A method for improving RF sensing performance

Abstract

A method performed by a node (300) for presence detection based on radio frequency, RF, sensing; wherein the node (300) is located in a first sensing area (S1) that is adjacent to a second sensing area (S2); the method (600) comprising: receiving messages from a first node (300') located in the first sensing area (S1); calculating a moving average of signal strengths based on a plurality of messages received from the first node (300'); performing presence detection by comparing signal strength of a newly received message from the first node (300') against a range defined by an upper threshold and a lower threshold; receiving RF sensing information about the second sensing area (S2); adaptively adjusting the upper threshold and / or the lower threshold based on both a change of the moving average and the RF sensing information about the second sensing area (S2), thereby reducing an impact on the change of the moving average caused by human presence or a motion in the second sensing area (S2).

Description

[0001] 2024PF80296

[0002] 1

[0003] A method for improving RF sensing performance

[0004] FIELD OF THE INVENTION

[0005] The invention relates to the field of a radio frequency based sensing system. More particularly, various methods, apparatus, and systems are disclosed herein related to improving RF sensing performance by reducing impact from neighboring areas.

[0006] BACKGROUND OF THE INVENTION

[0007] Radio frequency (RF) based sensing is a technique widely used for motion detection, which involves one or more transmitters arranged for transmitting RF signals on a regular basis, such as according to a certain transmission frequency, and one or more receivers for receiving the RF signals. These RF signals, which may also be used for data communication, when passing through a sensing volume, are affected by presence / movement of a person within the sensing volume. By monitoring such variations of RF signals, it is possible to infer presence or motion of the person. RF based sensing may also be extended to other applications such as location detection, fall detection, gesture detection, or vital signs detection, which are also based on monitoring how RF signals are affected in the sensing volume.

[0008] RF sensing based motion detection has been used in various applications. For example, in smart home automation scenario RF sensing can be used to detect motion in a room and trigger the lights, fans, and other appliances to turn on or off automatically, industrial automation. In a retail analysis scenario, RF sensing can be used to track the movement of customers in a store and analyze shopping patterns to improve sales and marketing strategies.

[0009] On the other hand, RF signals can penetrate walls, floors, and ceilings to certain extent, depending on the type of material and thickness of the barrier. This means that RF signals from a nearby room sometimes may interfere with the signal being received by a RF node in another room, even if there is a physical barrier between them. The ability of RF signals to penetrate walls depends on the frequency of the signal, the distance between the sensor and the source of interference, and the type of material used in the walls. For example, lower frequency signals can penetrate walls more easily than higher frequency signals, and 2024PF80296

[0010] 2 some materials like concrete or metal can block RF signals more effectively than others like dry wall or wood. Therefore, it's important to take into account the environment and potential sources of interference when setting up an RF sensing system to minimize false alarms and missed detections.

[0011] SUMMARY OF THE INVENTION

[0012] RF sensing based presence or motion detection has been commonly used for automatic control in smart home or smart building environments. By detecting motion, the system can trigger various actions, such as turning lights on or off, adjusting heating or cooling, or even unlocking doors. It is recognized by the inventor that it is beneficial to reduce the impact of activity in one or more neighbouring sensing regions on the current sensing region, in order to limit the chances of false alarms or missed detections.

[0013] More particularly, the goal of this invention is achieved by a method as claimed in claim 1, by anode as claimed in claim 10, by a system as claimed in claim 13, and by a computer program as claimed in claim 15.

[0014] In accordance with a first aspect of the invention a method is provided. A method carried out by a node for presence detection based on radio frequency, RF, sensing; wherein the node is located in a first sensing area that is adjacent to a second sensing area; the method comprising: receiving messages from a first node located in the first sensing area; calculating a moving average of signal strengths based on a plurality of messages received from the first node; performing presence detection by comparing signal strength of a newly received message from the first node against a range defined by an upper threshold and a lower threshold; receiving RF sensing information about the second sensing area; adaptively adjusting the upper threshold and / or the lower threshold based on both a change of the moving average and the RF sensing information about the second sensing area, thereby reducing an impact on the change of the moving average caused by human presence or a motion in the second sensing area.

[0015] Presence detection or motion detection may be used for automated control focuses on identifying the presence of individuals or objects in a given area for purposes such as turning on lights, adjusting temperature, or triggering other automated actions. One way to perform RF sensing based presence detection is to observe the variations of received signal 2024PF80296

[0016] 3 strength between two RF nodes in the sensing or detection area. Received Signal Strength Indicator (RS SI) is a measure of the power level of the received signal in wireless communication. There are several potential causes for RSSI fluctuations in practice. For example, RSSI will fluctuate when there are obstructions on the signal propagation path. The physical obstructions may be walls, doors, furniture, people, or another object that obstruct the wireless signal, leading to attenuation and fluctuations in RSSI. The fluctuations may also be caused by reflection and multipath effects. For example, a wireless signal may reflect off walls, floors, and ceilings, creating multiple paths for the signal to travel and arrive at the destination. This can lead to interference and signal cancellation, resulting in fluctuations in RSSI. The fluctuations may also be caused by interference from other electronic devices operating on the same frequency in the surroundings, such as from another RF sensing node in the neighbourhood. Furthermore, temperature and humidity in the environment may also affect the dielectric constant of the channel, which can affect the propagation of the wireless signal and lead to fluctuations in RSSI.

[0017] Since human presence also affects RSSI (e.g. human bodies absorb or reflect wireless signals), RF sensing based presence detection is designed to recognise RSSI changes caused by human presence rather than other factors.

[0018] As discussed above, due to the change of the communication channel and the environment, in practice the received signal strengths between two static nodes also fluctuate with time. By continuously observing the messages, the range of the received signal strengths may represent a moving bar around a moving average of the received signal strengths.

[0019] The moving average may be calculated or updated based on a plurality of messages received from the first node in a sliding window, and the change of the moving average is related to an updated moving average calculated or updated in a current sliding window as compared to a previous sliding window.

[0020] There are several approaches to calculate a moving average of received signal strength in RF sensing. As the most basic version, a moving average may be derived by calculating the average of the last n messages of the received signal strength, which is easy to implement and provides a smooth estimate of the signal strength, but it may not respond quickly to sudden changes in the signal. Alternatively, the moving average may be calculated via a weighted moving average (WMA) approach, where each RSSI value is given a weight based on its position in the moving window. The weights are usually higher for the most recent samples / messages and lower for the older ones. The WMA provides a better estimate of the current signal strength, but it may require more computational resources. Similarly, the 2024PF80296

[0021] 4 weight may be designed to have an exponentially decreasing weight for each sample / message in the window.

[0022] The moving average may also be calculated by applying a moving average filter, which is a temporal filter calculated as the weighted mean of data points within a specified window. In one option, the moving average filter is a Kalman Filter.

[0023] The moving average claimed in the present invention may not be the actual calculated moving average value but may also be a value linked to or representing the actual moving average value.

[0024] When a received signal strength drops below a lower bound of the moving bar, it may indicate there is something hindering the propagation of the signals, possibly a person in the room. If received signals “peak” above this bar, such as higher received signal strength than average, it may indicate there is less hindrance, possibly a person left the room.

[0025] Thus, being above or below the bar can be a value for detection. For example, an upper threshold or an upper bar boundary may be used as an “empty room” threshold, and a lower threshold or a lower bar boundary may be used as a “person in rom” threshold. It’s proposed here to add information from neighbouring areas in adjusting the upper and lower thresholds, thus, to reducing the impact on the change of the moving average caused by human presence or a motion in the second sensing area.

[0026] The messages may be sensing messages dedicated for RF based motion detection. Alternatively, the messages may also be messages for network control or data communication purposes. The messages or sensing messages for presence detection may be transmitted by each node via broadcast, multicast, or unicast. In one example, the messages may be transmitted by each node of a certain sensing area via multicasting to other nodes in the same sensing area, such that the sensing messages comprise a group address for a particular sensing group / area.

[0027] To detect the presence or motion in the sensing area timely, preferably the messages may be transmitted by each node on a regular basis. The nodes in a same subset, such as located in a same sensing area, may use a same duty cycle or transmission frequency for sending sensing messages repeatedly. It may also be an option that each node adopts a different duty cycle or transmission frequency for sending sensing messages.

[0028] The neighboring areas may be related to a sensing area next to, above or below the current sensing area or sensing area of interest.

[0029] The plurality of nodes may follow a wireless communication standard. The wireless communication standard may be related to a standard of Zigbee, Thread, or 2024PF80296

[0030] 5

[0031] Bluetooth Low Energy. In one example, the messages may be either InterP AN messages according to a Zigbee standard, or Bluetooth beacons according to a Bluetooth Low Energy standard.

[0032] The presence detection may be related to a binary presence indication, and a more advance detection regarding a type of the motion, such as gesture detection, activity detection, vital sign detection, etc, may be followed. The type of the motion may be detected by classification, or a pattern recognition based on a comparison against an established baseline.

[0033] The presence or motion detection may be performed in a distributed manner, such as by the node or another node in each sensing area, it is also possible to have the presence or motion detection algorithm performed in a centralized manner, such as by a control device close to the first sensing area, or by a server in the cloud, or via an APP in a user device. The user device may be a smartphone, a tablet, or a laptop. With the centralized option, one or more nodes are configured to provide information related to the detection of the messages, such as received signal strengths, to the control device, server, or user device regularly. And then, the presence or motion detection related to neighboring sensing area can be easily obtained via the centralized algorithm and applied to the adjustment of the upper threshold and lower threshold for the presence detection in an individual sensing area.

[0034] In one example, the RF sensing information about the second sensing area is related to one or more out of: a signal strength, a signal to noise ratio, a presence detection result, a confidence level of the presence detection, a motion detection result, a confidence level of the motion detection, a distance of the human presence or motion detected in the second sensing area to a boundary between the first and the second sensing area, a location of the human presence or motion detected in the second sensing area relative to the first sensing area.

[0035] Beneficially, the method further comprises: adjusting the difference between the upper threshold and the lower threshold based on the RF sensing information about the second sensing area.

[0036] Since there are different factors impacting the fluctuation of the received signal strengths, the difference between the two thresholds, or the width of the moving bar, may be tuned for an actual communication environment, the wireless communication technology, and / or the RF sensing information related to a neighboring sensing area.

[0037] Advantageously, the method further comprises: 2024PF80296

[0038] 6 depending on a detection result in the second sensing area, reducing an amount of adjustment to the lower threshold and / or higher threshold relative to the amount based only on the change of the moving average.

[0039] Preferably, the method further comprises: reducing the amount of adjustment to the lower threshold when a presence is detected by the second sensing area; and / or reducing the amount of adjustment to the higher threshold when an absence is detected by the second sensing area.

[0040] Note that the neighboring areas may be related to a sensing area next to, above or below the current sensing area or sensing area of interest. For example, a neighboring sensing area may be on the floor above or below the current sensing area, and its impact may be significant, such as in the case of wooden floors.

[0041] If the neighbouring area has a low value (and a person is detected) then don’t reduce the lower threshold in the present area or reduce it with a smaller value. Likewise, if the neighbouring area has a high value (no person) then don’t increase the higher threshold in this area or increase it with a smaller value. In this way, the impact of a human presence or motion is mainly limited to a relevant area, rather than one or more neighboring areas.

[0042] This is beneficial for RF sensing based home or building automation, which is designed to trigger actions that are related to the control of a relevant space, such as turning lights on or off or adjusting the thermostat. It is not desirable to turn on the light of the next room, which is not actually occupied.

[0043] In one example, when adjusting the upper threshold and / or the lower threshold, a weight is assigned to a parameter related the RF sensing information about the second sensing area, and the weight is determined by the confidence level of the presence detection or the motion detection in the second sensing area.

[0044] When taking the RF sensing information about the neighboring sensing area into account, it may further apply a different weight on using such information, and the different weight is determined based on a confidence level of the motion detection, a distance of the human presence or motion detected in the second sensing area to a boundary between the first and the second sensing area, or a location of the human presence or motion detected in the second sensing area relative to the first sensing area.

[0045] Advantageously, the method further comprises: receiving RF sensing information about a third sensing area different from the first and second sensing area; 2024PF80296

[0046] 7 adaptively adjusting the upper threshold and / or the lower threshold by additionally taking the RF sensing information about the third sensing area into account.

[0047] The third sensing area may be located further away from the first sensing area than the second sensing area. It is then also beneficial to apply a lower weight when taking the RF sensing information of the third sensing area into account as compared to a weight applied to the RF sensing information of the second sensing area.

[0048] Beneficially, the method further comprises: receiving RF sensing information about all other sensing areas belong to a same sensing zone or residential unit; adaptively adjusting the upper threshold and / or the lower threshold by additionally taking the RF sensing information about all other sensing areas into account.

[0049] For a house or a building comprising serval floors and each floor comprising more than one rooms, it is also preferred to apply the same mechanism to a whole house or whole building scale.

[0050] Preferably, the method further comprises: receiving feedback on one or more detection errors in the first sensing area; training a machine learning model with data on two or more parameters out of: the change of the moving average, the RF sensing information of the second sensing area, the upper threshold, the lower threshold, detection results in the first sensing area, and the feedback on one or more detection errors in the first sensing area; applying the machine learning model on making the adjustment to the upper threshold and / or the lower threshold to reduce an occurrence of detection errors in the first sensing area.

[0051] Since proper adjustments on the lower and upper thresholds are reflected on the accuracy of the presence detection, a machine learning model may be trained based on feedback on detection errors in the first sensing area.

[0052] In accordance with a second aspect of the invention a node is provided. A node for presence detection based on radio frequency, RF, sensing located in a first sensing area that is adjacent to a second sensing area; the node comprising: a radio configured to receive messages from a first node located in the first sensing area; a controller configured to calculate a moving average of signal strengths based on a plurality of messages received from the first node and perform presence detection by 2024PF80296

[0053] 8 comparing signal strength of a newly received message against a range defined by an upper threshold and a lower threshold; wherein the radio is further configured to receive RF sensing information about the second sensing area; and the controller is further configured to adaptively adjust the upper threshold and / or the lower threshold based on both a change of the moving average and the RF sensing information about the second sensing area, thereby reducing an impact on the change of the moving average caused by human presence or a motion in the second sensing area.

[0054] In one example, the controller is further configured to reduce an amount of adjustment to the lower threshold and / or higher threshold compared to an amount of adjustment determined based only on the change of the moving average, depending on a detection result in the second sensing area.

[0055] Beneficially, the radio is further configured to receive the RF sensing information about the second sensing area from a second node located in the second sensing area or a control device communicatively connected to a node in the second sensing area.

[0056] A control device may be used to control and manage the operation of the RF sensing system, which may be deployed to perform tasks such as collecting data from the nodes or sensors, analyzing the data, and sending instructions to the nodes based on the analysis. The control device may be a bridge device, a gateway device, or a central server. The control device may be further configured to perform RF sensing in addition to the system control and management tasks. Thus, the control device may also act as a RF sensing device.

[0057] The RF sensing information about the second sensing area may be received from a second node located in the second sensing area or a control device that is aware of the RF sensing information in the second sensing area.

[0058] In accordance with a third aspect of the invention a system is provided. A system for presence detection based on radio frequency, RF, sensing, wherein the system comprises a plurality of nodes, wherein the plurality of nodes comprises a node according to the present invention and a first node located in a first sensing area are configured to perform presence detection in the first sensing area; and one or more nodes located in a second sensing area adjacent to the first sensing area are configured to perform presence detection in the second sensing area.

[0059] The system may be a RF sensing system deployed for home or building automation. By reducing the impact of presence or motion among adjacent or nearby sensing areas, the detection performance can be improved, such as less false positive or false alarm. 2024PF80296

[0060] 9

[0061] The system may also be configured to provide control information to the plurality of nodes or collect sensing information from the plurality of nodes.

[0062] Beneficially, the system is configured to provide presence detection based control on an electronic device attached to or connected to at least one node out of the plurality of nodes.

[0063] The motion or presence detection may be used to trigger a control command to one or more electronic devices connected to or controlled by one or more nodes, or each node, of the plurality of nodes. It may also be possible that each node out of the plurality of nodes is integrated in, attached to or connected to an electronic device.

[0064] The electronic device may be a lighting device, a wall switch, a bridge, a speaker, a television, a thermostat, a wireless sensor, a power outlet plug, a doorbell, a smart home assistant device, and / or a smart home appliance. For example, the control command may be related to a turn on or power on command to the one or more electronic devices. The control command may be related to changing a setting or an operation mode of the one or more electronic devices.

[0065] BRIEF DESCRIPTION OF THE DRAWINGS

[0066] In the drawings, like reference characters generally refer to the same parts throughout the different figures. Also, the drawings are not necessarily to scale, emphasis instead generally being placed upon illustrating the principles of the invention.

[0067] Fig. 1 illustrates a system comprising a plurality of nodes according to the present invention;

[0068] Fig. 2 shows another example of a system according to the present invention;

[0069] Fig. 3 illustrates an example of a RF sensing system deployed on a single floor; and

[0070] Fig. 4 shows a flow chart of a method according to the present invention.

[0071] DETAILED DESCRIPTION OF EMBODIMENTS

[0072] The embodiments set forth below represent information to enable those skilled in the art to practice the embodiments. Upon reading the following description in light of the accompanying drawings, those skilled in the art will understand the concepts of the disclosure and will recognize applications of these concepts not particularly addressed herein. It should be understood that these concepts and applications fall within the scope of the disclosure. 2024PF80296

[0073] 10

[0074] FIG. 1 illustrates a system 100 configured to perform presence detection based on radio frequency (RF) sensing. The system 100 comprises a plurality of nodes 300, 300’, 300”, which are divided into different sensing groups for presence detection. One or more nodes belonging to a same sensing group are configured to send messages to one or more other nodes belonging to the same sensing group to carry out presence detection via one-hop direct communication. RF sensing is performed among nodes belonging to a same sensing group. The division of different subsets or sensing groups may be based on layout of the area or locations of the nodes, such as different sensing groups located in different rooms or on different floors. Each node of the sensing group may ignore sensing messages received from nodes belong to a different sensing group. The motion detection may be used for building automation to control sensors and actuators integrated in or co-located with the one or more nodes.

[0075] The plurality of nodes 300, 300’, 300” may follow a same wireless communication standard. The wireless communication standard may be related to a standard of Zigbee, Thread, or Bluetooth Low Energy. The nodes are able to detect motions, presence, and other physical activities by measuring the changes in radio frequencies caused by those activities.

[0076] The messages may be sensing messages dedicated for RF based presence or motion detection. Alternatively, the messages may also be data messages for communication or messages for network control purposes. In one option, the messages may be either InterP AN messages according to a Zigbee standard, or Bluetooth beacons according to a Bluetooth Low Energy standard.

[0077] In this example, the system 100 comprises a plurality of nodes 300, 300’, 300’ located in two adjacent sensing areas SI and S2 respectively. A node 300 and a first node 300’ out of the plurality of nodes are located in the first sensing area SI and form a sensing pair, and one or more nodes 300” out of the plurality of nodes are located in the second sensing area S2 adjacent to the first sensing area SI.

[0078] The node 300 comprises a radio configured to receive messages from a first node 300’ located in the first sensing area SI; and a controller configured to calculate a moving average of signal strengths based on a plurality of messages received from the first node 300’ and perform presence detection by comparing signal strength of a newly received message against a range defined by an upper threshold and a lower threshold. The radio is further configured to receive RF sensing information about the second sensing area S2; and the controller is further configured to adaptively adjust the upper threshold and / or the lower 2024PF80296

[0079] 11 threshold based on both a change of the moving average and the RF sensing information about the second sensing area S2, thereby reducing an impact on the change of the moving average caused by human presence or a motion in the second sensing area S2.

[0080] RF sensing by sending a plurality of messages between a pair of nodes is used to detect and measure the presence of objects or obstacles located between or around the two nodes. In this method, the node 300 is configured to receive a plurality of messages from the first node 300’. The messages are reflected and scattered by the objects in their path, and by monitoring the plurality of messages the node may use the received signals to estimate the distance and position of the objects. The messages may be sent periodically. The node may also detect the velocity and direction of the objects based on the received messages.

[0081] Due to the change of the communication channel and the environment, in practice the received signal strengths of the plurality of messages also fluctuate with time. The range of the received signal strengths may represent a moving bar, such as the average delayed signal strengths. If received signals “drop” below this bar, such as lower received signal strength than average, it may indicate there is something hindering the propagation of the signals, possibly a person in the room. If received signals “peak” above this bar, such as higher received signal strength than average, it may indicate there is less hindrance, possibly a person left the room.

[0082] It can be seen that being above or below the bar can be a value for detection. For example, an upper bar boundary may be used as an “empty room” threshold, and a lower bar boundary may be used as a “person in rom” threshold. The width of the bar, the difference between the two thresholds, may be tuned for an actual communication environment and / or the wireless communication technology. It’s proposed here to add information from neighbouring areas in adjusting the upper and lower thresholds. If the neighbouring area has a low value (and a person is detected) then don’t reduce the threshold in the present area or reduce it with a smaller value. Likewise, if the neighbouring area has a high value (no person) then don’t increase the threshold in this area or increase it with a smaller value.

[0083] The RF sensing information about the second sensing area S2 may be related to one or more out of: a signal strength, a signal to noise ratio, a presence detection result, a confidence level of the presence detection, a motion detection result, a confidence level of the motion detection, a distance of the human presence or motion detected in the second sensing area S2 to a boundary between the first and the second sensing area S2, a location of the 2024PF80296

[0084] 12 human presence or motion detected in the second sensing area S2 relative to the first sensing area SI.

[0085] The controller of the node 300 may be further configured to adjust the difference between the upper threshold and the lower threshold based on the RF sensing information about the second sensing area S2.

[0086] The controller of the node 300 may be further configured to reduce an amount of adjustment to the lower threshold and / or higher threshold according to a detection result in the second sensing area S2, compared to an amount of adjustment determined based only on the change of the moving average.

[0087] For example, the controller of the node 300 may be further configured to reduce the amount of adjustment to the lower threshold when a presence is detected by the second sensing area S2; and / or reduce the amount of adjustment to the higher threshold when an absence is detected by the second sensing area S2.

[0088] The radio of the node 300 may be further configured to receive the RF sensing information about the second sensing area S2 from a second node 300” located in the second sensing area S2 or a control device communicatively connected to a node 300” in the second sensing area S2. The control device may be a bridge device, a gateway device, or a central server of the system 100. The control device may also be configured to perform RF sensing in addition to the system control and management tasks, such as acting as a RF sensing node.

[0089] Instead of implementing the presence or motion detection in a distributed manner, such as by the node 300 or another node in each sensing area, it is also possible to have the presence or motion detection algorithm performed in a centralized manner, such as by a control device of the system, by a server in the cloud, or via an APP in a user device. The user device may be a smartphone, a tablet, or a laptop. With the centralized option, one or more nodes are configured to provide information related to the detection of the messages, such as received signal strengths, to the control device, server, or user device regularly. And then, the presence or motion detection related to neighboring sensing area can be easily obtained via the centralized algorithm and applied to the adjustment of the upper threshold and lower threshold for the presence detection in an individual sensing area.

[0090] Each node out of the plurality of nodes 300, 300’, 300 ’’may be integrated in, attached to, or connected to an electronic device to be controlled mainly or partially via motion detection. The electronic device may be a lighting device, a wall switch, a bridge, a speaker, a television, a thermostat, a wireless sensor, a power outlet plug, a doorbell, a smart home assistant device, and / or a smart home appliance. For example, a control command to be 2024PF80296

[0091] 13 generated upon motion detection may be related to a turn on or power on command to the one or more electronic devices. The control command may be related to changing a setting or an operation mode of the one or more electronic devices.

[0092] The system 100 may be configured to provide presence detection based control on an electronic device attached to or connected to at least one node out of the plurality of nodes 300, 300’, 300”.

[0093] With one or more nodes out of the plurality of nodes 300, 300’, 300” integrated in or attached to a luminaire, the RF sensing system may also be a lighting control system and is configured to carry out data communication in addition to RF sensing.

[0094] FIG. 2 shows another example of a system 100 according to the present invention. In this example, there is a third sensing area S3 nearby, which is different from the first SI and second sensing area S2. The controller of the node 300 is further configured to adaptively adjust the upper threshold and / or the lower threshold by additionally taking the RF sensing information about the third sensing area S3 into account.

[0095] According to the proximity between the present sensing area and different neighboring sensing areas, a different weight may be applied when adjusting the upper threshold and / or the lower threshold of the present sensing area by considering the RF sensing information about an individual neighboring sensing area. For example, human presence or motion in a nearby sensing area will typically have a greater impact on the moving average of the current sensing area than another, more distant neighboring sensing area. It may also be configured that only the RF sensing information related to a neighboring sensing area that is within certain distance is considered. The distance between the present sensing area and different neighboring sensing areas may be derived based on knowledge of the house or building layout or obtained from a user or software application. The distance information may also be obtained by overhearing messages sent from a different sensing group or area.

[0096] FIG. 3 illustrates an example of a RF sensing system 100 deployed on a single floor. For illustration purpose, each node of the RF sensing system is integrated in or attached to a luminaire in this example. Of course, each one of the plurality of nodes may also be integrated in or attached to a different type of electronic device or be deployed in a standalone manner.

[0097] The present invention aims to optimize a RF sensing system deployed in a house or building with multiple adjacent but separated sensing areas. The problems in such a system are as follows: 2024PF80296

[0098] 14

[0099] • Signal levels and signal to noise ratios (used for motion detection) fluctuate over time and cause false positives and false negatives on motion / presence detection. The detection quality can be improved by knowledge of the signal levels and signal to noise ratios of an adjacent area to set a more accurate base line.

[0100] • A person staying in one room influences the “base line” measurement in an adjacent room increasing the chance of false positives and false negatives on motion / presence detection in the adjacent room. The detection quality can be improved by knowledge of the estimated distance and position to an area.

[0101] It is therefore proposed to share signal levels and signal to noise ratios, together with motion / presence detection results, among adjacent areas, such that a detection algorithm adjusts its baseline for the current area also takes measurement value or detection results from one or more adjacent areas into account. For example, when there is a person in an adjacent area, if measurement values change in the current area (because the person is also close to the current area), it is proposed to avoid gradually increase the “empty room” threshold value as that might miss a person entering the room and lead to false negative). In another example, when there are no people in the adjacent area, if measurement values change, it is proposed to avoid gradually decrease the “people present” threshold value as that might trigger motion / presence while there is no person entering and lead to false positive.

[0102] As shown in FIG. 3, RF sensing for home automation or lighting control may treat each room as a separate sensing area or sensing zone. This is logical for users as they want to automate the lights in a room. However, imagine a person in living area B who also influences RF sensing measurements in area A, C, and D. Each area may have one “collector” node providing their measurements to a bridge or a control device where a centralized motion detection algorithm runs.

[0103] • The motion detection algorithm may use all measurement data from the same house. Allowing a more “home wide” threshold for “empty” and “people present” area. With awareness of the house layout, the algorithm may be configured to reduce the impact of non-adjacent areas (e.g. A on D). Such knowledge may be provided by a user via an APP regarding the RF-sensing area information to the bridge or the control device.

[0104] • Presence or motion in an adjacent area is used to prevent adjusting or over-adjusting “empty” and “people present” thresholds in a certain area of interest.

[0105] Alternatively, the motion detection algorithm may also be carried out by one or more nodes in the system in a distributed manner. The one or more nodes may receive RF 2024PF80296

[0106] 15 sensing information or other information (e.g. layout of the building) from a bridge or control device or from another node located in a different sensing area.

[0107] With RF sensing based presence or motion detection, a node or device broadcasts or multicast a message to other nodes or devices belonging to a same sensing group or area record the signal level. It may also be the case that there are one or more sensing pairs in each sensing area and messages are sent from one node of a sensing pair to another node via unicast. If a person moves between nodes, the signal levels between these nodes are affected more. If a person moves close to a node, the messages to or from that device get bigger impact. This is a simplified position detection. A user might inform the bridge or control device of the sensing system through the APP that a specific node in area C is close to the door towards area B. So, the algorithm knows that when passing from area C to area B the signals for which node or nodes likely get affected first. This knowledge can be used to increase the detection quality. Alternatively, when sufficient historic information is obtained, a machine learning algorithm can detect this relation as well.

[0108] FIG. 4 shows a flow chart of a method 600 according to the present invention. A method 600 performed by a node 300 for presence detection based on radio frequency, RF, sensing; wherein the node 300 is located in a first sensing area SI that is adjacent to a second sensing area S2; the method 600 comprising: receiving, in step S601, messages from a first node 300’ located in the first sensing area SI; calculating, in step S602, a moving average of signal strengths based on a plurality of messages received from the first node 300’; performing, in step S603, presence detection by comparing signal strength of a newly received message from the first node 300’ against a range defined by an upper threshold and a lower threshold; receiving, in step S604, RF sensing information about the second sensing area S2; adaptively adjusting, in step S605, the upper threshold and / or the lower threshold based on both a change of the moving average and the RF sensing information about the second sensing area S2, thereby reducing an impact on the change of the moving average caused by human presence or a motion in the second sensing area S2.

[0109] The RF sensing information about the second sensing area S2 is related to one or more out of: a signal strength, a signal to noise ratio, a presence detection result, a confidence level of the presence detection, a motion detection result, a confidence level of the 2024PF80296

[0110] 16 motion detection, a distance of the human presence or motion detected in the second sensing area S2 to a boundary between the first and the second sensing area S2, a location of the human presence or motion detected in the second sensing area S2 relative to the first sensing area SI. The method 600 may further comprise a step of adjusting the difference between the upper threshold and the lower threshold based on the RF sensing information about the second sensing area S2.

[0111] The method 600 may further comprise a step of reducing an amount of adjustment to the lower threshold and / or higher threshold according to a detection result in the second sensing area S2, compared to an amount of adjustment determined based only on the change of the moving average.

[0112] The method 600 may further comprise reducing the amount of adjustment to the lower threshold when a presence is detected by the second sensing area S2; and / or reducing the amount of adjustment to the higher threshold when an absence is detected by the second sensing area S2.

Claims

2024PF8029617CLAIMS:

1. A method (600) performed by a node (300) for presence detection based on radio frequency, RF, sensing; wherein the node (300) is located in a first sensing area (SI) that is adjacent to a second sensing area (S2); the method (600) comprising: receiving (S601) messages from a first node (300’) located in the first sensing area (SI); calculating (S602) a moving average of signal strengths based on a plurality of messages received from the first node (300’); performing (S603) presence detection by comparing signal strength of a newly received message from the first node (300’) against a range defined by an upper threshold and a lower threshold; receiving (S604) RF sensing information about the second sensing area (S2); and adaptively adjusting (S605) the upper threshold and / or the lower threshold based on both a change of the moving average and the RF sensing information about the second sensing area (S2), thereby reducing an impact on the change of the moving average caused by human presence or a motion in the second sensing area (S2).

2. The method (600) of claim 1 , wherein the RF sensing information about the second sensing area (S2) is related to one or more out of: a signal strength, a signal to noise ratio, a presence detection result, a confidence level of the presence detection, a motion detection result, a confidence level of the motion detection, a distance of the human presence or motion detected in the second sensing area (S2) to a boundary between the first and the second sensing area (S2), a location of the human presence or motion detected in the second sensing area (S2) relative to the first sensing area (SI).

3. The method (600) of claim 1 or 2 further comprising: adjusting the difference between the upper threshold and the lower threshold based on the RF sensing information about the second sensing area (S2).2024PF80296184. The method (600) of any one of previous claims further comprising: depending on a detection result in the second sensing area (S2), reducing an amount of adjustment to the lower threshold and / or higher threshold compared to an amount of adjustment determined based only on the change of the moving average.

5. The method (600) of claim 4 further comprising: reducing the amount of adjustment to the lower threshold when a presence is detected by the second sensing area (S2); and / or reducing the amount of adjustment to the higher threshold when an absence is detected by the second sensing area (S2).

6. The method (600) of any one of previous claims 2-5, wherein when adjusting the upper threshold and / or the lower threshold, a weight is assigned to a parameter related the RF sensing information about the second sensing area (S2), and the weight is determined by the confidence level of the presence detection or the motion detection in the second sensing area (S2).

7. The method (600) of any one of previous claims further comprising: receiving RF sensing information about a third sensing area (S3) different from the first (SI) and second sensing area (S2); adaptively adjusting the upper threshold and / or the lower threshold by additionally taking the RF sensing information about the third sensing area (S3) into account.

8. The method (600) of any one of previous claims further comprising: receiving RF sensing information about all other sensing areas belong to a same sensing zone or residential unit; adaptively adjusting the upper threshold and / or the lower threshold by additionally taking the RF sensing information about all other sensing areas into account.

9. The method (600) of any one of previous claims further comprising: receiving feedback on one or more detection errors in the first sensing area(Si); training a machine learning model with data on two or more parameters out of: the change of the moving average, the RF sensing information of the second sensing area2024PF8029619(S2), the upper threshold, the lower threshold, detection results in the first sensing area (SI), and the feedback on one or more detection errors in the first sensing area (SI); applying the machine learning model on making the adjustment to the upper threshold and / or the lower threshold to reduce an occurrence of detection errors in the first sensing area (SI).

10. A node (300) for presence detection based on radio frequency, RF, sensing located in a first sensing area (SI) that is adjacent to a second sensing area (S2); the node (300) comprising: a radio configured to receive messages from a first node (300’) located in the first sensing area (SI); and a controller configured to calculate a moving average of signal strengths based on a plurality of messages received from the first node (300’) and perform presence detection by comparing signal strength of a newly received message against a range defined by an upper threshold and a lower threshold; wherein the radio is further configured to receive RF sensing information about the second sensing area (S2); and the controller is further configured to adaptively adjust the upper threshold and / or the lower threshold based on both a change of the moving average and the RF sensing information about the second sensing area (S2), thereby reducing an impact on the change of the moving average caused by human presence or a motion in the second sensing area (S2).

11. The node (300) of claim 10, wherein the controller is further configured to reduce an amount of adjustment to the lower threshold and / or higher threshold compared to an amount of adjustment determined based only on the change of the moving average, depending on a detection result in the second sensing area (S2).

12. The node (300) of claim 10 or 11, wherein the radio is further configured to receive the RF sensing information about the second sensing area (S2) from a second node (300”) located in the second sensing area (S2) or a control device communicatively connected to a node (300”) in the second sensing area (S2).

13. A system (100) for presence detection based on radio frequency, RF, sensing, wherein the system (100) comprises a plurality of nodes (300, 300’, 300”), wherein the2024PF8029620 plurality of nodes comprises a node (300) according to any one of the previous claims 10-12 and a first node (300’) located in a first sensing area (SI) are configured to perform presence detection in the first sensing area (SI); and one or more nodes (300”) located in a second sensing area (S2) adjacent to the first sensing area (SI) are configured to perform presence detection in the second sensing area (S2).

14. The system (100) of claim 13 is configured to provide presence detection based control on an electronic device attached to or connected to at least one node out of the plurality of nodes (300, 300’, 300”).

15. A computer program product comprising instructions which, when the computer program product is executed on a processing unit of a node (300) according to any one of the claims 10-12, cause the node (300) to carry out the steps of the method (600) according to any one of the claims 1-9.

Images