Determine the position information about the drone

Abstract

A computer-implemented method for determining location information regarding the actual location of a drone in a communication network includes obtaining (302) a reported location of the drone at a first time point and obtaining (304) at the first time point a measurement of the radio conditions between the drone and a node in the telecommunication network. The method then includes predicting (306) the radio conditions at one or more locations related to the reported location of the drone and determining (308) location information regarding the actual location of the drone based on the measured radio conditions and the predicted radio conditions.

Description

Technical Field

[0001] The present disclosure relates to methods, nodes, and systems in a communication network. More particularly but not exclusively, the disclosure relates to determining location information regarding the actual location of an unmanned aerial vehicle (UAV). Background Art

[0002] Enhanced Long Term Evolution (LTE) support for UAVs is currently an area of research interest (see RAN#75, dated March 6 - 9, 2017, titled "Study on E-UTRA and E-UTRAN enhancements for Aerial Vehicles"). For example, whether an LTE network deployment can be used to serve UAVs, where base station antennas are targeted at terrestrial coverage to support Release 14 functionality.

[0003] A UAV (or an airborne user equipment UE) may experience radio propagation characteristics that can be different from those experienced by a UE on the ground. As long as the UAV is flying at a low altitude relative to the base station (BS) antenna height, its behavior will be like that of a conventional UE on the ground. However, once the UAV is flying above the BS antenna height, due to line-of-sight propagation conditions, the uplink (UL) signal from the UAV becomes visible to multiple cells. The UL signal from the UAV can increase interference in neighboring cells and the increased interference can have a negative impact on UEs on the ground, such as smartphones, IoT devices, etc. Similarly, the line-of-sight conditions to multiple cells can result in higher downlink (DL) interference to the UAV.

[0004] In addition, due to the downward tilt of the BS antenna, on the ground or below the BS antenna height, the UAV may be served by the main lobe of the BS antenna. However, when the UAV is flying above the line of sight, it is more likely to be served by the side lobe or the back lobe of the BS antenna, which have a reduced antenna gain compared to the antenna gain of the main lobe.

[0005] NR Beamforming: Multiple - antenna technology can improve signal quality. By spreading the total transmission power over multiple antennas, array gain for improving signal quality can be achieved. The transmitted signals from each antenna are formed in the following way: the received signals from each antenna are coherently added up for the user, which is called beamforming. Precoding describes how to form each antenna in the antenna array to form a "beam". The use of beamforming is a cornerstone in NR technology, and beamforming can be enabled in both the horizontal domain and the vertical domain using new advanced antenna systems. A UE or a drone can evaluate the beam quality in NR from a serving cell or an adjacent cell, for example, by measuring the synchronization signal block (SSB) or by measuring the channel state information reference signal (CSI - RS) resources.

[0006] RSRP Report : Reference Signal Received Power (RSRP) is a UE measurement. Assuming that the UE in the network sends an RSRP measurement report, it is an L3 measurement that contains the RSRP values of up to eight adjacent cells and the serving cell on the primary carrier in the LTE context. The RSRP value can be reported in the NR context by UE measurement of the SSB or CSI - RS.

[0007] UAV Trajectory Report : Drone trajectory reporting was introduced in Release 15 [36.331], and the drone trajectory reporting has the following format:

[0008] 。

[0009] Capable drones with available future location information can report their flight paths during connection setup. The report contains a sequence of location information elements with corresponding timestamps. Summary of the Invention

[0010] Drones registered in (or included in) a communication network can deliberately report false locations to the network. This can be for various reasons, including, for example: disrupting the terrestrial communication network by flying at a specific location to cause high interference in the uplink; flying in a "no - fly" zone such as an airport (a drone might do this, for example, to disrupt the communication network in such a zone or to capture sensitive video); flying at an altitude below or above the regulatory limit (e.g., a drone might want to travel at a different altitude for better received signal quality); flying at a speed higher than the maximum allowed speed limit; or simply to be able to fly over an illegal area to reach its destination faster (e.g., taking a shortcut).

[0011] In addition to deliberately reporting incorrect positions to the network, drones may also inadvertently (e.g., unknowingly) deviate from their reported position routes or report inaccurate positions due to, for example, inaccurate Global Navigation Satellite System (GNSS) position data. This may occur, for example, due to interfering transmitters or canyon effects associated with high-rise buildings. This is illustrated in Figure 1 wherein drone 104 reports its position to one or more nodes 102 as flying along the dashed flight path 106 at times t_1 to t_N, while actually flying at a higher altitude along flight path 108.

[0012] The aim of the embodiments herein is to be able to detect when a drone is reporting an inaccurate position and / or to determine the correct position of the drone.

[0013] A further aim of the embodiments herein is to provide improved connectivity and positioning services to drones in an LTE deployment, for example using existing Release 15 signaling.

[0014] Accordingly, in a first aspect herein, there is provided a computer-implemented method for determining position information regarding the actual position of a drone in a communication network. The method includes obtaining a reported position of the drone at a first time point and obtaining a measurement of the radio conditions between the drone and a node in the telecommunication network at the first time point. The method then includes predicting the radio conditions at one or more positions related to the reported position of the drone and determining position information regarding the actual position of the drone based on the measured radio conditions and the predicted radio conditions.

[0015] With this method, the predicted radio conditions near the reported drone position can be compared with the actual radio conditions measured between the drone and a node in the telecommunication network in order to determine information regarding the actual position of the drone. In some embodiments, the measured radio conditions at the position of the drone can be compared with the conditions that we would expect (e.g., predict) if the drone were actually at the position it has reported. If the predicted radio conditions match the measured conditions, then it is likely that the drone is at the position it has reported. If the measured conditions do not match the expected / predicted conditions, then this can provide an indication that the drone is not actually at the position it has reported. The actual position information can include, for example, the actual position of the drone, the actual flight path of the drone, and / or an indication of whether the drone has deviated from its reported flight path.

[0016] According to a second aspect, there is a node in a communication network for determining position information regarding the actual position of a drone, wherein the node includes a memory and a processor, the memory includes instruction data representing a set of instructions, the processor is configured to communicate with the memory and execute the set of instructions, wherein the set of instructions, when executed by the processor, causes the processor to: obtain a reported position of the drone at a first time point and obtain a measurement of the radio conditions between the drone and a node in the telecommunication network at the first time point. Further cause the node to predict the radio conditions at one or more positions related to the reported position of the drone, and determine the position information regarding the actual position of the drone based on the measured radio conditions and the predicted radio conditions.

[0017] According to a third aspect, there is a computer program product including a computer-readable medium having computer-readable code included therein, configuring the computer-readable code such that when executed by a suitable computer or processor, it causes the computer or processor to execute the method of the first aspect. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] To better understand and more clearly illustrate how the embodiments herein can be implemented, reference will now be made, by way of example only, to the accompanying drawings, in which:

[0019] Figure 1 illustrates an example of a drone deviating from a reported flight path;

[0020] Figure 2 illustrates an example node according to some embodiments herein;

[0021] Figure 3 illustrates an example method according to some embodiments herein;

[0022] Figure 4 illustrates an example method according to some embodiments herein;

[0023] Figure 5 illustrates an example mapping of predicted radio conditions according to some embodiments herein;

[0024] Figure 6 illustrates a further example mapping of predicted radio conditions illustrating a no-fly zone according to some embodiments herein; and

[0025] Figure 7 illustrates how the predicted mapping of radio conditions can be used to determine the actual flight path according to some embodiments herein. DETAILED DESCRIPTION

[0026] Figure 2Describes node 200 in a communication network according to some embodiments herein. Node 200 may be configured (e.g., adapted or programmed) to perform any of the embodiments of method 300 described below.

[0027] Generally, a communication network (or telecommunications network) may include any one or any combination of the following: wired links (e.g., ASDL) or wireless links such as New Radio (NR), Global System for Mobile Communications (GSM), Wideband Code Division Multiple Access (WCDMA), Long Term Evolution (LTE), WiFi, or Bluetooth wireless technologies. Those skilled in the art will appreciate that these are merely examples and the communication network may include other types of links.

[0028] Generally, node 200 may include any component or network function (e.g., any hardware or software module) suitable for performing the functions described herein in a communication network or may be included in any component or network function (e.g., any hardware or software module) suitable for performing the functions described herein in a communication network. For example, a node may include a device capable of, configured to, arranged to, and / or operable to communicate directly or indirectly with an unmanned aerial vehicle (also referred to as an aircraft or an aerial user equipment) and / or with other network nodes or devices in a communication network to enable and / or provide wireless or wired access to the unmanned aerial vehicle and / or to perform other functions (e.g., management) in the communication network. Examples of nodes include, but are not limited to, access points (APs) (e.g., radio access points), base stations (BSs) (e.g., radio base stations, Node B, evolved Node B (eNB), and NR Node B (gNB)). Further examples of nodes include, but are not limited to, core network functions such as, for example, core network functions in a fifth generation core network (5GC).

[0029] Node 200 may be configured or operable to perform the methods and functions described herein, such as method 300 described below. Node 200 may include a processor (e.g., processing circuitry or logic) 202. It will be appreciated that node 200 may include one or more virtual machines running different software and / or processes. Node 200 may thus include one or more servers, switches, and / or storage devices and / or may include a cloud computing infrastructure or an infrastructure configured to execute software and / or processes in a distributed manner.

[0030] The processor 202 can control the operation of the node 200 in the manner described herein. The processor 202 can include one or more processors, processing units, multi-core processors, or modules configured or programmed to control the node 200 in the manner described herein. In a particular implementation, the processor 202 can include multiple software and / or hardware modules, each configured to perform or for performing respective or multiple steps of the functionality of the node 200 as described herein.

[0031] The node 200 can include a memory 204. In some embodiments, the memory 204 of the node 200 can be configured to store program code or instructions that can be executed by the processor 202 of the node 200 to perform the functionality described herein. Alternatively or additionally, the memory 204 of the node 200 can be configured to store any requests, resources, information, data, signals, or the like described herein. The processor 202 of the node 200 can be configured to control the memory 204 of the node 200 to store any requests, resources, information, data, signals, or the like described herein.

[0032] It will be appreciated that in addition to Figure 2 those indicated in Figure 2 or alternatives to those indicated in

[0033] the node 200 can also include other components. For example, in some embodiments, the node 200 can include a communication interface. The communication interface can be used for communicating with other nodes in a communication network (such as other physical nodes or virtual nodes). For example, the communication interface can be configured to transmit requests, resources, information, data, signals, or the like to other nodes or network functions and / or receive requests, resources, information, data, signals, or the like from other nodes or network functions. The processor 202 of the node 200 can be configured to control such a communication interface to transmit requests, resources, information, data, signals, or the like to other nodes or network functions and / or receive requests, resources, information, data, signals, or the like from other nodes or network functions.

[0034] Figure 3A computer-implemented method 300 for determining location information regarding an actual location of a drone in a communication network according to some embodiments herein is described. Method 300 may be performed by a node 200 as described above. For example, the processor 202 may be caused or further caused to perform any of the steps or operations described in connection with any of the embodiments of method 300 as described below. In other embodiments, the method may be performed centrally, for example, in a distributed or cloud-based manner.

[0035] Briefly, in a first step 302, method 300 includes obtaining a reported location of the drone at a first time point. In a second step 304, the method includes obtaining a measurement of a radio condition between the drone and a node in the telecommunication network at the first time point. In a third step 306, the method includes predicting a radio condition at one or more locations associated with the reported location of the drone, and in a fourth step 308, the method includes determining location information regarding the actual location of the drone based on the measured radio condition and the predicted radio condition.

[0036] More specifically, the drone may include any wireless device capable of flying and further capable of being connected to a communication network. Examples of drones include, but are not limited to, aircraft, airborne user equipment (UE), or aviation-based devices (aviation cameras, sensors, or other devices). Those skilled in the art will be familiar with such drones and other objects that desire to have accurate location information.

[0037] In step 302, the method includes obtaining a reported location of the drone at a first time point. The reported location may be obtained directly or indirectly from the drone. For example, the drone may report its location at the first time point (e.g., a time instance or a time interval). The drone may use, for example, a drone trajectory report as described in the background section of this document to report its location. In some embodiments, the drone may report its location in response to a request from a network node (such as network node 200) for flight path information.

[0038] The obtained reported location may include a timestamped location or a series of timestamped reported locations (e.g., a sequence of timestamped reported locations). The reported location may be included in flight path information from the drone, and the flight path may include locations with associated timestamps (1...N).

[0039] As described above, the reported position may or may not accurately reflect the actual position of the drone. For example, the drone may be configured to deliberately report an inaccurate position. This may be to enable the drone to remain connected to the network while flying in, for example, a no-fly zone or a prohibited area. The drone may also inadvertently report an inaccurate position, for example, due to inaccurate GNSS position data. This may occur, for example, due to a jamming transmitter or a canyon effect associated with high-rise buildings. The nodes and methods herein can be used to determine whether the drone is reporting inaccurate position data or has deviated from its reported flight path. Some embodiments herein use standard reported channel measurements to achieve this (e.g., without incurring additional signaling overhead).

[0040] In step 304, a measurement of the radio conditions between the drone and a node in the telecommunications network is obtained at a first time point. In other words, the radio conditions are measured or estimated between the drone and the node at (near) the same time instance at which the drone reported its position (e.g., at the same time as the timestamp of the reported position obtained in step 302).

[0041] Those skilled in the art will realize that it is not necessary to perform the measurement of the radio conditions exactly at the first time point (e.g., at exactly the same time point as the reported position of the drone). For example, the radio conditions can be obtained at a time point approximately equal to the time point at which the drone reported its position; for example, the radio conditions can be measured at a time point adjacent to (e.g., slightly earlier or slightly later than) the first time point, or within a time window that overlaps, is adjacent to, or is simultaneous with the first time point.

[0042] In some embodiments, the radio conditions at the first time point can be obtained by interpolation or extrapolation of measurements of the radio conditions at a time different from the first time point (e.g., at a time on either side of the first time point). Generally, those skilled in the art will realize that the closer the measurement of the radio conditions is to the first time point, the more accurate the position information regarding the actual position of the drone can be.

[0043] In some embodiments, the measurement of the radio conditions can include radio measurements related to a reference signal transmitted by the drone. For example, the radio conditions can be obtained from the RSRP report as described above. As such, the method can be implemented without additional signaling overhead to obtain the measurement of the radio conditions. The radio conditions can thus include a reference signal received power (RSRP) measurement, e.g., an L3 measurement of the RSRP value.

[0044] A node (e.g., a base station) can generally use reference signals to obtain measurements of the radio conditions. For example, a drone performs measurements on the beams transmitted by the node, e.g., to evaluate the quality of the beams. Generally, the reference signals transmitted by the node to the drone can include at least one of the following: Channel State Information - Reference Signal (CSI-RS), SSB, Primary Synchronization Signal (PSS), Secondary Synchronization Signal (SSS), and Cell-Specific Reference Signal (CRS). More particularly, the drone can evaluate the beam quality by measuring the SSB (e.g., corresponding to the Synchronization Signal / Physical Broadcast Channel (PBCH) block) in a 5G (e.g., NR) network or by measuring the CSI-RS resources in a 5G (e.g., NR) network or a 4G (e.g., LTE) network. In the embodiments herein, the measurement of the radio conditions can include signal quality feedback on the above reference signals, such as RSRP, SINR, RSRQ, or SINR. The measurement of the radio conditions can also include the cell ID of the cells within range (e.g., an indication of which cells / nodes are within the range of the drone). The measurement of the radio conditions can also include timing advance or beamforming information such as a precoder index. The measurement of the radio conditions can also include radio signal quality measurements on the uplink signals (e.g., Sounding Reference Signal (SRS)) from the drone.

[0045] In some embodiments, the measurement of the radio conditions can include whether a UE can detect the node (e.g., having a signal from the node or being able to communicate with the node). For example, in step 304, the potential cells (cell IDs) that the drone can detect at a certain location can be obtained.

[0046] Turning to block 306, the method then includes predicting the radio conditions at one or more locations related to the reported location of the drone. Generally, the predicted radio conditions can be of the same type as the measurements of the radio conditions described above (or be transformed into the same type as the measurements of the radio conditions described above).

[0047] In some embodiments, in step 306, predicting the radio conditions at one or more locations related to the reported location of the drone can include using a channel model and deployment information (e.g., the known locations of the node(s) in the network) to predict the radio conditions. By using the channel model for the drone, the radio environment of the drone can be estimated. As an example, a channel model such as Free Space Path Loss (FSPL) can be used to predict the radio conditions. According to FPSL:

[0048]

[0049] where λis the signal wavelength, and d is the distance between the transmitter and the reported location of the drone. Using FSPL together with the antenna and noise power at the node and the drone, the RSRP of the drone at the reported location for each time instance can be predicted, or the potential cells (cell IDs) that the drone can detect at a certain location can be predicted.

[0050] In other embodiments, a model can be determined (or created) to predict radio conditions at different locations. For example, the network can build a radio signal quality prediction model of the environment that can map the drone location to radio measurements (such as the RSRP of one or more nodes). The prediction model can be built based on legitimate drone measurements (such as legally obtained drone measurements), and then a mapping from a set of drone locations to radio measurements is created. For example, in some embodiments, method 300 can further include obtaining ground truth location measurements and corresponding ground truth measurements of the radio conditions at the location. Such measurements can be obtained from drones (such as airborne UEs) that report trusted location information. Generally, measurements can be obtained from any terrestrial UE type that reports trusted location information such that it can be used as ground truth location information. For example, the measurements can be verifiable. Dedicated drones can be used to make such measurements or such measurements can be used to obtain the required ground truth data (such as survey drones), and such measurements can be aggregated from drone data available from drones at the site, a combination of both, or any other available data including radio conditions measured at different drone locations.

[0051] In some embodiments, a model trained using a machine learning process can be used to predict radio conditions. For example, the step of predicting radio conditions at one or more locations related to the reported location of the drone can include using a model trained using a machine learning process to predict radio conditions at one or more locations. As such, the ground truth data described above can be used as training data to train a machine learning model, for example, in the format (location, radio condition measurement).

[0052] Those skilled in the art will be familiar with a variety of machine learning models that can be trained to predict radio conditions based on location information. For example, a classification model can be trained in a supervised manner on the training data as described above. Examples of models that can be used include, but are not limited to, neural networks, decision trees (such as the random forest algorithm), logistic regression, and linear regression.

[0053] The model used to predict radio conditions can, for example, include a recurrent neural network that exhibits time-dynamic behavior and can thus process input sequences (such as in a path). The random forest algorithm can also be used.

[0054] Those skilled in the art will be familiar with neural networks, but in brief, a neural network is a type of supervised machine learning model that can be trained to predict a desired output for given input data. A neural network is trained by providing training data that includes example input data and the corresponding "correct" or desired ground truth results. A neural network includes multiple layers of neurons, and each neuron represents a mathematical operation applied to the input data. The output of each layer in the neural network is fed into the next layer to produce an output. For each piece of training data, the weights and biases associated with the neurons are adjusted until the best weighting is found that produces a prediction for the training example that reflects (e.g., best predicts) the corresponding ground truth.

[0055] Those skilled in the art will be familiar with methods for training neural networks using training data (such as gradient descent, etc.) and will appreciate that the training data can include hundreds or thousands of lines of training data obtained under different ranges of network conditions.

[0056] Typically, a model may have been trained using training data. Each piece of training data includes: i) an example drone position; and ii) a ground truth measurement of the radio conditions at the example drone position. The training data can be obtained as described above.

[0057] In some embodiments, the model can be trained to take a position as input and output a prediction of the radio conditions at that position. The position can be timestamped such that the model further takes time as input and outputs the predicted radio conditions at the specified position at the specified time. The model can be trained to take multiple input positions (e.g., along a flight path) or a sequence of input positions and output predictions of the radio conditions at each point along the flight path. In other embodiments, as described below, the model can be trained to take a region as input and output a map of the radio conditions in an area covering the input region. In another example, a neural network can take a position as input and output a map centered on the input coordinates (e.g., of a node in position [0,0]). As described below, an example map 500 of the predicted radio conditions around node 502 is illustrated in Figure 5 An example map 500 of the predicted radio conditions around node 502 is illustrated in

[0058] In some embodiments, the model can be trained to take as input one or more positions and output measurements of a plurality of different radio conditions between the drone and each of a plurality of different nodes in the telecommunications network. As such, the model can predict a "fingerprint" of the radio conditions that the drone can be expected to experience at the reported position, on the links between different nodes, and / or on different channels. In additional examples, the model can be trained to output a map that includes a plurality of such fingerprints at different positions (e.g., each pixel or position point on the map can be associated with a vector that includes predictions of the radio conditions at that point between the drone and the plurality of nodes and / or channels).

[0059] In some embodiments, the model can be trained to take as input one or more positions obtained as in step 302 and one or more corresponding measurements of radio conditions obtained as in step 304 and output, based on the measurements, the probability that the drone is actually at the specified input position. In other words, the model can further output the probability that the drone is out of path.

[0060] Those skilled in the art will realize that these are only examples and that other forms of input and output parameters are also possible. It will be realized that references to positions can generally involve three-dimensional coordinates including, for example, altitude. Examples of other inputs include, but are not limited to, the time of day and / or the serving cell ID.

[0061] Over time, due to external changes in the environment, the performance of the trained model may degrade. Therefore, an update / retraining of the model can be performed. In some embodiments, the model can be updated periodically. Based on the deployment location, traffic, and / or other external changes, the operator can decide on the period after which the model can be updated / retrained, regardless of the model performance. Training data for such an update can be collected, for example, by flying drones owned by the operator in a given coverage area.

[0062] In another embodiment, training updates can be performed on an ad hoc / occasional basis. If the number of off-path drones estimated by a given model during a given time period is above a threshold, this may indicate that the model's accuracy should be verified and the model can be verified / updated. The threshold may be set by the network operator and may include a trade-off between network performance (e.g., detection of off-path drones) and complexity (e.g., model updates). In such a scenario, model performance can be monitored based on a quality metric, such as the number of off-path drones detected during a given time period, as described above, and can be updated accordingly. Another metric may be the impact that off-path drones are having on the performance of the terrestrial network. For example, if off-path drones are intended to disrupt the terrestrial communication network by generating high interference levels, the network may set an interference threshold as the metric used for model updates.

[0063] In another embodiment, the model can be continuously updated by using ground truth data collected from trusted drones that commonly fly over the area, such as delivery drones that follow the same route every day.

[0064] In embodiments where the prediction in step 306 is performed by a trained machine learning model (e.g., an AI model), the model may be located in different nodes of the network. For example, the model may be located in a core node, such as a Mobility Management Entity (MME). In embodiments where method 300 is performed in a node serving a drone, the serving node may need to signal off-path detections to neighboring nodes during handovers. This may be necessary because off-path detections may be relevant over a longer period of time (e.g., if the drone does not immediately correct its trajectory) and multiple nodes may be serving the drone during its flight. Such off-path signaling between nodes during handovers may include, for example:

[0065] - Drone flight reports

[0066] -Probability of the drone being outside the path

[0067] - Similarity between reported and estimated radio measurements at previous time steps

[0068] - Estimated position of the drone

[0069] - Estimated flight track of the drone.

[0070] In such an embodiment, the serving node and the neighboring nodes can check the similarity between their respective out-of-path detection results and compare them accordingly. In other words, a cross-check can be performed between different nodes. For example, method 300 can be performed on the serving node and independently on the neighboring nodes, and the outputs (such as location information regarding the actual location of the drone) can be compared. The serving node can, for example, calculate a similarity metric between its result and the reported result from the neighboring node. Such a similarity check scheme can be used to overcome false result reports from spurious / compromised nodes that may signal false out-of-path signals to neighboring nodes during handover.

[0071] In other embodiments where the prediction in step 306 is performed by a trained machine learning model (such as an AI model), the model can be located, for example, in a cloud or distributed computing arrangement outside the core network and control the drone management function. In other words, method 300 can be performed in the cloud or other distributed computer arrangements. If the reported location of the drone is sent to the cloud in step 302, then the network measurement data can also be sent there in step 304, and the model can be built and maintained directly in the cloud.

[0072] Now turning to Figure 3 step 308 in, generally, the prediction can be made to determine whether the conditions that we would expect (e.g., predict) if the drone were actually at the location it has reported match the measured conditions at the actual location of the drone. In this way, it can be determined whether the drone is at the location it has reported.

[0073] Accordingly, step 308 can include comparing the measured radio conditions and the predicted radio conditions. The location information can then include a determination based on the comparison of whether the drone has deviated from the reported location. In some examples, the step of predicting 306 the radio conditions can include predicting the radio conditions at the reported location of the drone. For example, if the measured radio conditions deviate from the predicted radio conditions (e.g., by more than a threshold amount or by more than a statistically significant level), then it can be determined in step 308 that the drone has deviated from the reported location. As such, location information regarding the actual location of the drone can be provided that indicates whether the drone is out of path.

[0074] This is illustrated in Figure 4 where, Figure 4Node 402 is shown connected to drone 404. Drone 404 reports position data along flight path 406 at a first time point t1, a second time point t2, and subsequent time points tT as illustrated by the circles along the dashed flight path 406. Out-of-path determination can be performed for drone 404 using method 300 above. Reported positions are obtained 302, 304 via drone report 408, along with corresponding measurement conditions between drone 404 and node 402. In step 306, a model 410 trained using any of the processes described above is used to predict the radio conditions at positions (t1, t2,.., tT). In step 308, the predicted conditions are compared to the radio conditions measured by the drone (e.g., radio measurements 408 reported by the drone).

[0075] Thus, in step 308, if the measured radio conditions are similar to the predicted conditions, e.g., within a threshold amount, then it can be determined that drone 404 has accurately reported its position. If the measured radio conditions deviate from the predicted conditions, e.g., by more than a threshold amount, then it can be determined that drone 404 has inaccurately reported its position. For example, out-of-path determination can be performed for the drone.

[0076] As described above, in some embodiments, multiple radio conditions are obtained 302 (e.g., from different nodes / channels) and multiple radio conditions are predicted 306 for the reported positions of the drone, e.g., obtaining measured and predicted "fingerprints" as described above. For example, the drone has reported radio measurements [r1, r2,…rN] at time t1 (where, according to Figure 4 , r1 may correspond to the SINR of one node), and the node can determine the similarity s_1 between the reported measurement and the predicted by, for example, determining the Euclidian distance and using a threshold Euclidian distance to determine that the drone is out-of-path.

[0077] Alternatively, for example, out-of-path determination can be performed based on a statistical measure such as whether the mean / maximum / minimum of the differences between the predicted radio conditions and the measured radio conditions is above a certain threshold.

[0078] More generally, in some embodiments, in step 308 the method can include calculating a similarity measure between the measured and predicted radio conditions for one or more pairs of radio conditions. The measured and predicted radio conditions of the radio conditions can represent different time points and / or can be between different nodes in the drone and the network.

[0079] In some embodiments, the similarity measure can be compared to a threshold to determine whether the drone is out-of-path.

[0080] In other embodiments, the method may include providing one or more similarity measures to a second model trained using a machine learning process. Such a model may include any type of model of the types described above with respect to the model used in step 306 (e.g., neural network, random forest, etc.). In some embodiments, classification is performed. The output may include a binary indication of whether the drone is on / off the path or the probability that the drone is outside the path. The input may thus include a set of similarity measures for every T positions, and the output may include the probability of out-of-path detection.

[0081] As an example, in an embodiment where a series of T positions are reported at a series of T time points in step 302 and a corresponding series of radio condition measurements are obtained in step 304, the T samples may then be forwarded to a second model that takes as input a set of T samples and outputs the probability that the drone is outside the path.

[0082] Training data including an example set of similarity measures and a ground truth indication of whether the drone is outside the path may be used to train the second model. As such, using the second model may enable scenario-specific combinations of similarities to detect whether the drone is following its reported path. Since one combination of similarities may not be applicable to all scenarios. Such a model may be able to apply different criteria for different positions. For example, instead of applying a single measure of difference (e.g., a single Euclidean distance threshold), the model may be able to effectively learn different thresholds and / or different combinations of similarity measures for different positions. Thus, effectively, it can facilitate a more granular determination of whether the drone is outside the path.

[0083] In such an embodiment, the second model may be located in the cloud, while the model may be distributed across different regions of the network, e.g., between a network node (e.g., MME) and the location of the second model. For example, a machine learning model may be used to predict a sequence of radio conditions in a network node (e.g., the node serving the drone) in step 306. The serving node may then compare the measured radio conditions and the predicted radio conditions, e.g., to determine T measures of the similarity between the measured and predicted radio conditions. The results may then be sent to the second model in the cloud. The aforementioned aggregation of the T similarities may be performed in the second model to make a final decision. During handover between nodes of the drone, the nodes participating in the handover may communicate among themselves to continue sending data to the second model. On the other hand, the initiating (e.g., serving) node may send information about the handover and information about the target node to the second model, and the second model may decide to inform the target node to continue sending data.

[0084] Turning now to other embodiments, determining location information regarding the actual location of the drone can include determining whether the drone is flying in a no-fly zone (e.g., flying in restricted airspace or other space where the drone should not be flying, such as flying above an airport). For such areas, if the drone is not allowed to fly into the no-fly zone to collect such training data, the training data may not be available. As such, in embodiments where a machine learning model is used to perform the prediction in step 306, if no training data is available for the model, the model should not be able to predict the radio conditions in the no-fly zone. This is illustrated in Figure 6 which is shown in Figure 6 a map showing a prediction of the radio conditions near node 602 including a restricted area 604 where the prediction cannot be made. In such a scenario, anomaly detection can be used to estimate the similarity between the received samples and the samples collected from training data of legitimate drones.

[0085] For example, in some embodiments, in step 308, the method can include: determining that the actual location of the drone is in a no-fly zone if the measured radio conditions do not match the predicted radio conditions (e.g., statistically do not match). In particular, if the measured radio conditions do not match all of the predicted conditions (e.g., differ by more than a predetermined threshold amount).

[0086] In embodiments where multiple radio conditions are obtained in step 302 (e.g., from different nodes / channels) and corresponding multiple radio conditions (e.g., measured "fingerprint" and predicted "fingerprint") are predicted for the reported location of the drone in step 306, if the measured fingerprint has a different pattern of radio conditions compared to the predicted fingerprint, the measured fingerprint may not match the predicted fingerprint (or the predicted fingerprint for a different location). If the measured fingerprint has a different pattern of radio conditions compared to the predicted fingerprint, the measured fingerprint may not match the predicted fingerprint (or the predicted fingerprint for a different location), such that it is statistically impossible to find the measured fingerprint in the predicted fingerprint (e.g., considering errors in measurement and prediction).

[0087] In Figure 6 the scenario shown, an anomaly can be detected if, for example, the SINR is outside the range of other locations. If the SINR > 25 dB or the SINR < -5 dB, that is an example according to Figure 6 Note that radio measurements from more than one node result in more accurate detection.

[0088] Turning now to other examples, in some embodiments, the step of obtaining measurements of the 304 radio conditions includes obtaining a series of radio condition measurements between a node in a telecommunications network and a drone, the series of measurements being made at a series of locations along a flight path as reported by the drone (and thus at a series of time points). The step of determining 308 the position information can then include determining the actual flight path of the drone by pattern matching the obtained series of radio condition measurements to a pattern in the predicted radio conditions at one or more locations related to the reported positions of the drone.

[0089] For example, as described above, in some embodiments, the predicted radio conditions can include a mapping of the radio conditions covering an area including the flight path as reported by the drone. This is illustrated in Figure 5 which Figure 5 shows a prediction of the radio conditions 500, e.g., the signal quality at different locations around a node 502 with 3 sectors at an altitude of 100 meters. Thus, the predicted radio conditions along the reported flight path of the drone can be matched to a pattern in such a mapping to determine the actual position / flight path of the drone.

[0090] This is illustrated in Figure 7 which Figure 7 shows the trajectory of the reported flight path 706 color-coded according to the measurements of the radio conditions obtained at each point. The spatial pattern and the pattern in the measured radio conditions can be compared to the predicted pattern in the mapping of the predicted radio conditions 700 to determine the actual flight path 704 of the drone through the area.

[0091] Machine learning algorithms (such as convolutional neural networks) can be used to perform pattern recognition (as Figure 7 shown) using a method similar to image recognition by relying on an offline radio signal mapping in the same area in which the drone is flying. This, in turn, can enable the network to classify whether the reported future position of the drone at a given timestamp is expected or anomalous.

[0092] Those skilled in the art will appreciate that other pattern matching techniques can also be used, such as, for example, fuzzy matching, deep learning, and / or genetic algorithms.

[0093] Using pattern recognition techniques enables the network not only to detect drones outside the path but also to determine the most likely route being used by the drone. In this way, the position information can include the actual position or flight path of the drone (e.g., not only an estimate of whether the drone is outside the path).

[0094] Turning to another embodiment, method 300 may include obtaining ground truth measurements of radio conditions at different drone positions (e.g., obtained from a drone or a UE). The ground truth measurements can be used to train a model using a machine learning process to predict radio conditions at different drone positions. In inference, the method then includes obtaining 302 the reported position of the new drone at a first time point. The method may then include obtaining 304 a measurement of the radio condition between the new drone and a node in the telecommunications network at the first time point. In this embodiment, method 300 then includes providing the obtained reported position as an input to the model. The model then provides, as an output, a prediction 306 of the radio condition at the reported position of the new drone. In step 308, the measured radio condition is compared with the predicted radio condition to determine whether the new drone is off-path.

[0095] Once location information about the actual position of the drone has been obtained, for example, according to method 300 above, if it is determined that the drone has reported an inaccurate position (e.g., if an off-path detection is made), the drone may be warned, requested to return to the ground plane, or disconnected from the network. The warning may indicate the time when the connection will be terminated, so that the drone can first adjust its path to the reported path. Thus, generally, if the location information indicates that the drone has deviated from the reported position of the drone, method 300 may further include: sending a message to the drone, the message including one of the following: i) a warning to the drone that it has deviated from its reported position; ii) a request for the drone to return to the ground plane; and iii) an indication that the drone will be disconnected from the communication network if it fails to change its flight trajectory. The method may further (or alternatively) include disconnecting the drone from the communication network.

[0096] Thus, in this way, off-path UAV detection can be performed. False path reports (whether intentional or otherwise) can result in: disrupting the terrestrial communication network by causing high interference in the uplink if the UAV is flying at a specific location. UAVs flying in "no-fly" zones such as airports can disrupt the communication network in such neighborhoods or result in the capture (by the UAV) of sensitive video or other information. As described above, UAVs can also report false location information so that they can fly at altitudes above or below the permitted limits, or faster than actually permitted. Thus, the methods and nodes herein can be used to detect and stop such activities. If a UAV inadvertently deviates from its reported position route, for example due to inaccurate GNSS location, such as due to interfering transmitters or canyon effects associated with high-rise buildings, the methods herein can further be used to assist the UAV in flying back to its correct course. Additionally, the methods herein enable off-path UAV detection to be performed using standard signaling without, for example, any incurred additional signaling burden on the network.

[0097] In another embodiment, a computer program product is provided that includes a computer-readable medium having computer-readable code included therein that is configured such that when executed by a suitable computer or processor, causes the computer or processor to perform any of the embodiments of method 300 described herein.

[0098] Thus, it will be appreciated that the disclosure also applies to computer programs suitable for implementing the embodiments, in particular computer programs on or in a carrier. The program can be in the form of: source code, object code, intermediate source code, and object code in any other form suitable for use in the implementation of the method according to the embodiments described herein, such as in a partially compiled form.

[0099] It will also be appreciated that such a program can have many different architectural designs. For example, the program code implementing the functionality of the method or node can be broken down into one or more subroutines. For a person skilled in the art, many different ways of distributing the functionality among these subroutines will be obvious. The subroutines can be stored together in an executable file to form a self-contained program. Such an executable file can include computer-executable instructions, such as processor instructions and / or interpreter instructions (such as Java interpreter instructions). Alternatively, one or more or all of the subroutines can be stored in at least one external library file and, for example, linked to the main program statically or dynamically at runtime with one or more or all of the subroutines. The main program contains at least one call to at least one of the subroutines. The subroutines can also include calls to each other's functionality.

[0100] The carrier of a computer program can be any entity or device capable of carrying the program. For example, the carrier can include a data storage device, such as a ROM (e.g., a CD ROM or a semiconductor ROM) or a magnetic recording medium (e.g., a hard disk). Additionally, the carrier can be a transmissible carrier such as an electrical or optical signal that can be transmitted via a cable or an optical fiber cable or by radio or other components. When the program is included in such a signal, the carrier can be constituted by such a cable or other devices or components. Alternatively, the carrier can be an integrated circuit in which the program is embedded, and the integrated circuit is adapted to execute the relevant method or is used in the execution of the relevant method.

[0101] From the study of the drawings, the disclosure, and the appended claims, when implementing the claimed invention, changes to the disclosed embodiments can be understood and implemented by those skilled in the art. In the claims, the word "comprising" does not exclude other elements or steps, and the indefinite article "a" or "an" does not exclude a plurality. A single processor or other unit can implement the functions of several items recited in the claims. The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage. A computer program can be stored / distributed on a suitable medium such as a solid-state medium or an optical storage medium provided together with or as part of other hardware, but the computer program can also be distributed in other forms (such as via the Internet or other wired or wireless telecommunication systems). Any reference signs in the claims should not be construed as limiting the scope.

Claims

1. A computer-implemented method for determining location information regarding the actual location of a drone in a communication network, the method comprising: Obtaining (302) a reported location of the drone at a first point in time; Obtaining (304) at the first point in time a measurement of the radio conditions between the drone and a node in the communication network; Predicting (306) the radio conditions at one or more locations associated with the reported location of the drone; And Determining (308) the location information regarding the actual location of the drone based on the measured radio conditions and the predicted radio conditions, Wherein, if the location information indicates that the drone has deviated from the reported location of the drone, the method further comprises: Sending a message to the drone, the message including a request for the drone to return to the ground plane, When the method is executed by a serving node of the drone, the method further comprises: signaling a path-out detection result from the serving node to an adjacent node during a handover for cross-checking between the serving node and the adjacent node.

2. The method according to claim 1, wherein Determining the location information regarding the actual location of the drone based on a comparison between the measured radio conditions and the predicted radio conditions.

3. The method according to claim 2, wherein The step of predicting (306) the radio conditions comprises: predicting the radio conditions at the reported location of the drone; and Wherein, the step of determining (308) the location information comprises: determining whether the drone has deviated from the reported location based on the comparison.

4. The method according to claim 3, comprising: If the measured radio conditions deviate from the predicted radio conditions by more than a threshold amount, it is determined that the drone has deviated from the reported location.

5. The method according to claim 1, wherein Obtaining (304) measurements of radio conditions includes: Obtaining a series of measurements of the radio conditions between a node in the communication network and the drone, the series of measurements being made at a series of locations along the flight path reported by the drone; And Wherein, the step of determining (308) the location information comprises: determining the actual flight path of the drone by pattern matching the obtained series of radio condition measurements with a pattern in the predicted radio conditions at the one or more locations associated with the reported location of the drone.

6. The method according to claim 5, wherein, The predicted radio conditions include a mapping of the radio conditions covering the area including the flight path reported by the drone.

7. The method according to any one of the preceding claims, wherein, The measured radio conditions include measurements of a plurality of radio conditions between the drone and each of a plurality of different nodes in the communication network.

8. The method according to any one of claims 1 to 6, further comprising: Determining that the actual location of the drone is in a no-fly zone if the measured radio conditions do not match the predicted radio conditions.

9. The method according to any one of claims 1 to 6, wherein The step of predicting (306) the radio conditions at one or more locations associated with the reported location of the drone comprises: Using a channel model and deployment information to predict the radio conditions.

10. The method according to any one of claims 1 to 6, wherein, The step of predicting (306) radio conditions at one or more positions related to the reported position of the drone includes: Using a model trained using a machine learning process to predict the radio conditions at the one or more positions.

11. The method according to claim 10, wherein, The model has been trained using training data, where each piece of training data includes: i) an example drone position; and ii) a ground truth measurement of the radio conditions at the example drone position.

12. The method according to claim 10, wherein, The model includes a neural network or a random forest model.

13. The method according to any one of claims 1 to 6, wherein, If the position information indicates that the drone has deviated from the reported position of the drone, the method further includes: Sending a message to the drone, the message including one of the following: i) a warning to the drone that it has deviated from its reported position; and ii) an indication that the drone will be disconnected from the communication network if it fails to change its flight trajectory; and / or Disconnecting the drone from the communication network.

14. The method according to any one of claims 1 to 6, wherein, The method is performed by a base station, a network node, or a network function node in the communication network.

15. The method according to any one of claims 1 to 6, wherein The method is performed in a distributed manner or in the cloud.

16. A node (200) in a communication network for determining position information regarding the actual position of a drone, wherein, The node includes a memory (204), the memory (204) including instruction data representing a set of instructions; And A processor (202) configured to communicate with the memory and execute the set of instructions, where the set of instructions, when executed by the processor, causes the processor to: Obtain the reported position of the drone at a first time point; Obtain a measurement of the radio conditions between the drone and a node in the communication network at the first time point; Predict the radio conditions at one or more positions related to the reported position of the drone; Determine the position information regarding the actual position of the drone based on the measured radio conditions and the predicted radio conditions; and If the position information indicates that the drone has deviated from the reported position of the drone, send a message to the drone, the message including a request for the drone to return to the ground plane, Wherein the set of instructions, when executed by the processor, further causes the processor to: signal a path-out detection result from the node to an adjacent node during a handover for performing a cross-check between the node and the adjacent node.

17. A computer program product, comprising a computer-readable medium having computer-readable code included therein, configuring the computer-readable code such that, when executed by a suitable computer or processor, it causes the computer or processor to perform the method according to any one of claims 1 to 15.

Images