position estimation even when there is one transmitter and one receiver of Wi-Fi radio waves in the environment.

[0093] In this embodiment, as a method for removing a phase offset of CSI in AoD, AoA, and DV estimation, CSI is divided between receive antennas in AoD, and CSI is divided between transmit antennas in AoA estimation. Then, phase correction is performed using static components on the dynamic components of CSI after offset removal, thereby achieving highly accurate AoD and AoA estimation.

[0094] Furthermore, in this embodiment, in order to address the decrease in position estimation accuracy between devices, the variance of the distribution used to weight the particle filter between the transmitter and receiver is increased, and the movement trajectory is smoothed using fixed delay smoothing, thereby improving position estimation accuracy.

[0095] Summary of Embodiments This specification discloses at least the following estimation devices and estimation methods: (Item 1) An estimation device comprising: an input unit that inputs CSI acquired using multiple receiving antennas and multiple transmitting antennas, a first offset removal unit that divides the CSI between the receiving antennas to generate a first signal, a second offset removal unit that divides the CSI between the transmitting antennas to generate a second signal, a first phase difference correction unit that corrects the phase of a dynamic component extracted from the first signal using a static component extracted from the second signal to generate a third signal, a second phase difference correction unit that corrects the phase of the dynamic component extracted from the second signal using the static component extracted from the first signal to generate a fourth signal, an AoD estimation unit that estimates AoD, which is the direction of transmission of radio waves, using the third signal, and an AoA estimation unit that estimates AoA, which is the direction of arrival of radio waves, using the fourth signal. (2) The estimation device according to paragraph 1, comprising: a DV estimation unit that estimates a DV, which is a rate of change of a propagation path length, using the third signal and / or the fourth signal; and a position tracking unit that tracks a position of a person by a particle filter using the AoD, the AoA, and the DV. (3) The estimation device according to paragraph 2, wherein the position tracking unit increases the variance of a distribution used to weight particles close to devices that have acquired the CSI, and smooths a movement trajectory by fixed delay smoothing.(Clause 4) An estimation method comprising: a step of inputting CSI acquired using a plurality of receiving antennas and a plurality of transmitting antennas to an input unit; a step by a first offset removal unit dividing the CSI between the receiving antennas to generate a first signal; a step by a second offset removal unit dividing the CSI between the transmitting antennas to generate a second signal; a step by a first phase difference correction unit correcting the phase of a dynamic component extracted from the first signal using a static component extracted from the second signal to generate a third signal; a step by a second phase difference correction unit correcting the phase of the dynamic component extracted from the second signal using the static component extracted from the first signal to generate a fourth signal; a step by an AoD estimation unit estimating AoD, which is the direction of transmission of radio waves, using the third signal; and a step by an AoA estimation unit estimating AoA, which is the direction of arrival of radio waves, using the fourth signal.

[0096] Although the present embodiment has been described above, the present invention is not limited to such a specific embodiment, and various modifications and changes are possible within the scope of the gist of the present invention described in the claims.

[0097] REFERENCE SIGNS LIST 100 Estimation device 110 Input unit 120 Dynamic component extraction unit 121 First offset removal unit 122 Second offset removal unit 127 First phase difference correction unit 128 Second phase difference correction unit 130 Propagation path information estimation unit 131 AoD estimation unit 132 AoA estimation unit 133 DV estimation unit 140 Position tracking unit 150 Output unit

Claims

1. An estimation device having: an input unit that inputs CSI acquired using multiple receiving antennas and multiple transmitting antennas; a first offset removal unit that divides the CSI between the receiving antennas to generate a first signal; a second offset removal unit that divides the CSI between the transmitting antennas to generate a second signal; a first phase difference correction unit that corrects the phase of a dynamic component extracted from the first signal using a static component extracted from the second signal to generate a third signal; a second phase difference correction unit that corrects the phase of the dynamic component extracted from the second signal using the static component extracted from the first signal to generate a fourth signal; an AoD estimation unit that uses the third signal to estimate AoD, which is the direction of transmission of radio waves; and an AoA estimation unit that uses the fourth signal to estimate AoA, which is the direction of arrival of radio waves.

2. The estimation device of claim 1, comprising: a DV estimation unit that estimates DV, which is the rate of change of the propagation path length, using the third signal and / or the fourth signal; and a position tracking unit that tracks the position of a person using a particle filter, using the AoD, the AoA, and the DV.

3. The estimation device according to claim 2, wherein the position tracking unit increases the variance of the distribution used to weight particles close to the devices that acquired the CSI, and smooths the movement trajectory using fixed delay smoothing.

4. An estimation method comprising: a step of inputting CSI acquired using multiple receiving antennas and multiple transmitting antennas to an input unit; a step by a first offset removal unit dividing the CSI between the receiving antennas to generate a first signal; a step by a second offset removal unit dividing the CSI between the transmitting antennas to generate a second signal; a step by a first phase difference correction unit correcting the phase of a dynamic component extracted from the first signal using a static component extracted from the second signal to generate a third signal; a step by a second phase difference correction unit correcting the phase of the dynamic component extracted from the second signal using the static component extracted from the first signal to generate a fourth signal; a step by an AoD estimation unit estimating AoD, which is the direction of transmission of radio waves, using the third signal; and a step by an AoA estimation unit estimating AoA, which is the direction of arrival of radio waves, using the fourth signal.

Images