Method and device for identification of sensing applications and method and device for training a module for identification - Patents.com
Trained modules in wireless communication systems identify and predict sensing applications, improving spectral efficiency by reducing interference and optimizing resource allocation in shared spectrum environments.
Patent Information
- Application Number
- JP2023534669
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- Priority Date
- 2021-01-19
- Filing Date
- 2021-12-23
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2041-12-23
AI Technical Summary
Existing wireless communication and sensing systems lack efficient methods for identifying and predicting sensing applications, leading to inefficient spectrum usage and interference between communication and sensing devices, particularly in shared spectrum environments.
The implementation of trained modules, such as machine learning or deep learning models, to identify and predict sensing applications based on characteristics like periodicity, frame structure, bandwidth, and waveform of received wireless signals, allowing devices to schedule transmissions and utilize existing sensing signals for their own purposes.
Enhances spectral efficiency by reducing redundant sensing transmissions, minimizing interference, and optimizing resource allocation in shared spectrum environments.
Smart Images

Figure 0007584659000006 
Figure 0007584659000007 
Figure 0007584659000008
Abstract
Description
[Technical field]
[0001] The present disclosure relates to the field of wireless communication and sensing. In particular, the present disclosure provides an apparatus and method for a wireless communication and sensing system.
[0002] Wireless communications have been evolving for several decades now. Examples of prominent standardization bodies include the Third Generation Partnership Project (3GPP) and IEEE 802.11, commonly referred to as Wi-Fi. In practice, devices supporting such standards or having their own communications and sensing applications may share the spectrum, at least in part.
[0003] Cognitive radio is one of the emerging technologies for utilizing system spectrum. Cognitive radio devices are supposed to dynamically use the best wireless channels in their vicinity to improve spectrum efficiency. To achieve this, spectrum occupancy information may be desired, which helps to model and predict spectrum availability for efficient dynamic spectrum access. Spectrum occupancy prediction may be based on predicting future occupancy using information on previous spectrum occupancy. Such prediction is based on exploiting the inherent correlation between past and future occupancy. Some approaches exploit correlation in the time domain, positioning the spectrum prediction as a time series prediction. Some approaches further consider exploiting correlation along the frequency axis, thus exploiting time-frequency correlation. Correlation may also exist in the spatial domain. Therefore, it may be desirable to exploit correlation in all mentioned domains.
[0004] Some specific works use traditional statistical methods for spectral prediction, such as autoregressive model-based methods or Bayesian estimation-based methods. There are some approaches that use machine learning (ML) methods, such as wavelet neural networks and shallow artificial neural networks. Some more recent works refer to deep learning (DL)-based approaches to further improve the performance. DL-based methods can use correlations in multiple domains. Overview
[0005] Methods and techniques, and corresponding devices, that facilitate cognitive radio applications and other approaches to improving wireless spectrum usage are described. This disclosure specifically recognizes that identification of sensing applications may be used to further improve one or more aspects of wireless spectrum usage, particularly when the spectrum is shared by potentially disparate communication and / or sensing devices.
[0006] The invention is defined by the independent claims. Some exemplary implementations are provided in the dependent claims.
[0007] In some example implementations, a method is provided for identifying a sensing application, the method including: acquiring a received wireless signal; estimating, by a trained module, a presence of a sensing signal in the received signal, where the sensing signal is a signal generated by a sensing application; and performing wireless receiving, transmitting, or sensing based on the estimated presence of the sensing application.
[0008] Other exemplary implementations relate to training methods, training devices, and devices for identifying sensing applications, as well as programs and computer readable media having stored thereon programs for performing those methods.
[0009] These and other features and characteristics of the subject matter disclosed herein, together with the method of operation and function of the associated elements of construction, and the combination of parts and economies of manufacture, will become more apparent from a consideration of the following description and appended claims, with reference to the accompanying drawings, all of which form a part of this specification. It is to be expressly understood, however, that the drawings are for the purposes of illustration and description only, and are not intended to define the scope of the disclosed subject matter. As used in the specification and claims, the singular forms "a," "an," and "the" include plural referents unless the context clearly dictates otherwise. [Brief description of the drawings]
[0010] An understanding of the nature and advantages of various embodiments may be realized by reference to the following figures. [Figure 1] FIG. 1 is a block diagram illustrating a basic communication system. [Diagram 2] FIG. 2 is a flow diagram illustrating an example spectrum management method. [Diagram 3] FIG. 3 is a schematic diagram illustrating an exemplary pulsed radar signal. [Figure 4] FIG. 4 is a schematic diagram showing various frame formats depending on whether the application generating the frame is sensing, communication, or JSC. [Diagram 5] FIG. 5 is a block diagram of an example sensing application forecast. [Figure 6A] FIG. 6A is a schematic diagram illustrating a simple sensing scenario in which there is no object to be detected. [Figure 6B]FIG. 6B is a schematic diagram illustrating a simple sensing scenario when an object to be detected is present. [Figure 7] FIG. 7 is a schematic diagram illustrating a sensing scenario in which multiple devices communicate and sense via line of sight. [Figure 8] FIG. 8 is a schematic diagram illustrating a sensing scenario in which multiple devices communicate and sense via non-line-of-sight. [Figure 9] FIG. 9 is a flow diagram illustrating an exemplary method for detecting and / or identifying a sensing signal. [Figure 10] FIG. 10 is a flow diagram illustrating an exemplary method for training a module for detection and / or discrimination of a sensing signal. [Figure 11] FIG. 11 is a block diagram illustrating an exemplary apparatus for detecting and / or identifying a sensing signal. [Figure 12] FIG. 12 is a block diagram illustrating an example apparatus for training a module for detection and / or discrimination of a sensing signal. Detailed Description
[0011] For purposes of the following description, the terms "end," "upper," "lower," "right," "left," "vertical," "horizontal," "top," "bottom," "transverse," "vertical," and derivatives thereof, refer to the disclosed subject matter when oriented in the drawings. It is to be understood, however, that the disclosed subject matter may assume various alternative variations and step sequences unless expressly specified otherwise. It is also to be understood that the specific devices and processes illustrated in the accompanying drawings and described in the following specification are merely exemplary embodiments or aspects of the disclosed subject matter. Accordingly, specific dimensions and other physical characteristics related to the embodiments or aspects disclosed herein should not be considered as limiting, unless otherwise indicated.
[0012] As used herein, aspects, components, elements, structures, acts, steps, functions, instructions, and / or the like should not be construed as critical or required unless expressly stated. Additionally, as used herein, the articles "a" and "an" are intended to include one or more items and may be used interchangeably with "one or more" and "at least one." Additionally, as used herein, the term "set" is intended to include one or more items (e.g., related items, unrelated items, combinations of related and unrelated items, and / or the like) and may be used interchangeably with "one or more" or "at least one." When only one item is intended, the term "a" or similar language is used. Additionally, as used herein, terms such as "has," "have," "having," and the like are intended to be open-ended terms. Additionally, the phrase "based on" is intended to mean "based at least in part on," unless otherwise specified.
[0013] While the aforementioned techniques may work, they do not provide complete information or awareness of future spectrum or its use. Meanwhile, wireless sensing is becoming more popular in commercial devices for environmental monitoring, health monitoring, and many other applications. Military applications of wireless sensing, such as radar, are becoming ever more popular. This information is not very useful for sensing applications, since knowledge of the presence or absence of a signal or its characteristics is generally not sufficient to identify or predict behavior applications. In particular, sensing applications typically generate signals that may have patterns that differ from those of some communication applications. For example, most wireless sensing applications generate periodic signal transmissions of various periods. Effectively scheduling and managing these transmissions results in less wasted spectrum and power. Thus, if a device can determine which sensing applications are utilizing the spectrum, it can plan its transmissions to fit into free slots or utilize these transmissions for its own sensing, if appropriate. Further uses of this information are possible, for example, spectrum predictions may be provided to additional devices in the area to enable efficient resource access, regardless of whether the spectrum is shared or not. In summary, it would be desirable to provide an approach for sensing application identification and prediction to improve resource access and utilization, possibly without a single central resource management entity. In this way, some of the multiple coexisting sensing and communication devices can operate in a more efficient manner.
[0014] FIG. 1 shows an exemplary wireless system WiS, where Tx represents a transmitter of a wireless signal and Rx represents a receiver of a wireless signal. The transmitter Tx can transmit a signal to a receiver Rx or a group of receivers or broadcast a signal via an interface Itf. The interface may be any wireless interface. The interface may be specified by resources available for transmission and reception by the transmitter Tx and the receiver Rx. Such resources may be defined in one or more (or all) of the time domain, the frequency domain, the code domain, and the spatial domain. It should be noted that in general, both the "transmitter" and the "receiver" may be integrated in the same device. In other words, the devices Tx and Rx in FIG. 1 may include the functions of Rx and Tx, respectively.
[0015] The present disclosure is not limited to any particular transmitter Tx, receiver Rx, and / or interface Itf implementation. However, it can be easily applied to some existing communication systems as well as extensions of such systems or new communication systems. Exemplary existing communication systems may be, for example, 5G New Radio (NR) in current or future releases, and / or IEEE 802.11-based systems such as IEEE 802.11be that are being recently researched. The wireless signal is not necessarily a communication signal in the sense that it does not necessarily carry out human or machine communication. It may in particular be a sensing signal, for example a radar signal or a sounding signal, or any other kind of wireless signal from a sensing device, for example any signal that reports sensing results to other device(s).
[0016] For example, the IEEE 802.11bf amendment, i.e., Wireless Local Area Network (WLAN) Sensing, may include support for wireless sensing in WLAN networks. Some embodiments may be used to enhance the performance of devices conforming to this standard, for example, to reduce the amount of redundant sensing signals in an area or network. The fifth generation (5G) New Radio (NR) standard or the 6G standard may also apply wireless sensing as part of future cellular communication networks. Some embodiments of the present disclosure may help predict free space in unlicensed spectrum during opportunistic spectrum usage, where most wireless sensing is expected to occur. The present disclosure is also applicable to Long Term Evolution (LTE) / Unlicensed LTE (LTE-U) communication technologies.
[0017] As mentioned above, spectrum awareness is an important part of cognitive radio (CR). Some embodiments of the present disclosure can facilitate the identification and prediction of sensing transmissions. IEEE 802.22 and IEEE 802.15 standards support CR and may benefit from the present disclosure. The present disclosure is also applicable to low-power wide-area network (LPWAN) technologies, as it helps improve power efficiency by reducing the number of redundant sensing transmissions. Thus, it is relevant to LPWAN standards such as Wize, ZigBee, NarrowBand IoT, LoRaWAN, etc. In general, some embodiments can be used at high frequencies or millimeter waves (mm waves), because the spectrum availability and propagation characteristics are suitable for high-resolution wireless sensing. This can be used for wireless sensing resource management.
[0018] The devices Tx and Rx in FIG. 2 may include the functions of Rx and Tx, respectively. The transmitter Tx and receiver Rx may be implemented in any device, for example, a base station (eNB, AP) or a terminal (UE, STA), or any other entity of the wireless system WiS. A device such as a base station, an access point, or a terminal may implement both Rx and Tx. The present disclosure is not limited to any particular transmitter Tx, receiver Rx, and / or interface 200 implementation. However, it may be easily applied to some existing communication systems, as well as extensions of such systems, or new communication systems. An exemplary existing communication system may be, for example, a 5G New Radio (NR) in current or future releases, and / or an IEEE 802.11-based system, such as the recently researched IEEE 802.11be. The sensing application signal may also be embedded within resources provided by one or more of the known systems, such as some IEEE 802.11 standards or their possible specific extensions to support sensing applications.
[0019] One application of sensing application identification is spectrum occupancy. In general, spectrum occupancy may be based on using previous occupancies arranged in a time-frequency grid together with the spatial correlation of such occupancies to predict future spectrum occupancy states. This may be achieved, for example, by training some trainable models to predict the occupancy probability of the time-frequency occupancy grid. It should be noted that machine learning or deep learning and classifier models may be trained on real data acquired through measurements, after which they can be immediately used for spectrum prediction. Deep learning refers to trainable models having two or more layers. Machine learning refers to trainable models having any structure, including layer structures applying one or more layers. These models may be embedded in functional modules and / or physical modules. Thus, when referring to trainable modules herein, it means functional modules that implement trainable models such as machine learning, deep learning, or other types of models. Such models, if used, are typically pre-trained (in a training phase) and are not changed (updated) during operation. However, the present disclosure is not limited to any particular spectrum occupancy prediction. It may also be performed without a trained module, using statistical or other models.
[0020] One non-limiting application of spectrum occupancy prediction is for radio resource management and / or spectrum allocation in some wireless networks, such as mobile communication networks (e.g., 4G, 5G, 6G, etc.) or wireless local area networks (WLANs). In general, spectrum occupancy prediction may be utilized in cognitive radio systems. FIG. 2 shows an example of the use of spectrum sensing in cognitive radio. A radio environment 210 may be formed by multiple BSs and UEs and / or other entities, such as those shown in FIG. 1. One or more of the entities may then perform spectrum sensing 220. Spectrum sensing is a procedure in which a device listens (detects) whether other devices are already transmitting before transmitting a signal. Such an approach may be particularly useful in spectrum portions that may be used by different systems, such as unlicensed spectrum portions that may be used by WLANs and cellular or other systems. If a band for transmission is detected as available in step 220, it may be determined to use the spectrum for transmission in step 240. If the spectrum is not available in step 220, a different spectrum may be selected for use in step 230. Thus, the spectrum may be shared in an efficient manner in 250, and the spectrum usage information may be further considered by the wireless environment 210. Here, the wireless environment may perform spectrum occupancy prediction and select the spectrum to sense based on the prediction result. The spectrum occupancy prediction may use neural network(s), such as CNN or LSTM (generally, recurrent network).
[0021] Future wireless devices are expected to be sensing-enabled, or in some cases wireless sensors only, to support communication applications and / or provide a wide range of other applications, e.g., fully immersive augmented reality, improved quality of life by enabling smart environments, health-related applications by non-invasive testing and vital signs monitoring, etc. Wireless sensing applications may require periodic or continuous sensing transmissions. However, allowing all sensing / sensing-enabled devices to transmit their sensing transmissions may reduce the spectral efficiency and performance of networks operating in unlicensed bands. Furthermore, since sensing transmissions are periodic, they are likely to cause interference to communication transmissions if they are opportunistically scheduled or have opportunistic channel access mechanisms. This problem can be solved by including sensing-aware channel access and sensing coordination protocols in the standards. However, these only allow communication and coordination of sensing and devices within the same network. At the same time, this increases the overhead and complexity of signalization. The trend in wireless communication is toward distributed, minimally coordinated networks where multiple wireless networks coexist in the same area. Therefore, the standard requires a method or act to identify and predict future sensing transmissions before they are transmitted, allowing devices to better allocate resources and schedule transmissions.
[0022] Wireless sensing is a process of obtaining information or awareness of an environment through measurements of received (e.g., reflected or directly received) electromagnetic signals. With this definition, processes such as spectrum / channel sensing, radar, integrated radar and communication, WLAN sensing, and other methods can be considered as wireless sensing methods. Most wireless sensing methods have a periodic or continuous transmission pattern. One example can be radar, where a continuous signal or periodic pulses are transmitted. Another example can be channel state information (CSI)-based WLAN sensing, where packets are transmitted at a certain period. Sensing transmissions may include a particular frame design or transmission mechanism, which is specific to some sensing application(s) and may vary based on the requirements of the sensing application and environmental conditions. Periodicity is an important condition for sensing applications because any disruption in the period of the transmitted / received signals due to interference from other devices, inability to schedule transmissions, or inability to access the channel using a channel access protocol may disrupt the measurements. This may result in false alarms, missed detections, reduced resolution of the sensed information, and overall poor performance of the sensing application. Depending on the application, this could have significant financial consequences or even endanger lives.
[0023] Signal identification allows a device to identify several features of a signal, such as wireless technology (LTE, 5G, etc.), waveform, modulation type, etc., based on several characteristics of the signal, such as bandwidth, spectrogram image, etc. There are several applications where certain characteristics are identified for synchronization or authentication purposes. Also, spectrum sensing is known to be used to identify the spectrum occupancy status of key users. However, this may require continuous spectrum sensing. Alternatively, spectrum prediction techniques can be used to save the time, energy and computational overhead required for spectrum sensing.
[0024] Although the above techniques have been successful in many applications, they do not provide complete information about future and current sensing applications. These applications in a network can be predicted using higher layer information. However, problems arise when the spectrum is shared between different networks or devices. This is because the access point (AP) / base station (BS) / other devices are not aware of these transmissions and the transmissions may not be suitable for the sensing methods of the AP / BS / other devices or cause interference. In this situation, spectrum occupancy prediction techniques can be used to predict spectrum occupancy.
[0025] In future wireless communications, there may be several wireless sensing applications. Generally, sensing applications use continuous signals (with some periodicity), which may significantly reduce the spectrum efficiency. This is especially true when multiple sensing applications / devices are used simultaneously and / or in the same area. Some embodiments and exemplary implementations provide methods and apparatuses for the identification and prediction of the sensing application according to some characteristics of the received signal. If the sensing application is known, other devices can accordingly schedule their transmissions without interfering with other applications, or utilize the current sensing transmission for their own sensing application. Thus, spectrum, power, and other resources can be used efficiently.
[0026] There are two aspects of this disclosure: sensing application identification and sensing application prediction. In sensing application identification, the first step is to identify whether the signal is a communication signal or a sensing signal. To identify this, periodicity and frame structure as well as other features can be used. These are described in more detail below. Detecting the type of application: Periodic
[0027] The sensing signal may be transmitted at a specific rate or period. Depending on the characteristics to be sensed and the resolution or accuracy, the signal may be transmitted at various rates. On the other hand, communication signals are typically more random and do not have significant periodicity (due to the nature of scheduling, they may be periodic for a few packets, but are not periodic over the long term). An exemplary sensing signal is shown in FIG. 3.
[0028] There are a variety of radar waveforms that can be used. The most common are pulsed radar and frequency modulated continuous wave (FMCW) chirp radar. In pulsed radar, the signal is "on" (or transmitted) for a portion of the total frame time and "off" (not transmitted) for the rest of the time, as shown in Figure 3. The duration the signal is "on" or "off" (duty cycle) gives a trade-off between range and range resolution. A longer "on" period (wider pulse width) improves range but reduces resolution, and vice versa. FMCW chirps are continuous (no off time) and can also detect relative velocity (pulsed radar cannot do this without some modifications).
[0029] For purposes of determining whether a signal is a sensing or communication application signal, it may be sufficient to identify, for example, whether the signal is continuous and has a particular waveform, or whether the signal is a periodic signal, without determining the period of the signal. However, it should be noted that the disclosure is not limited in this respect and the period may also be identified, which may contribute to the reliability of the determination and may be used to further identify the sensing application. Detecting the type of application: Frame structure
[0030] Depending on the requirements of the sensing application, different frame structures can be used for wireless sensing. The sensing transmission may have a known frame structure in some wireless communication standards. This frame structure may be similar to the data transmission frame structure, except that in some cases, the frame structure may have a more suitable pilot or training sequence that may be used for sensing and may also support data transmission, as shown in the second and third frame structures of FIG. 4.
[0031] In particular, FIG. 4 shows an exemplary communication transmission frame structure in the first row, which includes a preamble portion, a data portion (where payload or higher layer signaling may be transmitted), and two training sequence (TRN SEQ) portions that may be used for channel estimation and / or synchronization, etc. The second row shows an exemplary sensing transmission frame structure. As can be seen, no data is transmitted in this frame structure. Instead of the training sequence TRN SEQ, a sensing sequence SENS SEQ is provided. In the third row, a joint sensing and communication (JSC) transmission frame structure is used, which is similar to the sensing frame structure, but also includes a DATA portion as shown in the communication frame structure. However, it should be noted that these three exemplary frame structures are only simplified examples, and an actual frame structure may include more portions (fields) carrying various preamble portions, training portions, sensing portions, data portions, and possibly even signaling portions or other portions (e.g., padding portions).
[0032] Additionally, devices that do not comply with wireless communication standards may use different frame designs that may differ from the frame design for data transmission. Detecting the type of application: Further features
[0033] In addition to the period and frame structure, any kind of feature that is different in the communication signal and the sensing signal can be used together with the period and frame structure to distinguish or contribute to the distinction between them. For example, the waveforms may be characterized by various different standards and may even be different in the sensing system and the communication system. In summary, as a first step, according to some embodiments, it may be determined whether the signal is a signal generated by a sensing application or a signal generated by a communication application. In some example implementations, the detection may also distinguish between the communication application, the JSC application, and the sensing application. Such a distinction may be useful for identifying patterns of resource usage and identifying the source application or standard or device.
[0034] If the signal is a sensing signal, the second step (after identifying whether the signal is generated by a communication application or a sensing application) may be to identify the sensing application to which the signal belongs using its characteristics, i.e., period, bandwidth, duration of sensing, and / or the like, some of which are described in more detail below. Application Identification: Periodic
[0035] Depending on the sensing application, the predefined or predetermined period may vary. As noted above, the period may include time intervals during which a signal is transmitted (on) and time intervals during which no signal is transmitted (off).
[0036] The period corresponds to the transmission rate of the sensing transmission and corresponds to the time resolution of the measurements (sensing task). In general, applications for detecting fast changing actions / motions / objects require a higher period or packet rate and vice versa. A low radar packet (wave packet, pulse) rate can lead to a longer maximum range. This may be because the radar listens for reflections between packets. In some implementations, the higher the packet rate, the shorter the time the receiver may need to listen for reflections. Any reflections that do occur may be subject to interference from the transmitted signal. Reflections from distant objects may take longer to reach the receiver. If other chirps or radar signals are transmitted at that time, it may be difficult to determine if a reflected signal is present and which transmitted signal the reflected signal belongs to. Furthermore, the maximum range may be related to the transmission power.
[0037] For example, for a Wi-Fi sensing application using CSI, received signal strength (RSS), or received signal strength indicator (RSSI), 100 packets per second (pps) may be used for gesture recognition, and 5-20 packets / sec may be used for sleep monitoring (position and respiration rate).
[0038] For example, in the case of gesture recognition, patterns of certain types of motions can be extracted from these measurements and used to detect the same motions in new measurements. Alternatively, learning algorithms can be applied to cluster or classify measurements into specific groups / actions. The packet rate here may determine the measurement resolution. If the action being detected is one that changes quickly, a higher resolution may be required.
[0039] In summary, the identification of the sensing application may be performed based on the period of the signal alone or in combination with further features. Application Identification: Bandwidth
[0040] A larger bandwidth may be required to obtain better range resolution and fine-grained CSI information that may be required for certain sensing applications. For example, applications such as high-resolution wireless imaging or detection of small objects and motions require a larger bandwidth. In some wireless sensing processes, the bandwidth may also affect the number and arrangement of subcarriers in the training / sensing sequence. In some applications, different subcarrier frequencies are affected differently by different motions / actions. Thus, a larger bandwidth may enable more information to be collected. Conversely, if the sensing method relies on other measurements such as RSS / RSSI, the device may perform measurements only at the carrier frequency and the bandwidth may not be as important. Intuitively, these sensing applications will use a narrower bandwidth transmission, for example to save power and possibly implementation costs. Thus, the bandwidth occupied by a sensing application may distinguish between sensing applications that target different object sizes as well as applications that do not sense objects at all. Application identification: duration and start / end time of sensing
[0041] It may not be reasonable for a Wi-Fi device to continuously sense the environment if the application does not require it or if the application is terminated. For example, a fully immersive video game may only last for a few hours at a certain time of the day. However, a home / room monitoring application may continue 24 / 7 or until the application is turned off by the user. Similarly, a sleep monitoring application may only continue at night. Monitoring vital signs during exercise may only last for a few hours and is most likely done on a schedule. Therefore, the transmission duration and start / end times of the sensing signal may also be indicative of the sensing application. Identifying Sensing Applications: Further Examples
[0042] In summary, a sensing application may be identified based on one or more or all of the above features. Additionally, additional features such as waveform, time of day, day of the week, or time of year may be used. When referring to identifying a sensing application, it may mean, for example, distinguishing between radar signals and medical / healthcare measurement reports, and / or other sensing applications.
[0043] For purposes of example, the environment for the identification of sensing applications may be a local or area network, which may be a home network or a network covering a street in a neighborhood with several buildings. However, the present disclosure is not limited to any particular scale. For example, the street is active and autonomous vehicles pass by it regularly. The autonomous vehicles are performing radar sensing using FMCW signals. In addition, roadside units of intelligent transportation systems are also sensing the road for vehicular and pedestrian traffic. The buildings may be private homes and offices, public / government buildings, schools, hotels, etc. The buildings include associated wireless environmental sensors such as home / office monitoring sensors, gesture recognition sensors, sleep monitoring sensors, etc., as well as a number of WLAN devices performing sensing for various applications, communications, or other.
[0044] In the following, an example of a private home with home monitoring sensors, gesture recognition sensors, and sleep monitoring sensors is described in detail. Also, there are many communication devices in all frequency bands. In summary, the sensing elements and example parameters of the initial scenario may be as listed below. (Mobile) Self-driving cars: Single Transceiver ●FMCW radar waveform ●Frequency = 6GHz Bandwidth (BW) = 40MHz ● Period = 100 packets / second (Stationary) Roadside machine: Single Transceiver ●FMCW radar waveform ●Frequency = 60GHz ●BW=1GHz ● Period = 100 packets / second Home Monitoring: ●2 APs and 5 sensing / responding nodes ●Wi-Fi Physical-layer Protocol Data Unit (PPDU: Physical-layer Protocol Data Unit) frame format ●Frequency = 2.4GHz ●BW=20MHz ● Period = 100 packets / second Sleep monitoring ●One transmitter and one receiver ●Wi-Fi PPDU ●Frequency = 60GHz ●BW=20MHz ● Period = 10 packets / sec Gesture Recognition ●1 AP and 5 sensing / responding nodes ●FMCW radar waveform ●Frequency = 60GHz Bandwidth = 120MHz ● Period = 100 packets / second
[0045] As can be seen from the above example, there are one or more features that allow for differentiation between applications. It should be noted that the above example is fictitious and the measurements may be altered, which may make accurate differentiation easier or more difficult. It should be noted that these five features (number of transmitters and / or receivers, frame structure / waveform, carrier frequency, bandwidth, and period) are merely exemplary. In general, depending on the required resolution for identifying the sensing application, one or more of these five features and / or any other feature that can distinguish (or contribute to distinguish) between the sensing applications may be used. In an exemplary and non-limiting implementation, the features may be detectable as described above. With regard to the number of transmitters and / or receivers, they may be detected by RF fingerprinting or other identification techniques used in PHY authentication to identify the transmitting device or AP / STA.
[0046] The transmission parameters as above may be used for null data packet (NDP) sounding and therefore may correspond to sensing operation parameters. Since channel measurements are typically performed by NDP, negotiation of NDP transmission parameters may be important to meet the performance requirements of the sensing application.
[0047] Within an environment, multiple sensing sessions belonging to unique sensing applications may occur simultaneously. Due to the presence of several devices with overlapping signal coverage, various configurations may exist. Some exemplary configuration options are: a) Standalone devices such as self-driving cars that sense the environment primarily through their own sensors b) AP-non-AP STA, non-AP STA-non-AP STA (point-to-point, P2P) sensing, and / or integrated sensing and communication. c) AP-multiple non-AP STAs, non-AP STA-multiple non-AP STAs (Point-to-Multipoint, P2M) sensing, and / or integrated sensing and communication
[0048] As already mentioned above, some example types of sensing (based on measurements) may include: Radar-based sensing of the above (which may be based on measuring delay, or delay and Doppler shift) CSI-based sensing RSSI-based sensing SNR-based sensing (e.g., sensing beam SNR)
[0049] For trigger-based sensing, sensing operation parameters may be negotiated during the sensing session setup. This may have the advantage that the parameters are only transmitted once during the session. On the other hand, they cannot be adapted to changing requirements during the session. To do this, the session must be terminated and renegotiated.
[0050] Another option is to negotiate the sensing operation parameters during the measurement setup, allowing the sensing operation parameters to be transmitted as frequently as required, while the responding STA (RSTA) may not be able to support these parameters.
[0051] In an advantageous implementation of a sensing device such as a STA or an AP, negotiation of operating parameters is forced to be performed at the sensing session setup and optionally at the measurement setup. If the operating parameters are not (re)defined at the measurement setup, they are assumed to remain unchanged compared to the operating parameters negotiated at the sensing session setup. Thus, an exemplary STA or AP is configured to negotiate sensing operating parameters at the start of a session and may be further configured to negotiate operating parameters within the measurement setup. For example, conditions based on channel conditions, etc. may exist according to which the sensing device determines whether or not to (re)negotiate the sensing operating parameters originally negotiated.
[0052] Among the operational parameters discussed above, the ones that can have a greater impact on the quality and rate of sensing (and possibly also on the computational complexity) are: A requested sensing rate (specifies how often sensing is performed, which may be given, for example, by a minimum number of frames per second transmitted by the transmitting device) Bandwidth or channel width (e.g., specifying an uninterrupted (continuous in the frequency domain) transmission bandwidth) ● Beamwidth (e.g., relative beamwidth (which is given by a level, e.g., for bands above 60 GHz, three levels: coarse, medium, fine)) Transmit output power (e.g., specifying the minimum and maximum transmit output power allowed for transmission)
[0053] These sensing operational parameters and their effects are discussed in more detail below.
[0054] The requested sensing rate may be expressed by the frequency of the NDP and the corresponding measurements, which determines the time domain resolution of the measurements. Different sensing applications may have different requirements for the time resolution. For example, gesture recognition requires more frequent measurements (e.g., 100 frames / sec) than the above sleep monitoring (which still works well with 5 frames / sec). If there are too few NDPs and corresponding measurements, the measurements may become stale (out of date), which may be due to changes in channel conditions within the NDP repetition period. On the other hand, if there are too many NDPs and corresponding measurements, network traffic may increase unnecessarily. Measurements beyond a certain number of measurements per hour may not further contribute to a significant improvement in sensing performance. It should be noted that transmitting NDPs at a strict (fixed) rate may not be appropriate due to equipment imperfections, lack of a free channel at the time, and / or possible collisions or further factors. For similar reasons, the time interval between successive transmissions may not always be exactly feasible. Therefore, a meaningful operational parameter specifying the requested sensing rate may be the minimum number of frames per second transmitted by the transmitting device. Alternatively, or in addition, an average number of frames per second may be specified.
[0055] Bandwidth affects the range resolution of radar sensing. The higher the bandwidth, the higher the resolution. In autonomous vehicle and V2V scenarios, radars using 802.11p PPDU have been studied and it has been found that the allowed 10 MHz bandwidth is not enough for the typically desired range resolution (and therefore target separation), and that around 100 MHz may be sufficient. Therefore, the IEEE 802.11bf amendment can be used in such applications, if the bandwidth is negotiable. Bandwidth also increases the number of measurements per frame for CSI-based sensing, so more information is captured even when compressed CSI metrics are reported.
[0056] Determining or standardizing a quantitative value of beamwidth may be highly hardware-dependent. Therefore, it is advantageous to determine a relative or qualitative term that reflects the capabilities of the transmitting device. For example, as described above, the beamwidth of a transmission in a bandwidth above 60 GHz is the beamwidth of the NDP relative to the capabilities of the device. Such relative values may be, for example, a coarse beam, a medium beam, and a fine beam. However, the present disclosure is not limited thereto, and in general, there may be only two levels (e.g., wide and narrow), or four or more values. It is also noted that in general, the embodiments of the present disclosure may also be implemented by directly indicating the beamwidth, rather than the relative values described above.
[0057] In radar-based sensing, the transmit power is directly related to the range or distance that can be measured. Some applications may only be interested in objects within a certain radius from the transmitter. Having the transmit output power as an operating parameter prevents unnecessary reflections and calculations, thus eliminating these reflections, saving power, reducing interference, improving spectrum reuse, etc. As with bandwidth, the NDP may include the Tx power in its signaling. However, if the sensing transmitter is not the same as the sensing receiver, it may still be necessary to negotiate the transmit power. Determining a fixed value for the transmit output power may not always be possible due to other constraints dictated by standards or interference mitigation techniques. To provide an operating parameter, the transmit power may be negotiated (and thus defined) as the minimum and maximum transmit output power of the signal required to achieve the sensing performance.
[0058] Generally, according to an embodiment of the present disclosure (which may be combined with the above-described embodiments relating to coexisting wireless sensing applications and integrated sensing and communication applications), negotiation of operational parameters is performed at the sensing session setup. Furthermore, the negotiation of operational parameters may be performed at the measurement setup. For example, the NDP transmitting and / or receiving devices may initiate negotiation at the measurement setup based on some event, such as measurement of significantly changing channel conditions. Whether such an event has occurred may be determined based on conditions such as changes in channel conditions or crossing of a certain threshold for received power, etc.
[0059] If an operating parameter is not defined in the measurement setup, it is assumed to be unchanged.
[0060] One or more (or all) of the following parameters are advantageously negotiated in the sensing session setup phase, and may optionally be negotiated in the measurement setup phase: Required sensing rate Bandwidth or channel width Beam width ●Transmission output power
[0061] It should be noted that the present disclosure does not preclude adding further parameters, for example the number of transmit and receive antennas can provide further information about the available hardware in addition to the beamwidth.
[0062] According to a particular exemplary implementation, the requested sensing rate is the minimum number of sensing frames per second transmitted by the transmitting device. The bandwidth is the contiguous bandwidth of the sensing transmission. The beam width of the transmission in a bandwidth above 60 GHz is the beam width of the NDP relative to the capabilities of the device. There may be values indicating three possible widths that can be distinguished (negotiated). The transmit output power may be the minimum and maximum transmit output power of the signal required to achieve sensing performance.
[0063] The table below shows known example use cases that show how different values can affect sensing or communication performance.
[0064] In particular, Table 1 relates to the required sensing rate. It shows the ratio of sensing transmissions to other transmissions for some applications. For example, for presence detection, sensing transmissions can cause up to 2% network load, which is related to the period and thus to the required sensing rate. [Table 1]
[0065] Table 2 below relates to continuous bandwidths or channel widths. In particular, the bandwidth B has the following effect on the range resolution (which relates to the resolution) ΔR, ΔR=c / (2B), and the maximum velocity v_max=λB / 2, where λ is the wavelength of the signal: [Table 2]
[0066] Furthermore, Table 3 shows the effect of beam width on angular separation (resolution) and therefore on various applications. For a bandwidth of 60 GHz (and beyond), the angular resolution Δθ is:
number
[0067] Table 4 shows the maximum range required for a particular application, which is proportional to the transmit power. [Table 4]
[0068] As can be seen from the above exemplary sensing application requirements (or more generally characteristics), they can distinguish a wide variety of different sensing applications. Thus, bandwidth, sensing rate (period), beam width, and / or transmit power may be used as characteristics input into sensing application identification as described in any of the embodiments and exemplary implementations of the present disclosure.
[0069] Below, we provide an example where the identification of a sensing application is used for its own sensing. For example, the sensing device and the communication device are first transmitting their own signals (for simplicity, we assume that there is no interference with each other; if there is interference, which is the normal case, the results may be affected in terms of identification accuracy). The roadside unit is continuously transmitting signals since it is a critical sensing application. The gesture recognition application is not critical and is an on-demand application. When the gesture recognition application is started, the sensing device first takes and analyzes a snapshot of the spectrum in the frequency band it requires (60 GHz). It notices that there are two periodic transmissions from the sleep monitoring device and the roadside unit. The other signals are discarded, and the gesture recognition device further analyzes these signals and compares their parameters with the transmission parameters it requires. Of the two transmissions, the roadside unit's transmission is more suitable for the gesture recognition application since it applies a similar waveform (FMCW radar type), frequency, etc. Therefore, the gesture recognition application may utilize the roadside unit's signal rather than transmitting its own signal. This is possible since some properties (waveform, frequency, period) of both applications are similar. Only the bandwidth requirements may differ. For example, the bandwidth of the roadside unit may typically be wider than that of gesture recognition. Therefore, the bandwidth of the roadside unit signal is sufficient for gesture recognition. As a result, the spectrum may be used more efficiently. Alternatively, without comparing its required and available signal parameters, the gesture recognition application (module) may enter a training phase, utilizing all sensing transmissions it can process (in terms of hardware suitability) and selecting the one that has the least error or best meets the performance metrics of the sensing application.
[0070] Now, in this example, the gesture identification device may need to allow the sensing transmission to continue in order to allow the sensing application to continue smoothly. This can be done by identifying the sensing application. From a dictionary of sensing signal parameters and applications, or by learning techniques, the gesture recognition device can recognize or determine that the sensing signal being used belongs to a critical sensing application, so the sensing transmission continues. When referring to a dictionary of sensing signal parameters and applications, we mean a pre-stored list populated with the application identification and the values of the features such as frequency, bandwidth, waveform, etc. mentioned above. The identification of the sensing application may be performed using a statistical algorithm. Alternatively, a machine learning or deep learning algorithm may be used to take the signal as input and classify the sensing application that emitted the signal as one of a set of pre-defined sensing applications. A hybrid approach is conceivable where the value of the features, rather than the signal as such, is fed to a trained (trainable) module to output the identification of the sensing application (classifying the sensing application that emitted the signal).
[0071] Similar embodiments and exemplary implementations may also be used for signal intelligence. In other words, enemy applications can be identified and some warning can be given before they occur. For example, if we learn by detection that the enemy wants to sense the motion of our soldiers or people in general, it is clear that they are trying to spy. Therefore, some warning can be given to prevent possible harm.
[0072] The present disclosure can also be used as a predictive control mechanism. For example, an AP can identify a sensing application and check if it matches a previously predicted signal. Such previous predictions can be present for any purpose in cognitive radio, e.g., more efficient use of spectrum, resource allocation, etc. To predict the spectrum, any of the methods presented in Aygul et al. can be used. On the other hand, when predicting a sensing application, the period of the signal may be known, which provides information about the future spectrum. Thus, it can be checked if the predicted spectrum (e.g., predicted by the method of Aygul et al.) matches a predicted spectrum obtained by a period-based technique.
[0073] As another example, in the environment defined above, other sensing devices or applications may enter the area. First, existing sensing applications can be identified and predicted. If similar or identical applications exist, these sensing signals can be utilized for the sensing application, and no new signals need to be generated. This may result in less power required and less interference. Also, the spectrum can be used more efficiently.
[0074] In summary, as indicated above, the present disclosure is by no means limited to the spectrum (occupancy) prediction first mentioned. However, spectrum prediction is also one of the applications to which the sensing application identification of the present invention can be applied, as provided below in an exemplary implementation. Sensing application prediction may be used for spectrum occupancy prediction (as well as for other purposes such as surveillance, threat and espionage detection). Spectral Prediction
[0075] Predicting sensing applications can be as important as identifying them. This has several advantages, which are detailed in the following sections. Predicting sensing applications has two steps. The first step is to predict spectrum occupancy. For this, the techniques used in Aygul, MA, Nazzal, M., Saglam, MI, daCosta, DB, Ates, HF, and Arslan, H. (2021), "Efficient Spectrum Occupancy Prediction Exploiting Multidimensional Correlations through Composite 2D-LSTM Models," Sensors2021, 21, 135: https: / / dx.doi.org / 10.3390 / s21010135 (which is incorporated herein by reference and hereafter referred to as Aygul et al.) can be used, but the applicable techniques are not limited to the specific techniques defined therein. If the spectrum is vacant (available), this means that there will be no sensing applications in the future. If it is filled (occupied), then it needs to be determined whether it is a sensing signal or another signal such as a communication signal present in the spectrum. The features and techniques described above for distinguishing between communication signals and sensing signals can be used for this purpose. If it is a sensing signal, then its application (the application that emitted the sensing signal) is identified. To predict the sensing application, the techniques and features defined for the identification of sensing applications can be used. A block diagram of this is shown in Figure 5.
[0076] In particular, the input to the trained module 520 is a measurement of the spectrum 510. In this exemplary diagram, the spectrum 510 is represented as a two-dimensional time-frequency grid, as is common in time-frequency resource-based systems, such as orthogonal frequency division modulation (or multiplexing) or generally various other frequency division multiplexing (FDM) approaches. The trained module 520 in this exemplary implementation is a learning module, such as a machine learning module or a deep learning module, trained for the specific task of spectrum occupancy detection. The result 530 of the trained module is a prediction of the spectrum occupancy of a particular spectrum portion for a particular time instance or for some subsequent time instances. Based on this result, in an evaluation step 540, if the spectrum is unoccupied (represented by a binary zero "0" in this example), no further action is taken. If the spectrum is occupied (represented by a binary one "1" in this example), it is determined whether the signal occupying the spectrum is a communication signal or a sensing signal. If, in evaluation step 550, the signal occupying the spectrum is a communications signal, then no further processing is necessary, at least in the context of this disclosure, which is focused herein on identifying sensing applications. If, in evaluation step 550, the signal occupying the spectrum is a sensing signal, then the sensing application that generated the signal is further identified.
[0077] Instead of first finding the spectrum occupancy and then finding the sensing application, a learning algorithm such as deep learning can be used to do both. Deep learning is capable of doing this integrated task. Details of an example of an integrated spectrum occupancy and sensing application are provided below.
[0078] There are two phases: training and testing. In the training phase, a dataset is first collected. This dataset includes a set of spectral measurements, the binary occupancy of these measurements, and the class of sensing applications. After training and testing, the inference phase (also called the production phase) corresponds to the actual operation of the trained modules for the identification and / or prediction of sensing applications. It is noted that the present disclosure is not limited to a particular network modularization. For example, it is conceivable to perform the recognition of the application type (sensing / communication) in two steps as described above. This may be performed by two different separately trained modules (e.g., two different neural networks or long short-term memories, etc.). It may also be performed by mixing machine learning (trained) modules and statistical algorithms. However, it is also possible to train a module to do both, i.e. to determine, based on an input, whether the input is a communication signal or one of a set of predefined sensing applications. In the following, the latter approach is discussed as an example.
[0079] Prior to training, the training data set can be obtained in various ways. For example, a binary spectral occupancy may be obtained by thresholding power spectral density (PSD) measurements (or other power or amplitude related measurements) with respect to a certain threshold. In some embodiments, the threshold is adjusted according to the power of the thermal noise. The noise level may be determined by measurements or estimated as known in the art. To account for unexpected effects and variations, a 3 dB margin may additionally be considered to determine the final threshold. If the measured PSD in a certain frequency band is above (exceeds) this threshold, this frequency band is reported as occupied. On the other hand, if the measured PSD in this frequency band is below (or equal to) the threshold, this frequency band is reported as unoccupied.
[0080] After the collection of measurements, the input data set can be obtained in time, frequency and space, as described in Aygul et al. As for the output, two different tasks may be learned: the first is the spectrum occupancy prediction, and the second is the sensing application. The spectrum occupancy prediction can be expressed by a binary value of one unit. In other words, for each spectral part, a binary value specifies whether this part of the spectrum is occupied or not. For the identification of the sensing application, since there are several possible sensing applications, the number of output units needs to be the same as the number of sensing applications to represent them all. It should be noted here that if the spectrum is assumed to be occupied (or there is prior information), only the sensing application as the output data set can be used. Otherwise, the spectrum occupancy can be used as a class of sensing application. Once the data set is obtained, the machine can be trained according to the input data set and the output data set. Then, in the testing phase, the spectrum occupancy prediction and its application can be found. It should be noted here that the spectrum can be predicted instead of the spectrum occupancy. In that case, the output unit of the spectrum occupancy is changed to the spectrum prediction, for example, from a binary value to a numerical value.
[0081] 6A and 6B show diagrams of a typical sensing application, where a wireless sensor (WS) transmitter (Tx) 610 is a wireless sensor transmitter and a WS receiver (Rx) 690 is a wireless sensor receiver. The WS Tx and WS Rx are synchronized and can communicate with each other through wireless signaling or through a wired connection. The WS Tx 610 and WS Rx 690 are located in a restricted area such as a room 600. The concentric circles indicate the electromagnetic field (wireless sensing signal) generated by the WS Tx 610. As can be seen, the WS Rx 690 is within the range of the wireless sensing signal. As discussed above, the wireless sensing signal may be a sensing pulse or a continuous signal such as a radar or sounding signal. However, it may also be a periodic measurement report in a packet format or the like.
[0082] FIG. 6A is only an example configuration of wireless sensing hardware. Other configurations may include two or more transmitters and / or two or more receivers, transceivers, etc. Here, the term transceiver refers to a receiver or transmitter or a combination of both. First, the WS Tx 610 may transmit sensing signals when there is no target (no object present within the monitored object size range) as shown in FIG. 6A. The WS Rx 690 receives these sensing signals and obtains measurements, such as, for example, measurements indicating CSI or any signal strength. Then, if there is a target, shown as a stick figure 650 in FIG. 6B, a change in the measurements indicates it. In other words, the presence of the object 650 changes the channel, which in turn changes the received signal, which can be detected by the WS Rx 690. Note that this is only one example where there is a cooperating transmitter and receiver WS. However, the disclosure is not limited in this respect, and instead of detecting a signal transmitted from a particular transmitter at a location different from the location of the receiver, some sensing applications may rely on a transceiver including a transmitter and receiver located at the same location, for example a pulsed radar, in which case the detected signal is a signal reflected from a detected object.
[0083] 7 and 8 show further different exemplary sensing scenarios. Different devices performing sensing and communication in the area may be only sensing devices, as shown as WS Tx and WS Rx. Alternatively or in addition, sensing may be performed between an AP (or other network controller / coordinator) and a station (STA), or between a STA and other STAs. At the same time, other network devices may communicate. The sensing devices may be line of sight (LOS) as shown in FIG. 7, or non-line of sight (nLOS) as shown in FIG. 8. For example, in FIG. 7, WS Tx transmits a first sensing signal (dotted line). AP1 transmits another sensing signal (solid line). STA1 may receive a sensing signal from AP1, while WS Rx may receive a sensing signal from WS Tx. However, in some scenarios, STA1 may also use a sensing signal from WS Tx, and WS Rx may also receive a sensing signal from AP1. Here, receiving means detecting the presence and possibly further processing, i.e., not simply receiving as part of the noise. Furthermore, AP2 and STA2 are in communication with each other, i.e., exchanging communication signals. In this simplified example, all these devices are within LoS. Some or all of these devices may operate in the same spectrum or at least partially overlapping spectrum.
[0084] As shown in the figure, the sensing signals may have different periods (indicated by different densities of the concentric circle portions showing the sensing signals). In this case, it may be beneficial to identify the sensing application without direct communication or coordination with other wireless devices. In response to the detection of the sensing signal, the device may use a signal suitable for itself rather than transmitting its own sensing signal. For example, when STA1 detects that WS Tx is transmitting a sensing signal, it may use that sensing signal in addition to or instead of the sensing signal from AP1. It is even possible that AP1 detects the sensing signal from WS Tx and stops transmitting its own sensing signal, since one sensing signal may be sufficient. Or vice versa, WS Tx detects that AP1 is transmitting a sensing signal and stops transmitting its own sensing signal. As will be apparent to those skilled in the art, various implementations of adjustment and adaptation of the sensing environment may be provided when a sensing application (signal from a sensing application) is detected in the area.
[0085] FIG. 8 illustrates a scenario where there is not necessarily LoS between some or all of the devices. In this figure, in fact, none of the devices STA3, STA4, and STA5 present in the sensing area have LoS to other devices, yet the sensing signals may still be received by the sensing signal receiving device. For example, STA5 has no LoS to STA3 and STA4, but may still receive their sensing signals. Therefore, STA5 may decide to turn off its own signal and use the sensing signals of STA4 or STA3, etc. By using the sensing application identification and prediction techniques discussed above, the device can know which sensing application the signal is from, predict its duration and future spectrum usage, and schedule (allocate resources) its own signals so that interference does not occur, or utilize these signals for its own sensing application.
[0086] In the above description, some specific examples are given. However, the present disclosure is not limited to those examples. Rather, variations and modifications may be advantageous in some scenarios. For example, different features of sensing signals and communication signals may be used to distinguish them, such as frame structure, period, resolution, RSS / RSSI value, or some sensing features defined in future standards, such as periodic channel access mechanism, back-off operation, special sensing sequence or waveform, etc. RSS / RSSI may be used instead of PSD to detect spectrum occupancy and / or measure signals to determine whether the signal is a communication signal or a sensing signal, or to determine the sensing application that originated the signal. For example, RSS / RSSI measurements are readily available in most communication devices and can provide a rough quantification of user activity, i.e., spectrum usage. The term "user" here refers more broadly to a specific application running on the device.
[0087] Some embodiments of the present disclosure may be used for military defense purposes or for civilian applications such as home monitoring, home appliances, entertainment, etc. For example, it can be known which sensing application is used by other parties. In addition, the location of the sensor can be found based on the PSD value and the application requirements. For example, if the sensing application is known (meaning that the sensing application is known and the trained module is trained to distinguish it), it is possible to obtain knowledge about the RSS / RSSI values, since different applications require different RSS / RSSI values. The RSS / RSSI depends on the distance between the transmitter and the receiver. Therefore, it is possible to at least approximately predict the location or relative location of the sensing device (e.g., the transmitter).
[0088] There are various deep learning (DL) techniques that can be used in the present disclosure for spectrum occupancy prediction and / or sensing application detection, identification and / or prediction. Examples include multi-layer perceptron (MLP), recurrent neural network (RNN) such as long short-term memory (LSTM), or convolutional neural network (CNN). Statistical methods can also be used instead of DL techniques. Sensing signal identification and prediction are not necessarily performed only by ML and DL techniques. Sensing application identification and prediction can be performed in several domains such as time, frequency, space, or code domains, jointly or separately.
[0089] The advent of DL reduces the need for human-based feature extraction / creation / engineering. Thus, it is possible to directly apply the above methods to real measurements (PSD or RSS / RSSI) without feature extraction. Conversely, the "deep" attributes of the ML used may be reduced at the cost of adding more feature extraction. In extreme cases, DL may be avoided altogether and ML may be used if it is possible to extract discriminatory features. Deep generally means that two or more processing layers are used, although machine learning is possible with a single layer. In the design of the training and trained modules, other types of extensions may be further considered, rather than extending the received signal in the form of a vector fed to the classifier. For example, it can be formed as other two-dimensional (2D) grids (e.g., time-frequency grids). Some Effects and Advantages
[0090] One of the advantages of the present disclosure may be the provision of improved spectrum efficiency. For sensing applications, there may already be a signal that can be used for sensing, so there may be no need to generate a new signal. Thus, spectrum waste may be avoided and the saved spectrum can be used for other things. Power efficiency may be another advantage of detection and / or identification of sensing applications. Since there may be no need to transmit additional signals, less power is wasted. This may be crucial for IoT and other power-constrained wireless sensing devices. Some of the above-mentioned embodiments may facilitate resource allocation. Existing WiFi standards lack control between inter-network APs. Therefore, coordination techniques between different APs are needed to eliminate interference while efficiently utilizing the spectrum. However, interference may also be reduced by reducing the transmission of sensing signals based on detection, identification, and / or prediction of sensing signals.
[0091] Knowledge of the period of the sensing signal may be used to aid in the spectrum occupancy prediction: if a signal is identified as a sensing signal and its period can be determined, future (near future) transmissions may be predicted and used for more accurate spectrum occupancy prediction.
[0092] Learning of the sensed information may also be performed: if we know which signals are sensing signals, we can process the signals to learn information about the environment just by using transmissions from other devices.
[0093] Blind signal analysis may be useful for (e.g., military) defense or civilian applications, where the information of the sensing signal may be used for military reconnaissance. For example, one may know what the other party wants to sense and what they are sensing. This can be done by analyzing the sensing signal using blind signal analysis (or other methods). In communication, the signals being communicated are typically dedicated to individual users, so if a node is not a rogue node (e.g., an eavesdropper), it may not be possible to use the signals of others. However, in sensing, any kind of signal that meets the requirements of the sensing application may be used, since the signals of sensing typically do not carry private / confidential data or information.
[0094] Blind signal analysis may also be used in commercial applications, and some embodiments of the present disclosure may be used to enable or facilitate CR and software-defined radio (SDR), public safety radio, improving spectral efficiency, optimizing the operation of various networks sharing spectrum, testing, and measurement of communication and sensing devices. This can be done by analyzing sensing signals using blind signal analysis (or other methods).
[0095] All of the above advantages help facilitate the integration of multiple sensing devices and unified sensing and communication (JSC) devices into current and next generation wireless networks.
[0096] Generally speaking, as mentioned above, according to one embodiment, a method for identifying a sensing application is provided. A flow diagram of the method is shown in FIG. 9. The method includes steps 910 of acquiring a received wireless signal and step 920 of estimating, by a trained module, the presence of (typically one or more) sensing signals in the received signal, the sensing signals being signals generated by the sensing application. In this context, acquiring refers to receiving via a wireless interface in some embodiments of a device, e.g., a device that also includes a transceiver front. However, a device of the present disclosure may be separated from an actual transceiver and simply have an interface for communicating (controlling and acquiring / providing received / transmitted signals). For example, acquiring may then be acquiring from such an interface or storage / buffer, etc. The trained module may be an ML or DL module, or another type of trained module. In general, the present disclosure may replace the trained module with a module that implements some kind of statistical or other decision algorithm, for example, based on criteria that distinguish between sensing applications based on some predefined characteristics (e.g., one or more of the above characteristics including period, carrier, bandwidth, waveform, etc.) A combination of trained modules and some decision-based modules is contemplated.
[0097] In other words, instead of or in combination with a trained (e.g., ML / DL) module, other types of methods, such as statistical or deterministic methods, may be used for estimation. For example, if a signal is observed to be periodically repeating, ML or DL methods may not be needed to detect the presence of such a sensing signal. It may be determined deterministically whether sensing is occurring or not. Similarly, in the case of sensing application identification, the identification may be performed considering a set of predefined features and their values for a particular application. Furthermore, if a sensing application uses specific header information for detecting the network type, this or other header information may be detected and deterministically used to determine the identification of the application. There may be drawbacks in defining parameters (features and their values) and application sets for each environment. Thus, depending on the deployment scenario, for example, in more complex scenarios where deterministic or probabilistic distinction is difficult or complex, a trained module may provide better results.
[0098] 9 further includes performing 930 wireless receiving, transmitting, or sensing based on the inferred presence of the sensing application. Some examples of such actions taken in response to or according to the inference include the following example implementations.
[0099] Based on the presence or absence of the sensing signal at multiple measurement time instances, a period of the sensing signal is determined. This may be performed by capturing the presence or absence detection of the sensing signal in the received signal and then analyzing the past captured signals. This may also be performed by correlating the waveform of a particular sensing application with the received signal, which may include noise and interference from other signals. This may be performed by a trained module. The period of the signal may be further used, for example, as an additional input or feedback to the spectrum occupancy prediction, to improve the spectrum occupancy prediction. Based on the spectrum occupancy prediction, signal scheduling may be performed in the wireless device (STA, AP, WS, etc.). For example, if the spectrum occupancy prediction (e.g., based on the period of the identified sensing signal) indicates that a particular portion of the spectrum is used by the identified sensing application, the wireless device may decide to transmit a signal (communication or sensing or both) in a different portion of the spectrum (than predicted by the sensing application) and / or at a different time instance. It should be noted that in some embodiments, the identification of the sensing application and / or the spectrum occupancy prediction may be performed in the same device using the identification result. However, there may be embodiments in which a first (e.g., controlling) device performs spectrum occupancy prediction and / or detection or identification of sensing applications. In that case, such a device may broadcast (or multicast or simply transmit) the results to one or more devices in the area. They may then adapt their behavior (e.g., channel access, transmission of sensing signals or reception and evaluation of sensing signals) accordingly.
[0100] In other words, the detected sensing application may be used to improve spectrum occupancy prediction. Furthermore, the characteristics of the detected sensing application may be used in scheduling. A device having knowledge about the presence of a particular sensing application (and prediction of future presence in a particular resource) may schedule its data or sensing signal to not use the same frequency and / or bandwidth and / or time. Alternatively or additionally, a device having a prediction of a sensing signal transmitted by a particular transmitter may use the sensing signal for its own sensing, such as evaluating the presence of an object in an area. In an unencrypted application, if the sensing application reports environmental or channel conditions, the report may be read by multiple devices that detect and predict the presence of such a signal. Alternatively or additionally, a transmitter of a sensing signal may decide to stop transmitting a sensing signal or reduce the period of its sensing signal if it has knowledge about a similar sensing application detected / identified in the same area (e.g., nearby).
[0101] In some exemplary implementations, a sensing application is an application that generates a sensing signal at a pre-set period or as a continuous transmission signal. Such applications may have characteristics that allow them to be easily distinguished from communication signals (repetitive nature and / or possibly a particular waveform from communication signals). Thus, detection and identification of such sensing application signals may be performed with some degree of reliability and may be used in any of the above exemplary scenarios.
[0102] The period may be preset, for example, at the transmitter of the sensing application, by a person (e.g., at a health monitoring device) or by an application, etc. When referring to continuity here, it should be noted that a continuous signal is continuous under a minimum time resolution. For example, a continuous signal may be an on / off signal pattern or a sinusoidal pattern that is continuously transmitted, etc., and does not mean that the signal must be constant. The minimum time resolution may depend on the capabilities of the hardware. In other words, even continuous transmissions typically have a periodic nature. The term continuous is used herein as in the art, and continuous means that there are no discernible gaps in the sensing signal, such as a radar chirp signal transmitted in one direction. Furthermore, the period mentioned above is for the case where there is a predefined time gap between the sensing signals (wave packets). Of course, the predefined gap may have some variation under imperfect hardware or wave propagation conditions (at the receiver).
[0103] In an exemplary implementation, the sensing application is one of radar sensing, wireless sensing, wireless local area sensing, and channel state information sensing.
[0104] Wireless sensing may include channel or spectrum sensing. Wireless local area sensing may include beamforming sensing. It should also be noted that the term wireless sensing may be considered more general and includes any type of sensing that uses wireless signals for the sensing itself or for (possibly periodic) reporting of sensing results. Such wireless sensing or wireless local area sensing may also include sensing using Wi-Fi technology. This usage may relate to devices (using STAs or APs), frequency bands, frame designs, communication-related measurements such as CSI, CIR, CFR, RSS, RSSI, etc. Some examples of wireless sensing and wireless local area sensing are provided, for example, in Yongsen Ma, Gang Zhou, and Shuangquan Wang, 2019, “WiFi Sensing with Channel State Information: A Survey,” ACM Comput. Surv, 52, 3, Article 46 (June 2019), p. 36, https: / / doi.org / 10.1145 / 3310194.
[0105] In some exemplary implementations based on any of the above embodiments or examples, estimating further includes determining the sensing application as one of a plurality of predefined sensing applications (e.g., identifying the application or application type). As described above, the detection of the presence of a sensing application and the more precise identification of such an application (or application type) may be performed jointly or separately. Here, the term application identification should be understood broadly. For example, in some implementations, it may mean a distinction between a set of predefined application types, such as health / fitness / sleep monitoring on the one hand and object presence / motion / velocity on the other hand. However, in some implementations, a finer distinction may be performed, such as the one illustrated in the detailed embodiment above. It may be possible to further distinguish gesture detection signals from health or sleep monitoring or signal reporting, etc.
[0106] In an exemplary embodiment, the sensing application is an application that generates a sensing signal with a preset period. Determining the sensing application then includes estimating the preset period based on a previously estimated presence of the sensing signal in the received wireless signal, and using the estimated preset period to identify the sensing application from among a plurality of predefined applications. The identification may be performed by a trained module (e.g., ML, DL). The detection of the presence (and possibly also the period) may be performed by a deterministic or statistical approach. However, as already mentioned above, other configurations are also possible, including a combined detection and identification by a trained module, two separate trained modules, or a pure statistical / deterministic approach.
[0107] For example, determining the sensing application is performed based on one or more of the characteristics, which are frame structure, bandwidth, duration of sensing, start time of sensing, end time of sensing, and waveform. The one or more characteristics are different for at least two of the multiple predefined sensing applications. Then, some inference can be made about the identity of the current application. The characteristics refer to features as exemplified in the above embodiments, including, for example, quantifiable parameters whose values are different for at least two applications.
[0108] As noted above, the presence detection and identification may be performed sequentially. In such a case, in one embodiment, the determining of the sensing application as one of a plurality of predefined sensing applications includes: - executed if the presence of a sensing signal in the received wireless signal is confirmed by the estimating step; - Not performed if the presence of a sensing signal in the received wireless signal is not confirmed by the estimating step.
[0109] This has already been described with reference to Figure 5. In particular, in an exemplary implementation, obtaining the received wireless signal includes measuring one or more of a power density spectrum, a received signal strength, or a received signal strength indicator. The method further includes determining the presence or absence of a communication signal or a sensing signal in the received wireless signal by the trained model. If it is determined that a communication signal or a sensing signal is present in the received wireless signal, estimating whether the signal is a sensing signal is performed. If the signal is a sensing signal, determining a sensing application as one of a plurality of predefined sensing applications is performed.
[0110] It should be noted that although the example described with reference to FIG. 5 showed a binary occupancy, in general, soft values may be used for occupancy detection / prediction. Such soft values may represent the reliability of the occupancy prediction. If it is determined that the sensing signal is not present, the method may include bypassing the estimation (identification) step. If the signal is a communication signal, the method may further include bypassing the application identification decision. Additionally or alternatively, if the signal is a communication signal, the method may include identifying the communication signal according to any method known in the art.
[0111] As noted above, in some embodiments, determining the sensing application and inferring presence are performed jointly by a trained module.
[0112] In some example implementations, predicting future resource occupancy by sensing applications may be performed based on the results of the estimating step. The prediction may be done by a trained module, or by deterministically assuming a certain periodicity, or by a mixture of both. The predictor can be changed based on the situation. When there are many sensing applications to transmit, it may be better to use machine learning. Otherwise, in simpler scenarios with clearly distinguishable sensing applications, parameter-based prediction may work better.
[0113] In any of the above embodiments and examples, performing the wireless reception, transmission, or sensing may include scheduling the wireless reception, transmission, or sensing in resources predicted to be unoccupied by the sensing application. Additionally or alternatively, performing the wireless reception, transmission, or sensing may include determining whether to transmit a sensing signal based on a result (presence and / or identity) of the estimating step, and performing or not performing sensing in response to the determining step.
[0114] Thus, an own use of sensing signals based on sensing identification may be performed, using knowledge of the presence and type of sensing application signals to determine whether to transmit its own sensing signal or to utilize an available signal, as already explained above. Within the network, this can be used to reduce sensing traffic by coordinating with other devices to serve sensing applications using a reduced or even minimal number of sensing transmissions.
[0115] In addition to the above approaches for detection and identification of sensing applications, a training method and device are provided for training a module used for detection / identification. For example, as shown in FIG. 10, a method for training a module for identifying a sensing application is provided. The method includes acquiring a received wireless signal in step 1010 and acquiring a ground truth in step 1020. The ground truth in this embodiment may be a determination of the presence or absence of a sensing signal, if a module for detecting the presence or absence is provided. The ground truth in another embodiment may be an identification of the application that emitted the sensing signal, if a module for identification (or for both presence detection and identification) is trained. The method includes combining an acquired representation of the received wireless signal (e.g., RSSI, or other measurement representative of the signal or its characteristics) and a desired determination of the presence or absence of a sensing signal in the representation. finger The desired indication and sensing application that generated the sensing signal. fingerThe method further includes inputting 1030 the training pairs and one or both of the indications into the module. After inputting the training pairs, the method includes modifying 1030 at least one parameter of the module according to the input. Such an approach may be repeated for multiple (possibly large amounts of) training data (training pairs in supervised learning). It is noted that this training method is merely exemplary and other approaches such as unsupervised learning may be applied as well.
[0116] It should be noted that although the embodiments and examples of the present disclosure are provided with respect to the above methods, corresponding devices providing the functionality described by these methods are also provided. For example, a device for identifying a sensing application is provided. The device may include a processing circuit configured to perform steps according to any of the above methods. For example, the processing circuit may be configured to obtain a received wireless signal and estimate the presence of a sensing signal in the received signal, the sensing signal being a signal generated by a sensing application. The device may further include a transceiver for performing wireless reception, transmission, or sensing based on the estimated presence of the sensing application. Instead of a transceiver, the processing circuit may control an external transceiver to perform wireless reception, transmission, or sensing based on the estimated presence of the sensing application.
[0117] 11 illustrates an exemplary device 1100 that may implement some embodiments of the present disclosure. For example, the device may be a device for detection and identification sensing applications. Such a device may include a memory 1110, a processing circuit 1120, a wireless transceiver 1140, and possibly a user interface 1130. The device may be, for example, (part of) a base station or a terminal / STA, or other device as described above.
[0118] The memory 1110 may store a program that may be executed by the processing circuitry 1120 to perform steps of any of the methods described above. The processing circuitry may include one or more processors and / or other dedicated or programmable hardware. The wireless transceiver 1140 may be configured to receive and / or transmit wireless signals. The transceiver 1140 may also include baseband processing that may detect, decode and interpret data according to some standard or predefined convention. However, this is not required and a device having only sensing applications may implement only the lower one or two protocol layers. For example, the transceiver may be used to perform measurements and communicate with other devices such as base stations and / or terminals. The device 1100 may further include a user interface 1130 for displaying messages or status of the device, etc., and / or for receiving user input. The bus 1101 interconnects the memory, the processing circuitry, the wireless transceiver, and the user interface.
[0119] 12 illustrates an example of a memory 1110 including a module 1160 for acquiring a received signal, a module 1170 for said estimation, and a module 1180 for controlling the transceiver 1140 to adapt wireless receiving, transmitting, or sensing based on the estimated presence of the sensing application. These modules 1160-1180 may be fetched from memory and executed by the processing circuit 1120.
[0120] The above examples are not limiting of the present disclosure, as there are many modifications and configurations that may be used in addition or as an alternative, as briefly described below.
[0121] Similarly, a device is provided for training a module for identifying a sensing application, the device comprising: a representation of a received wireless signal and a desired presence or absence of a sensing signal in the representation; fingerThe desired sensing application that generated the indication and sensing signal finger and a processing circuit configured to input the indication to the module and to alter at least one parameter of the module according to the input.
[0122] This disclosure can be used in any kind of device used for wireless sensing. For example, health monitoring, activity classification, gesture recognition, people counting, through-wall sensing, emotion recognition, attention monitoring, keystroke recognition, air drawing, imaging, step counting, speed estimation, sleep detection, traffic monitoring, smoking detection, metal detection, sign language recognition, humidity estimation, wheat moisture detection, fruit ripeness detection, sneeze sensing, etc. In addition to these applications, embodiments of the disclosure can be used in JSC technology. The disclosure can also be used in sensing applications to support communication applications, such as obstacle tracking for beam management. Thus, devices that can utilize the present invention can be devices such as smart home / office / city / factory, such as kitchen appliances, television sets, smart bus stops, smart office equipment (printers, etc.), lighting systems, WLAN and WiFi devices, etc. Other devices can be standalone wireless sensors such as heart rate monitors, motion detectors, smart watches, etc. In addition to these applications, the present invention can also be used in military operations, such as enemy sensors, to know the presence of enemy devices and what they are sensing, and to take some preventative measures. The present disclosure may be used in network controllers and management devices, such as APs, BSs, edge nodes, extended nodes, among others, for technologies such as CRs and reconfigurable wireless systems.
[0123] In some embodiments, the sensing signal is a continuous or periodic radar signal. In some embodiments, the sensing signal is a signal generated by a sensing application supporting wireless sensing, wireless local area sensing, and / or non-invasive medical sensing.
[0124] In general, wireless sensing is performed by measuring some characteristics of a received signal, while communication is performed by detecting information encoded therein at the transmitter from the received signal, where detection (e.g., demodulation and decoding) is performed using some characteristics of the received signal.
[0125] There are many frame designs, waveforms, and transmission methods used in wireless sensing and JSC. Their use depends on the method of wireless sensing being used. For example, wireless sensing may be performed in one of the following ways: ● Wireless Fingerprinting: Measurements are taken at different locations or by fixed sensors while different actions are being performed. These measurements may be pre-processed and stored in a look-up table with the respective location / action. Radar-based sensing: Sensing signals are transmitted at a specific rate and duty cycle, and reflections of these signals are processed to provide information about the object, such as its range, size, relative speed, material, etc. In radar applications, a waveform or frame design with high transmit power and very low receive power resulting in high autocorrelation and low peak to average power ratio (PAPR) is preferred. Common waveforms are pulsed waveforms or frequency modulated continuous wave waveforms. Frame design and transmission mechanism based factors that can affect radar performance are transmit power, duty cycle, pulse repetition frequency, frequency modulation, carrier frequency, autocorrelation, bandwidth, signal period, launch angle, beam width, beam sweep rate, etc. Environmental factors that can affect radar performance are amount of clutter, physical properties of the object (size, material, etc.), object location, atmospheric conditions (e.g. humidity), etc. Spectral conditions that can affect radar performance are occupancy, number of users, interference, channel conditions (time / frequency selectivity), etc. ● Pattern-based sensing: This is similar to wireless fingerprinting, except that instead of measurements being pre-stored in a look-up table, patterns are extracted from them that represent the action / object to be detected, and the patterns are used to detect the action / object in future measurements. Examples of frame designs and waveforms are Wi-Fi and OFDM Physical Protocol Data Unit (PPDU) packets and orthogonal time frequency space (OTFS) waveforms. Factors related to frame design and transmission mechanism that may affect sensing performance are sensing / training / pilot sequences, placement of these sequences (on which subcarriers), bandwidth, carrier frequency, packet length, packet repetition frequency, launch angle, beam width, beam sweep rate, RF obstructions, etc. Environmental related factors that may affect sensing performance are number of moving objects in the area (stationarity of the environment), number of users / objects to be detected, nature of the motion / action / activity to be detected (large / small displacement, slow / fast movement, etc.), atmospheric conditions, physical properties of the objects / users to be detected, etc. Spectral conditions that can affect radar performance are occupancy, number of users, interference, channel conditions (time / frequency selectivity), etc.
[0126] In general, the present disclosure is not limited to the above three types of sensing.
[0127] Received signal characteristics measured by wireless sensing may include, but are not limited to, time of flight, RSSI, CSI, etc. Transmitted signal parameters that affect wireless sensing performance include, but are not limited to, transmit frequency, bandwidth, waveform, power, training or sensing sequence, autocorrelation capability, pilots, beam angle and width (if beamforming is performed), duty cycle, transmission rate, etc.
[0128] It is desired that wireless sensing devices, communication devices, and JSC devices peacefully operate / coexist within the same (or at least partially overlapping) frequency bands with maximum efficiency in terms of spectrum usage, power, etc., and with maximum sensing and communication performance in terms of throughput, reliability, sensing accuracy, etc. This may be facilitated by detecting and / or identifying sensing signals as described herein. Software and hardware implementation
[0129] The methodologies described herein may be implemented by various means depending on the application. For example, these methodologies may be implemented in hardware, an operating system, firmware, software, or any combination of two or all of these. In a hardware implementation, any processing circuitry may be used, which may include one or more processors. For example, the hardware may include one or more of an application specific integrated circuit (ASIC), a digital signal processor (DSP), a digital signal processing device (DSPD), a programmable logic device (PLD), a field programmable gate array (FPGA), a processor, a controller, any electronic device, or other electronic circuit unit or element designed to perform the functions described above.
[0130] If implemented as program code, the functions performed by the sending device may be stored as one or more instructions or code on a non-transitory computer-readable storage medium, such as the memory 610 or any other type of storage. Computer-readable media includes physical computer storage media, which may be any available medium that can be accessed by a computer, or generally by the processing circuitry 620. Such computer-readable media may include RAM, ROM, EEPROM, optical disk storage, magnetic disk storage, semiconductor storage, or other storage devices. Some specific, non-limiting examples include compact discs (CDs), CD-ROMs, laser discs, optical disks, digital versatile discs (DVDs), Blu-ray (BD) discs, and the like. A combination of different storage media is also possible, in other words, distributed and heterogeneous storage may be used.
[0131] The above embodiments and example implementations provide some non-limiting examples. It is understood that various modifications may be made without departing from the claimed subject matter. For example, modifications may be made to adapt the examples to new systems and scenarios without departing from the central concepts described herein. Selected embodiments and examples
[0132] In summary, some embodiments of the present disclosure relate to the identification of a sensing application. In particular, a wireless signal is acquired and further processed. Then, the presence of a sensing signal in the received signal is estimated, the sensing signal being a signal generated by a sensing application. Then, wireless reception, transmission, or sensing is performed based on the estimation of the sensing signal. The detection and / or identification of the sensing application may be performed by a trained module, such as a machine learning based module. The wireless reception, transmission, or sensing performed according to the result of the estimation may further include channel access, resource allocation, utilization of the detected sensing signal for its own sensing purposes, etc.
[0133] According to one embodiment, a method for identifying a sensing application is provided, the method including the steps of acquiring a received wireless signal, estimating, by a trained module, the presence of a sensing signal in the received signal, where the sensing signal is a signal generated by a sensing application, and performing wireless receiving, transmitting, or sensing based on the estimated presence of the sensing application.
[0134] For example, a sensing application is an application that generates a sensing signal at a preset period or as a continuous transmission signal.
[0135] For example, the sensing application is one of radar sensing, wireless sensing, wireless local area sensing, and channel state information sensing.
[0136] In some exemplary embodiments, the estimating step further includes determining the sensing application as one of a plurality of predefined sensing applications.
[0137] According to an example implementation, the sensing application is an application that generates a sensing signal at a preset period, and determining the sensing application includes estimating the preset period based on a previously estimated presence of the sensing signal in the received wireless signal, and identifying the sensing application from among a plurality of predefined sensing applications using the estimated preset period.
[0138] For example, determining the sensing application is performed based on one or more of the characteristics, the characteristics being a frame structure, a bandwidth, a duration of sensing, a start time of sensing, an end time of sensing, and a waveform, and the one or more characteristics differ for at least two of the plurality of predefined sensing applications.
[0139] For example, determining the sensing application is performed based on one or more of the characteristics, the characteristics being bandwidth, beam width, transmission power, and sensing rate. Moreover, the one or more characteristics are different for at least two of the multiple predefined sensing applications. These characteristics correspond to some operating parameters of the sensing applications, providing a suitable distinction between the various different applications, as described above. Depending on the application one wishes to distinguish, one of the characteristics or a combination of two or more or all of these four characteristics may be used. Such characteristics may be negotiated at the start of (and optionally during) the sensing session.
[0140] In some embodiments, the determining of the sensing application as one of a plurality of predefined sensing applications is performed if the presence of a sensing signal in the received wireless signal is confirmed by the estimating step, and is not performed if the presence of a sensing signal in the received wireless signal is not confirmed by the estimating step.
[0141] In some example implementations, the step of acquiring a received wireless signal includes measuring one or more of a power density spectrum, a received signal strength, or a received signal strength indicator; determining the presence or absence of a communication signal or a sensing signal in the received wireless signal by a trained model; if a communication signal or a sensing signal is determined to be present in the received wireless signal, estimating whether the signal is a sensing signal; and if the signal is a sensing signal, determining a sensing application as one of a plurality of predefined sensing applications.
[0142] For example, the steps of determining the sensing application and inferring presence are jointly performed by a trained module.
[0143] The method may further include predicting future resource occupancy by the sensing application based on the results of the estimating step.
[0144] For example, performing wireless reception, transmission, or sensing may include scheduling the wireless reception, transmission, or sensing within resources that are predicted to be unoccupied by the sensing application.
[0145] For example, the step of performing wireless receiving, transmitting, or sensing includes determining whether to transmit a sensing signal based on the result of the estimating step, and performing or not performing sensing depending on the determining step.
[0146] According to one embodiment, a method for training a module for identifying a sensing application includes: finger The desired sensing application that generated the indication and sensing signal fingerand modifying at least one parameter of the module according to the input.
[0147] According to one embodiment, there is provided a device for identifying a sensing application, the device comprising: a processing circuit configured to acquire a received wireless signal; and infer a presence of a sensing signal in the received signal, where the sensing signal is a signal generated by a sensing application; and a transceiver for performing wireless receiving, transmitting, or sensing based on the inferred presence of the sensing application.
[0148] According to one embodiment, a device for training a module for identifying a sensing application is provided, the device comprising: a representation of a received wireless signal and a desired presence or absence of a sensing signal in the representation. finger The desired sensing application that generated the indication and sensing signal finger A device is provided that includes a processing circuit configured to input one or both of the indications into a module and alter at least one parameter of the module according to the input.
[0149] Additionally, corresponding methods are provided that include steps performed by any of the above processing circuit implementations.
[0150] Still further, there is provided a computer program stored on a non-transitory medium and comprising code instructions which, when executed by a computer or processing circuit, perform the steps of any of the methods described above.
[0151] According to some embodiments, the processing circuitry and / or the transceiver are incorporated in an integrated circuit IC.
[0152] While the disclosed subject matter has been described in detail for purposes of illustration based on what are presently considered to be the most practical and preferred embodiments, it is to be understood that such detail is for this purpose only and that the disclosed subject matter is not limited to the disclosed embodiments, but on the contrary, is intended to cover modifications and equivalent arrangements within the spirit and scope of the appended claims. For example, it is to be understood that the disclosed subject matter contemplates that, to the extent possible, one or more features of any embodiment can be combined with one or more features of any other embodiment. [Item of invention] [Item 1] 1. A method for identifying a sensing application, comprising: acquiring a received wireless signal; estimating, by a trained module, the presence of a sensing signal in the received signal, the sensing signal being a signal generated by a sensing application; performing wireless receiving, transmitting, or sensing based on the inferred presence of the sensing application; A method comprising: [Item 2] 2. The method according to claim 1, wherein the sensing application is an application that generates the sensing signal at a preset period or as a continuous transmission signal. [Item 3] 3. The method according to item 1 or 2, wherein the sensing application is one of radar sensing, wireless sensing, wireless local area sensing, and channel state information sensing. [Item 4] the estimating step further includes determining the sensing application as one of a plurality of predefined sensing applications. The method according to any one of items 1 to 3. [Item 5] the sensing application is an application that generates the sensing signal at a preset period, determining the sensing application, estimating the predetermined period based on the presence of the sensing signal previously estimated in the received wireless signal; identifying the sensing application from among the plurality of predefined sensing applications using the estimated predetermined period; The method according to item 4, comprising: [Item 6] The determining of the sensing application is performed based on one or more of the characteristics, the characteristics being: Frame structure, Bandwidth, Sensing duration, Sensing start time, The end time of sensing, and waveform, and the one or more characteristics differ for at least two of the plurality of predefined sensing applications; Item 6. The method according to item 4 or 5. [Item 7] The determining of the sensing application is performed based on one or more of the characteristics, the characteristics being: Bandwidth, Beam width, Transmit power, and Sensing rate, and the one or more characteristics differ for at least two of the plurality of predefined sensing applications; The method according to any one of Items 4 to 6. [Item 8] determining the sensing application as one of the plurality of predefined sensing applications; is executed if the presence of the sensing signal in the received wireless signal is confirmed by the estimating step; is not performed if the presence of the sensing signal in the received wireless signal is not confirmed by the estimating step. The method according to any one of items 4 to 6. [Item 9] The step of obtaining the received wireless signal further comprises: Measuring one or more of a power density spectrum, a received signal strength, or a received signal strength indicator; determining, via a trained model, the presence or absence of a communication signal or a sensing signal in the received wireless signal; if a communication signal or a sensing signal is determined to be present in the received wireless signal, estimating whether the signal is a sensing signal; If the signal is a sensing signal, determining the sensing application as one of a plurality of predefined sensing applications; The method according to any one of items 1 to 8, comprising: [Item 10] 7. The method according to any one of items 4 to 6, wherein the steps of determining the sensing application and estimating the presence are jointly performed by the trained module. [Item 11] predicting future resource occupancy by the sensing application based on the results of the estimating step. The method according to any one of items 1 to 10, further comprising: [Item 12] Item 12. The method of item 11, wherein the step of performing wireless reception, transmission, or sensing includes scheduling the wireless reception, transmission, or sensing within resources predicted to be unoccupied by the sensing application. [Item 13] The step of performing wireless receiving, transmitting, or sensing further comprises: determining whether to transmit a sensing signal based on a result of the estimating step; performing or not performing the sensing depending on the determining step; Including, The method according to any one of items 1 to 12. [Item 14] 1. A method for training a module for identifying a sensing application, comprising: inputting into said module a representation of a received wireless signal and one or both of a desired indication of the presence or absence of a sensing signal in said representation and a desired indication of a sensing application that generated said sensing signal; modifying at least one parameter of said module in accordance with said input; A method comprising: [Item 15] A device for identifying a sensing application, comprising: A processing circuit, Obtaining a received wireless signal; estimating the presence of a sensing signal in the received signal, the sensing signal being a signal generated by a sensing application; and a processing circuit configured to execute a transceiver for performing wireless receiving, transmitting, or sensing based on the estimated presence of the sensing application; A device comprising: [Item 16] A device for training a module for identifying a sensing application, comprising: inputting into said module a representation of a received wireless signal and a desired indication of the presence or absence of a sensing signal in said representation and / or a desired indication of a sensing application which generated said sensing signal; modifying at least one parameter of said module according to said input; A device comprising: a processing circuit configured to execute the
Claims
1. A method executed by a device for identifying a sensing application executed by another device, the other device being other than the device for identifying the sensing application, acquiring a received wireless signal; estimating, by a trained module, the presence of a sensing signal in the received signal, the sensing signal being a signal generated by a sensing application; receiving, transmitting, or sensing wireless signals based on the estimated presence of the sensing application; A method comprising:
2. The method of claim 1 , wherein the sensing application is an application that generates the sensing signal at a preset period or as a continuous transmission signal.
3. The method of claim 1 or 2, wherein the sensing application is one of radar sensing, wireless sensing, wireless local area sensing, and channel state information sensing.
4. the estimating step further comprises determining the sensing application as one of a plurality of predefined sensing applications. The method according to any one of claims 1 to 3.
5. the sensing application is an application that generates the sensing signal at a preset period, determining the sensing application, estimating the predetermined period based on the presence of the sensing signal previously estimated in the received wireless signal; identifying the sensing application from among the plurality of predefined sensing applications using the estimated predetermined period; The method of claim 4 , comprising:
6. The determining of the sensing application is performed based on one or more of the characteristics, the characteristics being: Frame structure, Bandwidth, Sensing duration, Sensing start time, The end time of sensing, and waveform, and the one or more characteristics differ for at least two of the plurality of predefined sensing applications; The method according to claim 4 or 5.
7. The determining of the sensing application is performed based on one or more of the characteristics, the characteristics being: Bandwidth, Beam width, Transmit power, and Sensing rate, and the one or more characteristics differ for at least two of the plurality of predefined sensing applications; The method according to any one of claims 4 to 6.
8. determining the sensing application as one of the plurality of predefined sensing applications; is executed if the presence of the sensing signal in the received wireless signal is confirmed by the estimating step; is not performed if the presence of the sensing signal in the received wireless signal is not confirmed by the estimating step. The method according to any one of claims 4 to 6.
9. The step of obtaining the received wireless signal further comprises: Measuring one or more of a power density spectrum, a received signal strength, or a received signal strength indicator; determining, via a trained model, the presence or absence of a communication signal or a sensing signal in the received wireless signal; if a communication signal or a sensing signal is determined to be present in the received wireless signal, estimating whether the signal is a sensing signal; If the signal is a sensing signal, determining the sensing application as one of a plurality of predefined sensing applications; The method according to any one of claims 1 to 8, comprising:
10. The method according to any one of claims 4 to 6, wherein the steps of determining the sensing application and estimating the presence are jointly performed by the trained module.
11. predicting future resource occupancy by the sensing application based on the results of the estimating step. The method of any one of claims 1 to 10, further comprising:
12. 12. The method of claim 11, wherein the step of performing reception, transmission, or sensing of wireless signals comprises scheduling the reception, transmission, or sensing of wireless signals within resources predicted to be unoccupied by the sensing application.
13. The step of receiving, transmitting, or sensing a wireless signal further comprises: determining whether to transmit a sensing signal based on a result of the estimating step; performing or not performing the sensing depending on the determining step; Including, The method according to any one of claims 1 to 12.
14. A method for training a module for identifying sensing applications executed by other devices, the other devices being other than a device executing the module for identifying the sensing application, inputting into said module a representation of a received wireless signal and one or both of a desired indication of the presence or absence of a sensing signal in said representation and a desired indication of a sensing application that generated said sensing signal; modifying at least one parameter of said module in accordance with said input; A method comprising:
15. A device for identifying a sensing application executed by another device, the other device being other than the device for identifying the sensing application, A processing circuit, Obtaining a received wireless signal; estimating the presence of a sensing signal in the received signal, the sensing signal being a signal generated by a sensing application; and a processing circuit configured to execute a transceiver for receiving, transmitting, or sensing wireless signals based on the estimated presence of the sensing application; A device comprising:
16. A device for training a module for identifying a sensing application executed by another device, the other device being other than a device executing the module for identifying the sensing application, inputting into said module a representation of a received wireless signal and a desired indication of the presence or absence of a sensing signal in said representation and / or a desired indication of a sensing application that generated said sensing signal; modifying at least one parameter of said module in accordance with said input; A device comprising: a processing circuit configured to execute the
