Apparatus, method, and program for estimating the number of targets

Abstract

This invention provides a target number estimation device that can improve the accuracy of target number estimation through mathematical optimization using a single sensor type. [Solution] According to the embodiment, the target number estimation device comprises a data acquisition unit and a target number estimation unit. The data acquisition unit acquires a first signal containing information about pairs of a first region, which is a division of a predetermined area, and the target number, and a second signal containing information about pairs of a second region, which is a division of a predetermined area, and the target number. The target number estimation unit generates a third signal containing information about pairs of a third region, which is a division of a predetermined area, and the target number, by applying mathematical optimization processing to the target number for each first region included in the first signal and the target number for each second region included in the second signal. Both the first signal and the second signal are signals generated based on the same first wireless system. The first signal is based on first information of the first wireless system. The second signal is based on second information of the first wireless system.

Description

Technical Field

[0001] Embodiments of the present invention relate to an object number estimation device, method, and program.

Background Art

[0002] There is a technique for estimating the number of objects (e.g., the number of people) for each area by performing mathematical optimization processing on the number of objects (e.g., the number of people) counted within their respective measurement ranges by a plurality of types of sensors installed in a building (Non-Patent Document 1).

[0003] Mathematical optimization is an approach to mathematically solve an optimization problem. More specifically, it is a method of obtaining a solution to minimize or maximize a specific objective function under given constraints using mathematical techniques and properties.

[0004] In object number estimation technology, when there are multiple sensor types, that is, in a multi-sensor configuration, one or more sensors of each sensor type are installed in a target space. Each sensor observes the number of objects in an area defined as its observation range. Then, for all sensor types, information indicating the relationship between the area and the observed number of objects is input to the object number estimation device.

[0005] In object number estimation, for the area for which the number of objects is to be estimated, the observed number of objects in the overlapping part of the area of each sensor of each sensor type is given as a variable, and an equation indicating the relationship between the observed number of objects of each sensor and the variable is established. At this time, the error of each sensor is also given as a variable. By solving the system of simultaneous equations consisting of the equations for the total number of sensors so as to minimize the variable of the error, the object number estimation device can derive the estimated number of objects in the area for which the number of objects is to be estimated without learning even when there is an excess or deficiency in the sensor information.

[0006] This object number estimation technology can improve the accuracy of the number of objects in the area to be estimated by various types of sensor information.

Prior Art Documents

Non-Patent Documents

[0007] [Non-Patent Document 1] K. Kondo et al., “Equation-based modeling and optimization-based parameter estimation in multimodal virtual sensing platforms for smart buildings,” Build.Environ.,Vol.243,110620,2023 [Overview of the project] [Problems that the invention aims to solve]

[0008] However, the conventional object number estimation techniques using mathematical optimization described above assume the presence of multiple sensor types, i.e., a multi-sensor configuration. In contrast, in practical applications using existing sensors, only one type of sensor, such as a wireless LAN (Local Area Network), may be available. When the number of observed objects input to the mathematical optimization process is limited to only one sensor type, the advantages of performing mathematical optimization are lost.

[0009] One embodiment of the present invention provides a target number estimation device, method, and program that can improve the accuracy of target number estimation by mathematical optimization using a single sensor type. [Means for solving the problem]

[0010] According to the embodiment, the target number estimation device comprises a data acquisition unit and a target number estimation unit. The data acquisition unit acquires a first signal containing information about pairs of first regions, which are divisions of a predetermined geographical area, and the number of targets corresponding to the first regions, for the number of first regions, and a second signal containing information about pairs of second regions, which are divisions of a predetermined geographical area, and the number of targets corresponding to the second regions, for the number of second regions. The target number estimation unit generates a third signal containing information about pairs of third regions, which are divisions of a predetermined geographical area, and the number of targets corresponding to the third region, for the number of third regions, by applying mathematical optimization processing to the number of targets for each first region included in the first signal and the number of targets for each second region included in the second signal. Both the first signal and the second signal are signals generated based on the same first wireless system. The first signal is based on first information of the first wireless system. The second signal is based on second information of the first wireless system. [Brief explanation of the drawing]

[0011] [Figure 1] A figure showing an example of an application of the target number estimation device according to the embodiment. [Figure 2] A diagram showing one example configuration of the target number estimation device according to the embodiment. [Figure 3] A diagram illustrating a method (Method 1) for determining the location of STAs in a wireless LAN, which may be applied to the target number estimation device of the embodiment. [Figure 4] A diagram showing an example of an AP (Access Point) installation on a wireless LAN MLD. [Figure 5] A diagram illustrating a method (Method 2) for determining the location of STAs in a wireless LAN, which may be applied to the target number estimation device of the embodiment. [Figure 6] A diagram illustrating a method (Method 3) for determining the location of STAs in a wireless LAN, which may be applied to the target number estimation device of the embodiment. [Figure 7] A diagram illustrating a method (Method 4) for determining the location of STAs in a wireless LAN, which may be applied to the target number estimation device of the embodiment. [Figure 8]A diagram illustrating a method (Method 4) for determining the location of STAs in a wireless LAN, which may be applied to the target number estimation device of the embodiment. [Figure 9] This figure schematically illustrates how the target number estimation device of the embodiment improves the accuracy of target number estimation through mathematical optimization. [Figure 10] A flowchart illustrating the processing procedure of the target number estimation device according to the embodiment. [Modes for carrying out the invention]

[0012] The embodiments will be described below with reference to the drawings. Figure 1 shows an example of an application of the target number estimation device 4 according to the embodiment.

[0013] The object number estimation device 4 is a device that can derive the number of observed objects from different perspectives, as if multiple sensor types existed, even though only one sensor type exists, and determine the number of objects in the target area through mathematical optimization.

[0014] The number of people can be expressed as the number of people (the number of people in a given area can also be expressed as the number of people present), or it can be the number of specific devices, etc. In this embodiment, we will proceed with the explanation using the number of people present.

[0015] In this embodiment, a wireless sensor is given as an example of a single sensor type. While the explanation will focus on wireless LAN, sensors using other wireless communication methods such as Bluetooth® (including Bluetooth Low Energy) or cellular are also acceptable. Figure 1 shows an example using wireless LAN.

[0016] Here, the wireless LAN uses, for example, the IEEE 802.11 standard (including extended standards such as 11a / b / g / n / ac / ax / be, etc.), or has obtained Wi-Fi certification (including Wi-Fi CERTIFIED 4 / 5 / 6 / 7, etc.) from the Wi-Fi Alliance. In the IEEE 802.11 standard, STA (station) is usually used in a concept that includes AP (access point). In the case of STA excluding AP, it is expressed as non-AP STA, but in this embodiment, unless otherwise noted, it is used as a concept excluding AP. Note that STA is also expressed as a client.

[0017] As shown in FIG. 1, the target number estimation device 4 is connected to a wireless LAN controller (WLC) 1. The WLC 1 is a device that integrally controls a plurality of access points (APs) 2. The WLC 1 collects various data from each of the plurality of APs 2 and provides it to the target number estimation device 4. In other words, the target number estimation device 4 acquires data related to the estimation of the target number of the plurality of APs 2 via the WLC 1.

[0018] The AP 2 is a device that performs wireless communication to accommodate terminals (STA) 3 existing within a certain range in the wireless LAN. The plurality of APs 2 are arranged to be scattered in a predetermined geographical area such as an office building. That is, a wireless service area is formed within the predetermined geographical area by the plurality of APs 2. The plurality of APs 2 are basically fixedly arranged.

[0019] STA 3 is a device such as a PC (personal computer) or smart phone used within a predetermined geographical area such as the aforementioned office building. STA 3 has a wireless communication function and can be connected to the Internet, etc. by performing wireless communication with the AP 2.

[0020] The target number estimation device 4 acquires data from the plurality of APs 2 via the WLC 1 and uses the data to estimate the number of STA 3 (the people using STA 3).

[0021] Figure 2 shows an example configuration of the target number estimation device 4.

[0022] The target number estimation device 4 includes a CPU (central processing unit) 11, main memory 12, communication device 13, input device 14, display device 15, and storage device 16.

[0023] The CPU 11 is a processor that loads and executes firmware (program) 200 stored in the storage device 16 into the main memory 12. By executing firmware 200, the CPU 11 can operate as a data acquisition unit 101 and as a target number estimation unit 102. The data acquisition unit 101 performs processing to acquire data related to target number estimation. The target number estimation unit 102 uses the data acquired by the data acquisition unit 101 to perform processing to estimate the number of people in a predetermined geographic area. Mathematical optimization is used for the estimation process. Note that the data acquisition unit 101 and the target number estimation unit 102 may be implemented as hardware, such as an electrical circuit.

[0024] Main memory 12 is a volatile storage device such as DRAM (dynamic RAM [random access memory]). Main memory 12 stores programs and data as a workspace for the CPU 11.

[0025] The communication device 13 is a device that performs wired or wireless communication with the WLC1. The input device 14 is a device that handles the input side of a user interface, such as a keyboard or pointing device. The display device 15 is a device that handles the output side of a user interface, such as an LCD (liquid crystal display). The display device 15 displays, for example, the target number estimation result. In addition to being displayed, the target number estimation result can be output as data to other devices, storage, the cloud, etc.

[0026] The storage device 16 is a non-volatile storage device such as an HDD (hard disk drive) or an SSD (solid state drive). As mentioned above, the storage device 16 stores the firmware 200 executed by the CPU 11. The storage device 16 may also store data of the target number estimation results.

[0027] One method for detecting the presence or absence of people using wireless LAN is, for example, in an office where laptops are connected to the wireless LAN for work, a method for detecting laptops can be used. Furthermore, since the location of the laptops can indicate the approximate location of people, this can be used to determine the number of people present in a certain area.

[0028] In this case, the laptop connects to the wireless LAN AP2 as wireless LAN STA3. AP2 and STA3 communicate wirelessly, so the location of STA(laptop)3, i.e., the location of the person, can be determined by AP2 receiving the wireless signal transmitted by STA3, or by STA3 receiving the wireless signal transmitted by AP2. Alternatively, STA3 may transmit a wireless signal before connecting to AP2, and the location of STA(laptop)3, i.e., the location of the person, can be determined using that signal.

[0029] For example, methods include using the Receive Signal Strength Indicator (RSSI) of the radio signal, or the difference between the Time of Departure (TOD) and the Time of Arrival (TOA) of the radio signal. The former utilizes the property that radio signals attenuate with distance (however, the amount of attenuation varies depending on the transmission path environment), while the latter utilizes the time required for the radio signal to propagate through space. Signal-to-Noise Ratio (SNR) or other metrics may be used instead of RSSI. When determining propagation time from the difference between ToD and ToA, in wireless LANs, the prevailing method involves the STA3 (including AP2 in this case) on the side attempting to acquire the time difference transmitting a Management frame, and the receiving STA3 (including AP2 in this case) receiving the Control frame, which is a response frame (Acknowledgement (Ack) frame in IEEE 802.11 wireless LANs), transmitting a Control frame after a fixed time has elapsed since receiving the frame. The Round Trip Time (RTT) is then measured by adding the spatial occupancy time related to the frame length of each frame, the fixed time before the transmission of the Control frame, and twice the time required for spatial propagation (due to round-trip frame exchange). The propagation distance can be determined by multiplying the speed of light by the propagation time.

[0030] Another method involves using wireless LAN to detect the location of the person themselves, rather than the STA (notebook PC) 3. This method utilizes wireless LAN signals for sensing, sending and receiving wireless LAN signals between AP2-AP2 or AP2-STA3, and obtaining the transmission path status, i.e., CSI (Channel State Information), at the receiving end to determine whether a person is present or absent on the transmission path.

[0031] There are several methods for determining the location of STA(person)3, such as the following. The target number estimation device 4 (target number estimation unit 102) can input each of these separately into the mathematical optimization process. That is, it can be treated as if it were input to the mathematical optimization process by a different type of sensor. Within each method, there may be multiple methods rather than just one, and these can each be input to the mathematical optimization process as separate methods. In addition, some methods may be merged and input to the mathematical optimization process. The methods for determining the location of multiple STA(person)3 are performed at a certain period, and the input to the mathematical optimization process is performed simultaneously. The input to the mathematical optimization process is time-series data. In each location determination method, a certain period within a certain period (up to a period equal to one cycle) may be observed, and statistical processing may be performed to generate data. For example, if the same STA3 is observed during a certain period, the count of that STA3 is treated as 1 (no duplicate counting).

[0032] Information about the connected STA3 acquired by each AP2, as well as observational information of the wireless signals received from each STA3 (it is possible to receive wireless signals even if there is no connection), are collected by WLC1. The observational information of the wireless signals received from each STA3 includes the MAC address of the STA3 that transmitted the wireless signal, the IP address assigned at the IP layer, user information used for authentication when connecting to the wireless LAN, the RSSI of the wireless signal from the STA3 (which can be expressed as a time average due to smoothing in the implementation), and the SNR (the noise power used to calculate the SNR may be the average power when there is no channel reservation, i.e., NAV (Network Allocation Vector: a mechanism for virtual carrier sensing in IEEE802.11 wireless LAN, or the period during which it is activated; used here in the latter sense) is not set and no transmission or reception is taking place, or a fixed value may be used). The target number estimation device 4 (data acquisition unit 101) can acquire this information via WLC1.

[0033] (Method 1: Count STAs in the area of ​​the AP to which the STA is connected) As shown in Figure 3, multiple APs (AP1 to AP3) are located in the target space. The observation range for each AP is defined. Since the observation range of each AP affects the output power and propagation of wireless signals, it is desirable to define it taking into account the environment in which it is located. The mathematical optimization process will require inputting multiple pairs of people present in each AP's observation range.

[0034] The output power could be, for example, the maximum output power, or the output power when transmitting a beacon frame. If the radio signal propagation environment is similar, an AP with higher output power will be able to transmit radio signals further than an AP with lower output power. Therefore, it is desirable that the observation range of an AP with higher output power be wider than that of an AP with lower output power.

[0035] A beacon frame is a management frame that is scheduled to be sent periodically from an AP. It broadcasts synchronization information, a Service Set Identifier (SSID), and capability information such as optional functions supported by the AP. The STA confirms the presence of APs based on beacon frames and other information from APs. If multiple APs are confirmed, the STA connects to the AP with the highest RSSI (Relationship Speed ​​Index) among them. Here, "connected" means that the STA can send and receive data frames with the AP, and this state is achieved after the authentication process, association process, and security settings (if necessary) have been completed as part of the connection process.

[0036] Wireless signals are fundamentally attenuated by the distance between the transmitter and receiver. In a line-of-sight (LOS) environment, attenuation occurs according to the square law of distance. In a non-line-of-sight (NLOS) environment, attenuation occurs according to the N-th power law (N>2) of distance, and the propagation loss coefficient N depends on the propagation environment. For example, when a wireless signal reaches the receiver by reflecting off ceilings, floors, walls, etc., the propagation loss coefficient N changes due to multipath. Also, if there is a wall between the transmitter and receiver and the wireless signal passes through that wall, the transmittance changes depending on the wall material, frequency, and angle of incidence, causing attenuation. Therefore, if the output power of two access points (APs) is the same, it is desirable for APs with more line-of-sight environments in the surrounding area to have a wider observation range.

[0037] In Method 1, if one STA is connected to an AP, the number of people present within the observation range of that AP is counted as one. In Figure 3, since STA1 is connected to AP1, the number of people present is counted as one within the observation range of AP1. If the number of people present within the observation range of AP1 has already been tallied, one is added for STA1. In Figure 3, since STA1 is not connected to AP2 or AP3, the number of people corresponding to STA1 within the observation ranges of AP2 and AP3 is not counted.

[0038] Normally, an STA connects to a maximum of one AP. However, IEEE 802.11be has a feature called Multi-Link Operation (MLO), and when this feature is supported, a single chassis can have multiple STAs or multiple APs. A device with multiple STAs is called a non-AP MLD (Multi-Link Device), and a device with multiple APs is called an AP MLD. Non-AP MLDs and AP MLDs can connect to each other. In this case, an AP under an AP MLD and an STA under a non-AP MLD communicate one-to-one over a certain frequency link (frequency channel) (e.g., Figure 4(a)). Up to the number of such frequency links can be used for communication, as long as there are pairs of APs and STAs that can be formed on different frequencies. Each AP in an AP MLD and each STA in a non-AP MLD have their own unique MAC (Medium Access Control) address, and each AP MLD and non-AP MLD also have its own MAC address (called the MLD MAC address). In this connection configuration, the AP observation range should be replaced with the AP MLD observation range, and the counts corresponding to STA should be replaced with the corresponding counts for non-AP MLD units.

[0039] As a wireless LAN, it is conceivable that the APs in an AP MLD may not necessarily be in the same enclosure (e.g., Figure 4(b)). If we call an AP MLD with such APs a non-co-located AP MLD, then the aforementioned AP MLD where all APs are in the same enclosure becomes a co-located AP MLD. In the case of a non-co-located AP MLD, one or more APs in the same enclosure, i.e., the same location, are treated as a single unit, and the observation range of each virtual AP is defined. When an STA MLD connects to a non-co-located AP MLD, if one or more STAs under the STA MLD connect to one or more APs in the same location under the non-co-located AP MLD, the number of people present should be counted as one within the observation range of the virtual AP composed of those APs in the same location. On the other hand, if STA11 under STA MLD1 is connected to AP11 under non-co-located AP MLD1, and STA12 under STA MLD1 is connected to AP12 under non-co-located AP MLD1, and AP11 and AP12 are not in the same enclosure but are installed separately, then the count for STA should be divided among the observation ranges of each virtual AP. For example, half a person should be counted for each virtual observation range: AP11's virtual observation range (including other APs in the same location as AP11 under non-co-located AP MLD1) and AP12's virtual observation range (including other APs in the same location as AP12 under non-co-located AP MLD1). This method divides one person by the number of virtual APs in question and allocates it to the observation range of each virtual AP. Note that if the frequency channel width or the traffic capacity allocated to the frequency link differs, the count may be apportioned according to the ratio of the channel width or the ratio of the traffic capacity.

[0040] (Method 2: Number of surrounding STAs observed at each AP (regardless of whether they are connected or not)) Method 1 describes how to count the number of people present in the observation range of an AP when an STA is connected to that AP, and also describes how to count when the AP is MLD. In Method 2, regardless of whether the STA is connected to an AP, the STA observed by each AP is counted directly in the observation range of that AP.

[0041] In Figure 5, AP1 and AP3 are operating connected to STA via frequency channel 1, while AP2 is operating connected to STA via frequency channel 2.

[0042] Operating with an STA on a specific frequency channel means, in other words, that in an IEEE 802.11 wireless LAN, a Basic Service Set (BSS) is configured on that frequency channel. All the STAs (including APs) that make up the BSS are synchronized. A BSS configured by APs is called an infrastructure BSS and is sometimes distinguished from an independent BSS which consists only of STAs without APs.

[0043] In Figure 5, AP1 is connected to two STAs, AP2 is connected to three STAs, and AP3 is connected to four STAs. In Method 2, each AP observes the radio signal from an STA even if no STA is connected, and identifies the source STA. Identifying the source STA means receiving the radio signal with the antenna, converting it to a baseband signal, decrypting the acquired physical packet, extracting the MAC frame, and saving the MAC address listed in the source address (Transmitter Address: TA) of the MAC frame as the observed STA. Identifying the observed STA helps avoid duplicate counting. Each AP also observes other frequency channels (channels other than the operational channel are also called off-channels). As a result, AP1 observes two STAs connected to its own AP on operational channel ch1, plus one STA connected to AP3 (assuming a situation where observing ch2 does not allow observation of the STA connected to AP2), observing a total of three devices (people). AP2 observes a total of 4 STAs (people) on operational channel ch2, including 3 STAs connected to its own AP and 1 STA connected to AP1 by observing ch 1. AP3 observes 4 STAs connected to its own AP on operational channel ch1 (assuming a situation where observing ch2 does not allow observation of STAs connected to AP2), and observes a total of 4 STAs (people) including other channels. As a result, in Method 2, the input to the mathematical optimization process is 3 people in the observation range of AP1, 4 people in the observation range of AP2, and 4 people in the observation range of AP3. In contrast, in the case of Method 1 described above, the input would be 2 people in the observation range of AP1, 3 people in the observation range of AP2, and 4 people in the observation range of AP3.

[0044] In Method 2, when observing STAs on different frequency channels, it is desirable to define the observation range of each AP to take into account propagation across those multiple frequency channels. Radio signals, being radio waves, undergo frequency-dependent attenuation. In LOS, if the frequency doubles, the received power becomes 1 / 4, and conversely, if the frequency is halved, the received power becomes 4. Therefore, lower frequency bands generally have less propagation attenuation of radio signals and can transmit over longer distances. The frequency bands used for wireless LANs include the 2.4GHz, 5GHz, and 6GHz bands. For example, if observation is limited to the 2.4GHz band as in Method 2, it is desirable to define a wider observation range for APs compared to when observation is limited to the 5GHz or 6GHz bands. When observing using all three bands (2.4GHz, 5GHz, and 6GHz), the observation range of APs may be defined to consider propagation in the wider 2.4GHz band, or the observation range of APs may be defined by considering the weight of each frequency band, based on which frequency band has the most STA observations.

[0045] The following describes a method for deriving the number of occupants from a different perspective, using the RSSI obtained when the AP receives the wireless signal transmitted by the STA.

[0046] Furthermore, although the following explanation uses RSSI, SNR may be used instead of RSSI, in which case it will be treated as if it were an input to mathematical optimization processing using a different sensor type. When replacing RSSI with SNR, you can replace the magnitude of RSSI with the magnitude of SNR, or replace the value greater than or equal to / less than RSSI with the value greater than or equal to / less than SNR. Also, the information is not limited to information acquired via a wireless LAN controller; as mentioned above, the propagation time of the wireless signal may be used, in which case it will also be treated as an input to mathematical optimization processing using a different sensor type. When replacing RSSI with the propagation time of the wireless signal, the RSSI tends to be larger and the propagation time tends to be shorter when the distance between AP and STA is short, so you can replace the magnitude of RSSI with the shortness or longness of the propagation time, or replace the value greater than or equal to / less than RSSI with the propagation time less than or equal to / greater than

[0047] (Method 3: Number of connections at each AP × RSSI) Method 3 is similar to Method 1, but as shown in Figure 6(a), when counting connected STAs, a threshold is set for the RSSI of the radio signal received from the STA, and only STAs with an RSSI above that threshold are counted. It is desirable that the radio signal from the STA does not have transmit power control. For example, the AP can observe radio signals that transmit management frames such as Probe Request frames, or other management frames that the AP sends to the STA for observation, and determine whether to count based on the RSSI. In Figure 6(a), it is shown as X dBm or higher, but in reality, a unique value is set. Ideally, it is desirable to set an RSSI threshold for each STA according to the STA's transmit power, but there are cases where the STA's transmit power cannot be determined, so a single value may be used provisionally. Also, if the device type of the STA can be determined, the RSSI threshold may be set according to the device type. For example, the threshold should be set taking into account that PCs tend to have high transmit power, while smartphones tend to have low transmit power. In Figure 6(a), since there is only one STA receiving a wireless signal with an RSSI above the set threshold, it is counted as one unit, whereas in Method 1 it would be counted as two units. It is appropriate to set the AP observation range to be smaller than in Method 1.

[0048] Method 3 may be input into the mathematical optimization process along with Method 1, treating it as if it were a different sensor type from Method 1.

[0049] Furthermore, along with Method 2, Method 3 may be input into the mathematical optimization process as if it were a different sensor type from Method 2.

[0050] In contrast to Figure 6(a) described above, Figure 6(b) may also be used, where counting occurs when the RSSI falls below a threshold. This method may also be treated as a separate method from Figure 6(a). When using a method like Figure 6(b), it is appropriate to define the AP observation range as one with an open center.

[0051] The above describes the processing of wireless signals received from the STA, but conversely, the STA observes wireless signals received from the AP and notifies the AP whether its RSSI exceeds a threshold (in this case, the AP counts the STA only if it is notified that the threshold has been exceeded, if it expects operation similar to Figure 6(a), or only if it is notified that the threshold has been exceeded, if it expects operation similar to Figure 6(b). The AP must notify the connected STA of the threshold in advance. Also, if it is possible to select either above or below the threshold, the connected STA must be notified of this instruction in advance). Alternatively, if the RSSI of the wireless signal received from the AP is notified to the AP (in this case, the AP considers the RSSI threshold to decide whether to count), the observed wireless signal can be observed on the AP side even if it is a downlink from the AP to the STA instead of an uplink from the STA to the AP. The AP may then forward the information from the STA directly to the wireless LAN controller, and the wireless LAN controller, or the target number estimation device, may ultimately determine the count based on the threshold and RSSI information via the wireless LAN controller. Here, it is desirable that the wireless signal from the AP observed by the STA is one that does not have transmit power control. For example, an STA can observe wireless signals, such as Beacon frames which are management frames from an AP, and send a notification to the AP according to its RSSI. When an STA observes wireless signals from an AP, the threshold used for the same AP is naturally common among multiple STAs. In a wireless LAN network consisting of multiple APs (Extended Service Set (ESS) in IEEE 802.11 wireless LAN), if the transmission power is set commonly across multiple APs, for example, for Beacon frames and frames that are the target of observation by the STA, the value used as the threshold for wireless signals from the APs can also be common.

[0052] The method of determining whether or not to count an STA by setting a threshold for the RSSI of the wireless signal from the STA may be incorporated into Method 2, which counts the number of STAs observed by the AP regardless of whether or not the STA is connected to the AP.

[0053] (Method 4: Rearrange the AP counts taking RSSI into account) Method 4 counts the STA at the AP that receives the most radio signals from the STA, regardless of whether the STA is connected or not, when multiple APs (APs) are observed as an STA. Figure 7 shows that although STA11 is connected to AP1, AP2 and AP3 also observe radio signals from the STA (for example, radio signals that send management frames such as Probe Request frames, as mentioned above, and radio signals that send other management frames that the APs send to the STA for observation), and the radio signal from the STA observed at AP2 is greater than those observed at AP1 and AP3, meaning that the radio signal from the STA observed at AP2 has the largest RSSI. In this case, STA11 is counted at AP2 (whereas in Method 1, it would be counted at AP1). Therefore, in Figure 7, which shows a situation where there is only one STA, STA11 is counted at AP2, and not at AP1 and AP3.

[0054] Furthermore, when observing an STA with multiple APs, including off-channels (on-channels refer to the channels that connect to the STA and send notifications, while off-channels refer to channels other than on-channels), the observation period becomes longer. In fact, while one AP may observe the state of the STA after it has moved, the observation information of the STA at other APs used for comparison may be from before the STA moved, making the comparison unsuitable. Therefore, for example, if the RSSI observed when observing an STA at a certain AP fluctuates significantly, it can be determined that the STA has moved. It is desirable to assume that the time when the RSSI fluctuation subsides represents the completed state of the move, and to compare it with the RSSI observed by other APs only after the completion of the move (in this case, it can be treated as if the observations were almost simultaneous). Alternatively, if the observation time of the STA at an AP has elapsed for more than a certain amount of time (treated as old observation information), that observation information may not be used.

[0055] Furthermore, if a radio signal from an STA can only be observed by one AP and cannot be observed and compared by multiple APs (including cases where the STA's movement is estimated and STA observation information is not used, as described above, or where STA observation information is not used due to the setting of an effective observation period), the radio signal will be counted at the AP that observed it. Even if a radio signal from the STA is not observed, the count at the last AP that counted it may be maintained for a certain period, but since the STA may actually be gone, it is desirable to limit this certain period to an average time during which multiple APs can be expected to observe the STA.

[0056] In this method 4 as well, a method may be incorporated to set a threshold and limit the RSSI used by each AP, similar to method 3.

[0057] Furthermore, although the above explanation described the case where the AP observes the RSSI of the STA's wireless signal, the STA may also observe the RSSI of wireless signals from multiple APs and count within the observation range of the AP with the highest RSSI. This can be done in the same way as described in Method 3, where the STA notifies the APs of the RSSI of wireless signals from multiple APs, and the count is performed at the AP with the highest RSSI via the wireless LAN controller, or the STA itself may determine the AP with the highest RSSI and notify the AP, and the AP or wireless LAN controller may then enable the wireless LAN controller or ultimately the target number estimation device via the wireless LAN controller to perform the count at that AP.

[0058] In Method 4, unlike Methods 1 and 3, the definition of the AP observation range does not depend on whether or not a connection is established. Therefore, it is appropriate to define the AP observation range more narrowly than in Methods 1 and 3. Generally, when an STA connects to an AP, it may maintain a connection with the old AP even if another AP is more suitable for wireless connection (for example, if the received signal strength from another AP is higher than the received signal strength from the AP being connected to). This is known as the sticky terminal problem. In Methods 1 and 3, it is desirable to define the AP observation range broadly to account for this sticky terminal problem. However, in Method 4, since the STA is counted at the AP that observes it with the highest RSSI, the sticky terminal problem does not affect the STA if a sufficient number of APs can observe it almost simultaneously.

[0059] (Method 5: Tripoint positioning using RSSI from multiple APs) Method 5 is similar to Method 4 in that it observes multiple access points (APs) that form the STA, but each AP estimates the distance between the STA and the AP using the RSSI of the wireless signal from the STA, and estimates that the STA is located at the intersection of these estimated distances from each AP. If there is a difference in height between the AP and the STA, three APs are required for estimating the STA's position because it involves three-dimensional parameters of length, width, and height. This is a position estimation method that uses triangulation. Instead of observing the RSSI of the wireless signal from the STA with multiple APs, the RSSI of wireless signals from multiple APs can be received at the STA, the distance to each AP can be estimated, and the position of the STA can be estimated. In any case, for example, by collecting the acquired RSSI with a wireless LAN controller, the position of the STA can be estimated from the collected RSSI information and the positions of the APs.

[0060] The distance between each AP and STA is derived from a site-general propagation model, such as those provided in ITU-R P.1238-12. Alternatively, the propagation loss coefficient N may be estimated and used in advance by propagation measurements in the target area. Received power Pr [dBm] is expressed as Pr = Pt + Gr + Gt - L, where Pt is the transmit power [dBm], Gt and Gr are the antenna gains of the transceiver [dBi], and L is the propagation loss [dB]. While RSSI is a relative index, received power is an absolute value and cannot be treated as the same. However, by focusing on L, which depends on the distance d, in the region where the RSSI value is not saturated, the relationship RSSI = A - 10N log10(d) (where A is a constant) holds, and by appropriately setting A and N, the distance can be estimated from the RSSI.

[0061] When actually estimating the distance and estimating the STA's position using tripoint positioning, the RSSI fluctuates over time due to radio wave attenuation such as multipath and noise. Therefore, techniques such as Kalman filters may be used to mitigate these effects. Furthermore, due to the propagation characteristic where propagation loss increases logarithmically with distance, the accuracy of distance estimation is higher for shorter distances and lower for longer distances. Therefore, there are techniques to improve position estimation accuracy, such as applying weights to the estimated distance. These existing accuracy improvement techniques may also be used.

[0062] It is also possible to estimate the distance from each AP to the STA from the RSSI of the STA's radio signal, using the received signal strength of the radio signals between APs and the actual distance between APs as reference. However, in this case, while APs may be out of range (LOS) due to, for example, ceiling mounting, LOS is not necessarily guaranteed between APs and STAs, and on the other hand, when observing the radio signals between APs, the horizontal antenna gain at each AP should be considered. A method in which measurement points are defined in a specific space (i.e., the locations of the measurement points are known), the RSSI of the radio signals from APs at each measurement point (e.g., radio signals transmitting Beacon frames; the same applies hereafter), an RSSI map is created within that specific space, and the RSSI of the radio signals from each AP at the terminal to be measured is compared with the RSSI map to estimate that the terminal is located in a place with similar characteristics, that is, pattern matching of the RSSI at the terminal to be measured with the RSSI map is performed to estimate the location of the terminal. This method is called fingerprinting. Since the locations of the APs are known and the relative positions between APs are clear, the above method, which uses the received signal strength of the radio signals between APs and the actual distance between APs as reference, can be said to be utilizing fingerprinting.

[0063] In Method 5, the area in which STAs are counted differs from that of previous methods. Instead of counting STAs within the observation range of each AP as in previous methods, it is assumed that sensors that detect STAs, each with its own observation range, are placed within the measurement range, and the STA is counted within the measurement range that includes the estimated position of the STA. For example, the observation range may be divided into blocks of a certain size, and the STA may be counted in the block that includes the estimated position of the STA. Alternatively, the STA may be counted in a block and its surrounding blocks, taking into account the deviation of the estimated position of the STA. In this case, the count for each block is obtained by dividing the total number of blocks in a given block and its surrounding blocks by 1. Alternatively, the same range as each area representing the number of people present in the output from the person-to-person estimation device (i.e., the output after mathematical optimization) may be used as the observation range of each pseudo-sensor, and the STA is counted within the observation range that includes the estimated position of the STA.

[0064] Figure 8(a) shows an example of estimating the location of STA1 using three APs, specifically AP1, AP2, and AP3. Considering the uncertainty in the estimated location of STA, it is also possible to count STA in a region of a certain size as described above. For example, as shown in Figure 8(b), if the distance to STA1 can only be estimated using two APs, AP1 and AP2, it may not be possible to narrow down the estimated location to a specific region, and it may only be possible to narrow it down to the possibility that STA is likely to be in one of the two regions. In this case, for example, the count of STA 1 is divided equally between these two regions. That is, since there are two regions, 1 / 2 is counted in each region. If each region is further divided into smaller blocks, the count is further divided equally by the number of blocks within each region. As shown in Figure 8(c), if the distance to STA can be estimated using more than three APs, inconsistencies may arise between the distances estimated using RSSI etc. from each AP due to the influence of the propagation environment, and it is possible that the estimated region where STA may be located will not be narrowed down and will become wider. In such cases, the number of blocks to be counted is increased, and the STA count of 1 is equally divided among those blocks for counting. Note that in all examples in Figure 8, APs for which the distance to the STA has not been estimated are omitted. APs that are omitted may be those that have not observed a radio signal from the STA in the first place, or for which too much time has passed since the observation, or for which the RSSI of the observed radio signal is below a certain threshold (this can be said to be a method that combines the case shown in Figure 6(a), one of the implementation examples of Method 3).

[0065] The target number estimation device 4 (target number estimation unit 102) inputs two or more occupancy counts derived from different perspectives, as described in methods 1 to 5 above, into the mathematical optimization process to determine the number of occupants in the target area. In this way, even when only one type of sensor can actually be used, as shown in Figure 9, the number of targets is observed from different perspectives and treated as observation counts for different sensor types, as if there were multiple sensor types, thereby improving the accuracy of target number estimation through mathematical optimization.

[0066] Figure 9 shows an example in which mathematical optimization is applied to the number of people present in each of the following areas: the number of people present in each of the first areas (A'1~) derived using sensor type A from the first perspective (sensor type A'), the number of people present in each of the second areas (A''1~) derived using sensor type A from the second perspective (sensor type A''), and the number of people present in each of the third areas (A''1~) derived using sensor type A from the third perspective (sensor type A''''), in order to estimate the number of people present in each of the fourth area.

[0067] As described in each of the methods above, the definition of the observation range when inputting data into the mathematical optimization process is not the same across different methods. Furthermore, the observation range used in the input method for the mathematical optimization process and the areas representing the number of people in the output from the mathematical optimization process do not necessarily have to be the same.

[0068] Figure 10 is a flowchart showing the processing procedure of the target number estimation device 4 in the embodiment. The target number estimation device 4 acquires the first and second information from the wireless LAN (S1).

[0069] The target number estimation device 4 calculates the target number for each of the multiple first regions obtained by dividing a predetermined geographical area, using the first information by the first method (S2).

[0070] Furthermore, the target number estimation device 4 calculates the target number for each of the multiple second regions obtained by dividing a predetermined geographical area using the second information by the second method (S3). Steps S2 and S3 may be processed in parallel.

[0071] The target number estimation device 4 uses the target number calculated for each first region by the first method and the target number calculated for each second region by the second method to estimate the target number for each of several third regions obtained by dividing a predetermined geographic region through mathematical optimization (S4).

[0072] As described above, the object number estimation device 4 of the embodiment can derive the number of observed objects from different perspectives as if there were multiple types of sensors, even though only one type of sensor exists, and can determine the number of objects in the target area through mathematical optimization.

[0073] In other words, the target number estimation device 4 of this embodiment can improve the accuracy of target number estimation through mathematical optimization using a single sensor type.

[0074] While several embodiments of the present invention have been described, these embodiments are presented as examples only and are not intended to limit the scope of the invention. These novel embodiments can be carried out in a variety of other forms, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. These embodiments and their variations are included in the scope and spirit of the invention, as well as in the claims of the invention and its equivalents. [Explanation of symbols]

[0075] 1...Wireless LAN controller (WLC), 2...Access point (AP), 3...Terminal (STA), 4...Target count estimation device, 11...CPU, 12...Main memory, 13...Communication device, 14...Input device, 15...Display device, 16...Storage device, 101...Data acquisition unit, 102...Target count estimation unit, 200...Firmware.

Claims

1. A data acquisition unit that acquires a first signal containing information about pairs of first regions, obtained by dividing a predetermined geographical area, and the number of targets corresponding to the first region, for a number of first regions, and a second signal containing information about pairs of second regions, obtained by dividing the predetermined geographical area, and the number of targets corresponding to the second region, for a number of second regions. A target number estimation unit generates a third signal containing information about pairs of third regions obtained by dividing a predetermined geographical region and the number of targets corresponding to the third region, for the number of third regions, by applying mathematical optimization processing to the number of targets for each first region included in the first signal and the number of targets for each second region included in the second signal. It is equipped with, Both the first signal and the second signal are signals generated based on the same first wireless system. The first signal is based on the first information of the first wireless system, The second signal is based on the second information of the first wireless system. A device for estimating the number of targets.

2. The first region or the second region is a predetermined observation range of the base station of the first wireless system that forms a wireless service area within the predetermined geographical region. The number of targets corresponding to either the first or second region is the number of slave stations located within the observation range of the master station and connected to the master station. The target number estimation device according to claim 1.

3. The number of targets corresponding to either the first or second region is the number of slave stations connected to the master station whose first value of the radio signal received by the master station is equal to or greater than the threshold. The target number estimation device according to claim 2.

4. The number of targets corresponding to either the first or second region is the number of slave stations connected to the master station in which the first value of the radio signal received by the master station is less than the threshold. The target number estimation device according to claim 2.

5. The first value is RSSI (received signal strength indicator) or SNR (signal-to-noise ratio). The target number estimation device according to claim 3 or 4.

6. The first region or the second region is a predetermined observation range of the base station of the first wireless system that forms a wireless service area within the predetermined geographical region. The number of targets corresponding to either the first or second region is the number of slave stations that are within the observation range of the master station and are observed by the master station, regardless of whether or not they are connected to the master station. The target number estimation device according to claim 1.

7. The master station receives a radio signal of a predetermined frequency and, based on the radio signal, observes the slave stations that are within the observation range of the master station. The target number estimation device according to claim 6.

8. If the same slave station is observed by two or more of the aforementioned master stations, the slave station is counted as the corresponding number associated with the master station that has the largest second value of the radio signal received from the slave station among the two or more master stations. The target number estimation device according to claim 7.

9. The second value mentioned above is RSSI. The target number estimation device according to claim 8.

10. Either the first or second region is a section obtained by dividing the predetermined geographical area into blocks of a predetermined size. The number of objects corresponding to either the first region or the second region is: Based on the signals received by each of the multiple base stations forming a wireless service area within the predetermined geographical area from the same slave station, the distance between each of the multiple base stations and the slave station is estimated. Based on the distances between the aforementioned multiple base stations and the aforementioned slave stations, the position of the slave station is estimated. Identify the area including the estimated location of the aforementioned substation. The number of substations counted for each section is as follows: The target number estimation device according to claim 1.

11. The position of the slave station is estimated by triplicating positioning using the distances between three of the parent stations and the slave station. The target number estimation device according to claim 10.

12. The first wireless system is a wireless LAN (local area network), The aforementioned base station is an access point for the wireless LAN, The aforementioned substation is a station of the wireless LAN. The target number estimation device according to claim 2, 3, 4, 6, 8, or 10.

13. A method for estimating the number of targets within a given geographical area, To obtain a first signal containing information about pairs of first regions obtained by dividing the predetermined geographical area and the number of targets corresponding to the first region, for a number of first regions, and a second signal containing information about pairs of second regions obtained by dividing the predetermined geographical area and the number of targets corresponding to the second region, for a number of second regions, By applying mathematical optimization processing to the number of targets for each of the first regions included in the first signal and the number of targets for each of the second regions included in the second signal, a third signal is generated that contains information about pairs of third regions obtained by dividing the predetermined geographic region and the number of targets corresponding to the third region, for the number of third regions. It is equipped with, Both the first signal and the second signal are signals generated based on the same first wireless system. The first signal is based on the first information of the first wireless system, The second signal is based on the second information of the first wireless system. method.

14. Computers, A data acquisition unit that acquires a first signal containing information about pairs of first regions, obtained by dividing a predetermined geographical area, and the number of targets corresponding to the first region, for a number of first regions, and a second signal containing information about pairs of second regions, obtained by dividing the predetermined geographical area, and the number of targets corresponding to the second region, for a number of second regions. A target number estimation unit that generates a third signal containing information about pairs of third regions obtained by dividing a predetermined geographical region and the number of targets corresponding to the third region, for a number of times equal to the number of third regions, by applying mathematical optimization processing to the number of targets for each first region included in the first signal and the number of targets for each second region included in the second signal. It is a program designed to function as such. Both the first signal and the second signal are signals generated based on the same first wireless system. The first signal is based on the first information of the first wireless system, The second signal is based on the second information of the first wireless system. program.

Images