CN116027800A - A switching method, device, network device and computer storage medium - Google Patents

Abstract

The invention discloses a switching method, which comprises the following steps: receiving current flight state information of the unmanned aerial vehicle, inputting the current flight state information into a trained deep neural network model to determine a parameter set of a switching event of the unmanned aerial vehicle, and transmitting the parameter set of the switching event of the unmanned aerial vehicle to a base station. The embodiment of the invention also discloses a switching device, network equipment and a computer storage medium, so that the communication quality of the unmanned aerial vehicle can be ensured when the unmanned aerial vehicle performs cell switching, and the communication quality of the unmanned aerial vehicle is improved when the unmanned aerial vehicle performs cell switching.

Description

A switching method, device, network device and computer storage medium

technical field

The present invention relates to the technical field of cell handover of unmanned aerial vehicles, in particular to a handover method, device, network equipment and computer storage medium.

Background technique

At present, cell handover refers to the process in which a terminal moves from one cell to another during network communication. In order to ensure that the user terminal does not drop calls and maintain better communication services during the handover process, an optimal handover algorithm is required. .

Existing cell handover algorithms include optimizing cell handover by reserving channel resources and marking priorities in the handover strategy, using a handover algorithm based on signal-to-noise ratio to formulate handover strategies, and communicating cell handover messages through communication terminals Issue and control.

However, the existing handover algorithm is based on the configuration of ground users, which is often inconsistent with the coverage of the air network. This will lead to frequent handover of network cells during low-altitude operations, which will affect the communication quality of air drones and other terminals. It can be seen from this that, The existing unmanned aerial vehicle has the technical problem of poor communication quality after cell switching.

Contents of the invention

In view of this, the present invention provides a handover method, device, network equipment and computer storage medium to solve the technical problem of poor communication quality after the UAV performs cell handover in the prior art.

Technical scheme of the present invention is realized like this:

In a first aspect, an embodiment of the present invention provides a handover method, including:

Receive the current flight state information of the UAV; wherein, the flight state information includes: flight direction, flight speed, signal quality parameters and distance from adjacent cells;

Inputting the current flight state information into the trained deep neural network model to determine the parameter set of the switching event of the unmanned aerial vehicle;

Sending the parameter set of the handover event of the UAV to the base station; wherein the parameter set of the handover event of the UAV is used by the base station to instruct the UAV to perform cell handover.

In the above method, the method also includes:

Acquiring a training data set from the collected sample data set; wherein, the sample data set is: a parameter set of a switching event corresponding to the flight state information and the flight state information;

Inputting the training data set into a preset deep neural network model for training to obtain a trained deep neural network model;

According to the trained deep neural network model, the trained deep neural network model is determined.

In the above method, the described according to the trained deep neural network model, determining the trained deep neural network model includes:

obtaining a test data set from said sample data set;

The flight status information in the test data set is input into the trained deep neural network model to obtain a parameter set of the switching event;

The error between the parameter set of the calculated switching event and the parameter set of the switching event corresponding to the flight state information in the test data set;

When the error satisfies the second preset condition, the trained deep neural network model is determined as the trained deep neural network model;

When the error does not satisfy the second preset condition, return to the execution of obtaining the training data set from the collected sample data set.

In the above method, also include:

When the error is less than or equal to a preset error threshold, determining that the error satisfies the second preset condition;

When the error is greater than the preset error threshold, it is determined that the error does not satisfy the second preset condition.

In the above method, also include:

Based on the preset flight direction, the data set of the flight direction is determined according to the step size of the preset flight direction;

Based on the preset flight position, determine the data set of the distance between the preset flight position and the adjacent cell;

Based on the preset flight speed, according to the preset flight speed step size, determine the data set of the flight speed;

According to the ratio of 1:1:1, the data set of the flight direction, the data set of the flight speed and the data set of the distance are respectively used to form a three-dimensional vector;

forming an array of correspondences between each vector in the three-dimensional vectors and each set of parameter sets in the preset switching event parameter set;

From the array, select the data set of the sample.

In the above method, selecting the data set of the sample from the array includes:

According to each array in the array, call the LOS path propagation model to obtain the signal quality parameter corresponding to each array;

From the signal quality parameters corresponding to the array and each array, select for each flight direction, when the handover failure rate in the signal quality parameters and the channel bit error rate in the signal quality parameters meet the first preset condition, Correspondence between the flight status information and the parameter set of the handover event;

The selected flight state information and the parameter set of the switching event corresponding to the selected flight state information are determined as the sample data set.

In the above method, also include:

When the handover failure rate in the signal quality parameter is less than or equal to a preset failure rate threshold, and/or the channel bit error rate in the signal quality parameter is less than or equal to a preset bit error rate threshold, determine the channel quality parameter The medium switching rate and the channel bit error rate meet the first preset condition;

When the handover failure rate in the signal quality parameter is greater than the preset failure rate threshold, and the channel bit error rate in the signal quality parameter is greater than the preset bit error rate threshold, determine the handover rate and the channel in the channel quality parameter The bit error rate does not satisfy the first preset condition.

In a second aspect, the present invention provides a switching device, comprising:

The receiving module is used to receive the current flight state information of the drone; wherein, the flight state information includes: flight direction, flight speed, signal quality parameters and distance from adjacent cells;

A determination module, configured to input the current flight state information into the trained deep neural network model, to determine the parameter set of the switching event of the drone;

The handover module is configured to send the parameter set of the handover event of the UAV to the base station; wherein, the parameter set of the handover event of the UAV is used by the base station to instruct the UAV to perform cell handover.

In the above device, the above device is also used for:

Acquiring a training data set from the collected sample data set; wherein, the sample data set is: a parameter set of a switching event corresponding to the flight state information and the flight state information;

Inputting the training data set into a preset deep neural network model for training to obtain a trained deep neural network model;

According to the trained deep neural network model, the trained deep neural network model is determined.

In the above device, the above device determines the trained deep neural network model according to the trained deep neural network model, including:

obtaining a test data set from said sample data set;

The flight status information in the test data set is input into the trained deep neural network model to obtain a parameter set of the switching event;

The error between the parameter set of the calculated switching event and the parameter set of the switching event corresponding to the flight state information in the test data set;

When the error satisfies the second preset condition, the trained deep neural network model is determined as the trained deep neural network model;

When the error does not satisfy the second preset condition, return to the execution of obtaining the training data set from the collected sample data set.

In the above device, the above device is also used for:

When the error is less than or equal to a preset error threshold, determining that the error satisfies the second preset condition;

When the error is greater than the preset error threshold, it is determined that the error does not satisfy the second preset condition.

In the above device, the above device is also used for:

Based on the preset flight direction, the data set of the flight direction is determined according to the step size of the preset flight direction;

Based on the preset flight position, determine the data set of the distance between the preset flight position and the adjacent cell;

Based on the preset flight speed, according to the preset flight speed step size, determine the data set of the flight speed;

According to the ratio of 1:1:1, the data set of the flight direction, the data set of the flight speed and the data set of the distance are respectively used to form a three-dimensional vector;

forming an array of correspondences between each vector in the three-dimensional vectors and each set of parameter sets in the preset switching event parameter set;

From the array, select the data set of the sample.

In the above device, the above device selects the data set of the sample from the array, including:

According to each array in the array, call the LOS path propagation model to obtain the signal quality parameter corresponding to each array;

From the signal quality parameters corresponding to the array and each array, select for each flight direction, when the handover failure rate in the signal quality parameters and the channel bit error rate in the signal quality parameters meet the first preset condition, Correspondence between the flight status information and the parameter set of the handover event;

The selected flight state information and the parameter set of the switching event corresponding to the selected flight state information are determined as the sample data set.

In the above device, the above device is also used for:

When the handover failure rate in the signal quality parameter is less than or equal to a preset failure rate threshold, and/or the channel bit error rate in the signal quality parameter is less than or equal to a preset bit error rate threshold, determine the channel quality parameter The medium switching rate and the channel bit error rate meet the first preset condition;

When the handover failure rate in the signal quality parameter is greater than the preset failure rate threshold, and the channel bit error rate in the signal quality parameter is greater than the preset bit error rate threshold, determine the handover rate and the channel in the channel quality parameter The bit error rate does not satisfy the first preset condition.

In the third aspect, the embodiment of the present invention also provides a network device, the network device includes: a processor and a storage medium storing instructions executable by the processor, and the storage medium depends on the processor through a communication bus Executing an operation, when the instruction is executed by the processor, execute the switching method described in one or more embodiments above.

In a fourth aspect, an embodiment of the present invention provides a computer storage medium, which stores executable instructions, and when the executable instructions are executed by one or more processors, the processors perform one or more of the above implementations. The switching method described in the example.

A switching method, device, network device and computer storage medium provided by the present invention, the method includes: receiving the current flight state information of the drone, wherein the flight state information includes: flight direction, flight speed, signal quality parameters and The distance from the adjacent cell, the current flight status information is input into the trained deep neural network model to determine the parameter set of the handover event of the UAV, and the parameter set of the handover event of the UAV is sent to the base station , where the parameter set of the handover event of the UAV is used by the base station to instruct the UAV to perform cell handover; that is, in the embodiment of the present invention, the trained deep neural network model can determine the current The parameter set of the switching event corresponding to the flight state. In this way, the parameter set of the switching event determined by using the trained deep neural network model combined with the current flight state information of the UAV is more suitable for the flight environment of the UAV. Then, The cell is switched based on the determined parameter set of the handover event, so that the UAV can ensure the communication quality of the UAV when it performs cell handover, thereby improving the communication quality when the UAV performs cell handover.

Description of drawings

FIG. 1 is a flow chart of 5G handover signaling in the related art;

FIG. 2 is a schematic flowchart of an optional handover method in an embodiment of the present invention;

FIG. 3 is a schematic diagram of an optional distance from adjacent cells provided in an embodiment of the present invention;

FIG. 4 is a schematic diagram of an optional unmanned aerial vehicle flight provided by an embodiment of the present invention;

5 is a schematic diagram of an optional DNN model establishment and correction process provided by an embodiment of the present invention;

FIG. 6 is a schematic flowchart of construction and prediction of an optional handover method provided by an embodiment of the present invention;

FIG. 7 is a schematic diagram of a hardware environment of a switching method provided by an embodiment of the present invention;

FIG. 8 is a schematic structural diagram of an optional switching device provided by an embodiment of the present invention;

FIG. 9 is a schematic structural diagram of an optional network device provided by an embodiment of the present invention.

Detailed ways

The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention.

At present, for cell handover, the handover process mainly includes three processes: handover measurement, handover judgment, and handover execution. Table 1 shows the new air interface (5G NR, 5G New Radio) measurement event judgment:

Table 1

Figure 1 is a flow chart of 5G handover signaling in the related art, as shown in Figure 1, including: a trigger link, a measurement link, a target cell decision link and a handover link; wherein,

In the triggering link, after the user equipment (UE, User Equipment) completes access or handover is successful, the base station (gNB) will immediately send measurement control information to the UE through radio resource control (RRC, Radio Resource Control).

In the measurement link, according to the current backbone configuration of the measurement control, the UE detects the wireless channel, but when the measurement report condition is met, it reports to the gNB through an event.

In the decision process of the target cell, the gNB uses measurement as the basic resource, selects the handover cell in the manner of reporting first, and selects the corresponding handover strategy.

In the handover link, the original base station applies for and allocates resources to the target base station, and then the source gNB makes a handover execution judgment, sends a handover command to the UE, and the UE performs handover and data forwarding.

However, the above handover method is based on the signal strength in the measurement report reported by the terminal and the handover parameters set by the handover event, which cannot be changed according to the actual situation, and the existing handover parameters are configured based on the ground user. Inconsistent with the air network coverage, this will lead to frequent switching of network cells during low-altitude operations, which will affect the communication quality of end users such as aerial drones.

Embodiment one

An embodiment of the present invention provides a handover method. FIG. 2 is a schematic flowchart of an optional handover method in an embodiment of the present invention. As shown in FIG. 2 , the handover method may include:

S201: receiving current flight status information of the drone;

In order to improve the communication quality of UAVs during cell handover, an embodiment of the present invention provides a handover method, which is applied to a network device, and the network device is connected to a base station. First, the UAV receives The reported current flight status information, where the flight status information includes: flight direction, flight speed, signal quality parameters and distance from adjacent cells. The flight direction refers to the direction the UAV is pointing at during flight ; The signal quality parameters may include: RSRP of the primary serving cell, SINR of the primary serving cell, RSRP of the neighboring cell, SINR of the neighboring cell, channel bit error rate and handover success rate.

Here, the distance from the adjacent cell refers to the distance between the current position of the UAV and the base station of the adjacent cell. Figure 3 is an optional distance from the adjacent cell provided in the embodiment of the present invention As shown in Figure 3, the formula for calculating the distance Di from the base station of adjacent cell i is:

Among them, H is the vertical distance between the UAV and the ground, h is the height of the base station of the adjacent cell, and _Dt is the distance between the projection of the UAV on the ground and the base station of the adjacent cell.

In this way, the distance between the UAV and the adjacent cell can be calculated by formula (1).

In addition, during the flight, the UAV can report its own flight status information to the network device according to the preset time interval, and can also report its own flight status to the network device according to the change of the UAV's flight status information. Information, here, this embodiment of the present invention does not specifically limit it.

In this way, the network device can receive the current flight status information reported by the UAV, so as to know the UAV's flight direction, flight speed, signal quality parameters and the distance to adjacent cells.

S202: Input the current flight state information into the trained deep neural network model to determine the parameter set of the switching event of the drone;

S203: sending the parameter set of the switching event of the drone to the base station;

Specifically, after receiving the current flight state information of the UAV through S101, the current flight state information is input into the trained deep neural network model, because the trained deep neural network model can determine that there is no The parameter set of the handover event of the man-machine, so the trained deep neural network model can determine the parameter set of the handover event of the UAV, and send the determined parameter set of the handover event of the UAV to the base station , where the parameter set of the handover event of the UAV is used by the base station to instruct the UAV to perform cell handover.

In this way, after the base station receives the parameter set of the handover event of the UAV, it applies for and allocates resources to the target base station according to the parameter set of the handover event, and then sends the handover command to the UAV, and the UAV performs the handover and Data forwarding to realize cell handover. Since the determined parameter set of the handover event is obtained in combination with the network environment of the UAV, the parameter set of the handover event determined in this way is more accurate, which improves the time when the UAV performs cell handover. communication quality.

In order to determine the parameter set of the switching event that is consistent with the network environment of the drone, it is necessary to determine the trained deep neural network model. In an optional embodiment, the above method may also include:

Obtain a training data set from the collected sample data set;

Input the training data set into the preset deep neural network model for training, and obtain the trained deep neural network model;

According to the trained deep neural network model, a trained deep neural network model is determined.

Specifically, the training data set is first obtained from the collected sample data. Here, a part of the sample data set can be selected from the sample data set as the training sample set. For example, the sample data set has a total of 1000 sets of data, and 800 sets of data can be selected. as a sample dataset.

Wherein, the sample data set is: the parameter set of the switching event corresponding to the flight state information and the flight state information; since the sample data set includes the parameter set of the switching event corresponding to the flight state information and the flight state information, the set of corresponding relations is composed of The data is input into the preset deep neural network model for training to optimize the model parameters in the deep neural network model, so that the trained deep neural network model can be obtained, and finally, the predicted value can be obtained according to the trained deep neural network model The error between the actual value and the actual value is used to determine the trained deep neural network model.

In order to determine the trained deep neural network model, in an optional embodiment, according to the trained deep neural network model, determine the trained deep neural network model, including:

Obtain a test dataset from the sample dataset;

Input the flight state information in the test data set into the trained deep neural network model to obtain the parameter set of the switching event;

The error between the parameter set of the calculated switching event and the parameter set of the switching event corresponding to the flight state information in the test data set;

When the error satisfies the second preset condition, the trained deep neural network model is determined as the trained deep neural network model;

When the error does not meet the second preset condition, return to the execution of obtaining the training data set from the collected sample data set.

Specifically, in order to optimize the training effect of the trained deep neural network model, the trained deep neural network model can be tested. In order to realize the test of the trained deep neural network model, first, obtain the test from the sample data set The data set, here, the test data set can be obtained from the sample data set except the training data set, so as to improve the accuracy of the test.

After obtaining the test data set, the test data set can include one or more sets of test data, and the flight state information in the test data set is input into the trained deep neural network model to obtain the switching event corresponding to the flight state information. The parameter set, and then calculate the error between the parameter set of the switching event obtained and the parameter set of the switching event corresponding to the flight state information in the test data set, so as to determine whether the error meets the second preset condition.

If it is satisfied, it means that the trained deep neural network model has passed the test. Therefore, the trained deep neural network model is directly determined as the trained deep neural network model; The model is trained until it passes the test, so as to determine the trained deep neural network model.

In order to determine whether the trained deep neural network model passes the test, in an optional embodiment, the above method may also include:

When the error is less than or equal to a preset error threshold, it is determined that the error satisfies a second preset condition;

When the error is greater than the preset error threshold, it is determined that the error does not meet the second preset condition.

Specifically, the calculated error is compared with the preset error threshold. Here, it should be noted that the calculated error can be one or more. When the error is one, compare whether the error is less than or equal to If the preset error threshold is less than or equal to, it is determined that the test is passed, and then it is determined that the error meets the second preset condition; if it is greater than, it means that the test fails, and it is determined that the error does not meet the second preset condition.

When the number of errors is multiple, compare whether each error is less than or equal to the preset error threshold, if they are all less than or equal to, it is determined that the test is passed, then it is determined that the error meets the second preset condition, if they are all greater than, it means that the test is not successful If passed, it is determined that the error does not meet the second preset condition. It can also be determined that the number of errors less than or equal to the preset error threshold occupies the proportion of the total number of errors. When the proportion is greater than or equal to the preset ratio threshold, it is determined that the test is passed, and then it is determined that the error meets the second preset condition. If the ratio is less than When the ratio threshold is preset, it means that the test fails, and it is determined that the error does not meet the second preset condition.

In this way, through the setting of the above-mentioned second preset condition, the effect of the trained deep neural network model can be further improved, so as to obtain a more accurate parameter set for switching events.

In order to improve the effect of the trained deep neural network model, it is necessary to filter the collected data when obtaining the sample data set, so as to obtain the optimal sample data set that is conducive to model training. In an optional implementation In an example, the above method may also include:

Based on the preset flight direction, the data set of the flight direction is determined according to the step size of the preset flight direction;

Based on the preset flight position, determine the data set of the distance between the preset flight position and the adjacent cell;

Based on the preset flight speed, according to the preset flight speed step size, determine the data set of the flight speed;

According to the ratio of 1:1:1, the data set of flight direction, the data set of flight speed and the data set of distance are respectively used to form a three-dimensional vector;

The corresponding relationship between each vector in the three-dimensional vector and each set of parameter sets in the preset switching event parameter set is formed into an array;

From the array, select the dataset for the sample.

Specifically, the flight direction, flight position and flight speed are set in advance, and then the data set of the flight direction is determined according to the step size of the preset flight direction. For example, the preset flight direction is the direction from cell A to cell B , and the direction is represented by 0°, and a flight direction is taken every 1° within the direction of plus or minus 60° to obtain the data set of the flight direction.

Similarly, set the flight position in advance, and then determine the distance between the flight position and the adjacent cell, for example, if there are 3 adjacent cells, then the distance to the first adjacent cell can be obtained by formula (1), and the distance to the second The distance from the first adjacent cell and the distance from the third adjacent cell constitute the data set of the distance from the adjacent cell.

For the data set of flight speed, set the flight speed in advance, for example, when the flight speed is 0, the step size of the flight speed is 0.1m/s, then take the flight speed on [0, Vmax] to get the data of the flight speed set.

Based on the determined flight direction data set, flight speed data set and distance data set, a three-dimensional vector can be formed according to the ratio of 1:1:1, and multiple three-dimensional vectors can be obtained by combining them, and then multiple three-dimensional vectors can be obtained The correspondence between each vector in the vector and the parameter set of the preset switching event forms an array, and finally, the data set of the sample can be selected from each array.

Since the above-mentioned three-dimensional vectors collect as much flight state information of the UAV in various situations as possible, so that the flight state information included in the three-dimensional vectors is as comprehensive as possible, then the data set of samples selected from each array should be as comprehensive as possible. The comprehensiveness is conducive to optimizing the parameters of the deep neural network model.

In order to improve the optimization effect of the deep neural network model, in an optional embodiment, the data set of samples is selected from the array, including:

According to each array in the array, call the channel model (LOS, Line of Sight) path propagation model to obtain the signal quality parameters corresponding to each array;

From the array and the signal quality parameters corresponding to each array, select the flight status information and Correspondence between parameter sets of switching events;

The selected flight state information and the parameter set of the switching event corresponding to the selected flight state information are determined as a sample data set.

That is to say, when selecting a data set of samples from each array and the signal quality parameters corresponding to each array, it can first be determined that the UAV is in each array and is in the signal quality parameters corresponding to each array.

After knowing the signal quality parameters in each flight state, from each array and the signal quality parameters corresponding to each array, first select the array in each flight direction, the corresponding signal quality parameters of the array, and from From the signal quality parameters, find out the flight status information whose handover failure rate and signal error rate meet the first preset condition and the parameter set of the handover event corresponding to the flight status information, and combine the flight status that meets the first preset condition with the first preset condition. The parameter set of the switching event corresponding to the flight state of the preset condition is determined as a sample data set.

In this way, the data in the sample data set are all data whose switching failure rate and signal error rate meet the first preset condition, wherein the following method is used to determine whether the first preset condition is met, which is conducive to optimizing the performance of the deep neural network. parameters, thereby improving the optimization effect of the deep neural network model.

Further, in order to improve the optimization effect of the deep neural network model, in an optional embodiment, the above method also includes:

When the handover failure rate in the signal quality parameter is less than or equal to the preset failure rate threshold, and/or, when the channel bit error rate in the signal quality parameter is less than or equal to the preset bit error rate threshold, determine the handover rate and the channel error rate in the channel quality parameter The code rate meets the first preset condition;

When the handover failure rate in the signal quality parameter is greater than the preset failure rate threshold, and the channel bit error rate in the signal quality parameter is greater than the preset bit error rate threshold, it is determined that the handover rate and the channel bit error rate in the channel quality parameter do not meet the first threshold. a preset condition.

Specifically, comparing the handover failure rate with a preset failure rate threshold, comparing the channel bit error rate with a preset bit error rate threshold, when the handover failure rate is less than or equal to the preset failure rate threshold, and/or , when the channel bit error rate in the signal quality parameter is less than or equal to the preset bit error rate threshold, it means that the channel vector is better at this time, so it is determined that the switching rate and the channel bit error rate in the channel quality parameter meet the first preset condition, and select the The flight state information under the condition and the parameter set of the handover event corresponding to the flight state information under the condition are used as the sample data set; when the handover failure rate in the signal quality parameter is greater than the preset failure rate threshold, and the channel error in the signal quality parameter When the rate is greater than the preset bit error rate threshold, it means that the channel vector is poor at this time, so it is determined that the switching rate and channel bit error rate in the channel quality parameters do not meet the first preset condition, and the flight status information under this condition is compared with the The parameter set of the switching event corresponding to the flight state information under the condition is not used as the sample data set.

In this way, by filtering the collected data sets to obtain sample data sets, the parameters of the deep neural network model can be further optimized, and the effect of the trained deep neural network model can be improved.

The following examples are used to describe the handover method described in the foregoing one or more embodiments.

Fig. 4 is a schematic diagram of an optional UAV flight provided by the embodiment of the present invention. As shown in Fig. 4, during the flight of the UAV from cell A to cell B, the flight speed, no The relative position of the man-machine and the cell, the parameter set of the handover event, etc. will all affect the communication quality, handover failure rate and bit error rate of the UAV at the current position.

By changing the parameter sets of different positions, different flight speeds and different switching events, the signal quality parameters of the UAV measured in different states can be obtained. There is a certain correspondence between these parameters, and the deep learning model can be used to represent complex nonlinear relationships. When using a deep learning algorithm, for example, a deep neural network (DNN, DeepNeural Networks) model to invert the parameter set of the switching event , it is necessary to prepare a large number of labeled sample data sets for training the model and predicting the inversion effect of the model.

In this example, the model training and inversion steps of the parameter set of the optimal low-altitude cell handover event are as follows:

Step 1: Modify the model parameters of the network simulation platform;

Specifically, due to the differences between UAVs operating in the air and ground operations, it is necessary to set the propagation algorithm model of the network simulation platform to use the LOS path propagation model in low altitude.

Step 2: According to the input parameters in the simulation platform, according to the different states of the UAV, set different step length parameters to collect the data set.

Step 3: The collected data set is: the signal quality parameter S of [V, D] in different states, and the tag value is the parameter set P of the corresponding handover event.

Step 4: Preprocess the collected data set, and select the corresponding relationship between [S, V, D] and P when the handoff failure rate and channel bit error rate in S are the smallest in each test direction A The composed data set is obtained to obtain the preprocessed data set.

Step 5: Use the preprocessed data set to train the DNN model;

Step 6: Use the trained DNN model to invert the parameter set of the optimal cell handover event in real time according to the real-time measurement of the flight situation.

Among them, the acquisition of the optimal training data set may include:

During the flight of the drone, the flight route is generally consistent with the direction of the nose. Referring to the above figure 4, assuming that the drone flies from cell A to cell B, the initial access cell is cell A, and the detected neighbor cell For cell B, by measuring and reporting the aircraft position, flight height and nose direction of the drone in real time, the flight angle of the drone is centered on the nose direction, within the direction of plus or minus 60° of the aircraft direction (assuming that a cell covers 120°) at intervals of 1°, and in each flight direction, increase the flight speed at the interval of [0, Vmax] (the maximum flight speed of the UAV is Vmax) at a speed of 0.1m/s to obtain each Velocity array V for directions.

V=[V ₁₁ ,V ₁₂ ,V ₁₃ ,...,V _in ] (2)

Among them, i represents flying at the speed Vin when it deviates from the aircraft heading i°, and records the array D of the straight-line distance between the UAV and the i-th neighbor in the current altitude state.

D＝[D ₁ ,D ₂ ,D ₃ ,...,D _i ] (3)

Wherein, the calculation formula of the distance D _i is shown in the above formula (1).

At the same time as the flight course is changed each time, the parameter set of the switching event is also changed according to a certain step size, and the parameter set P of the switching event is obtained.

P＝[A1hys, A1time, A2hys, A2time, A3ocn, A3hys, A3ofs, A3time, A4hys, A4thr, A4time, A5the, A5hys, A5the2, A5time, A6hys, A6time, B1hys, B1time, B2hys, B2time, B2the] (4)

Among them, Aitime indicates the duration of the event Ai continuously meeting the event entry condition, that is, the time lag; Aihys indicates the amplitude hysteresis of the event Ai measurement result; Aiocn indicates the cell-specific offset CIO of the neighboring cell in the event Ai system; Aiofs indicates the event Ai measurement result Bias; Aithe indicates the threshold value of the event corresponding to the event configuration; and the signal quality parameter S1 in each state is measured.

S1＝[RSRPs,SINRs RSRPni,SINRni,BLERi,HOi] (5)

Among them, RSRPs represents the RSRP of the primary serving cell, SINRs represents the SINR of the primary serving cell, RSRPni represents the measured RSRP of the i-th neighbor cell, SINRni represents the measured SINR value of the i-th neighbor cell; BLERi represents the parameter status of the i-th cell The channel bit error rate under , HOi represents the handover success rate under the ith parameter state.

In this way, the initial sample space [S1,P] of the corresponding relationship can be obtained. First, the data set [S1,P] is preprocessed, and the handover failure rate BLER and the channel bit error rate HO are selected to be the smallest in each test direction The training data set S:

S = [RSRPs, SINRs, RSRPni, SINRni] (6)

In this way, combining formula (2), formula (3) and formula (6), the input vector M of the final network model can be obtained:

M = [S, V, D] (7)

After the training data set of the DNN model is obtained, the DNN model is established and corrected. Fig. 5 is a schematic diagram of an optional DNN model establishment and correction process provided by the embodiment of the present invention. As shown in Fig. 5, a deep DNN is used The model structure is shown in Figure 5. It mainly consists of three parts: input layer, hidden layer and output layer. The input vector of the input layer is M=[S, V, D], the output layer outputs the parameter set P of switching events, and the hidden Layer neurons receive input data from the previous layer, and then use the activation function to perform nonlinear transformation, and then continue to output to the next layer.

The output vector a ⁱ (i=1,2,3...15) of the i-th layer can be expressed as:

Among them, m represents the number of neurons in the i-1th layer; w ^i-1 , b ^i-1 represent the weight coefficient matrix and bias vector of the i-1th layer respectively. The input vector of the input layer is the S parameter as follows:

a ⁰ =[M] ^T (9)

Wherein, M=[S, V, D], as shown in formula (7).

The vector of the output layer is the parameter set of the switching event:

a ^P = [P] ^T (10)

Among them, f(x) represents the activation function, and a linear rectification function (ReLU, Rectified LinearUnit) can be used as the activation function. The RMSprop (The Root Mean Square Prop) algorithm is used to update the weight coefficient matrix and bias vector parameters of the network, which are used to establish the main structural parameters of the DNN model.

Among them, FIG. 6 is a schematic flow diagram of the construction and prediction of an optional switching method provided by the embodiment of the present invention. The entire algorithm construction and prediction process is shown in FIG. 6, which is divided into three parts. The first part is training sample collection , the second part is training the DNN model, and the third part is switching parameter inversion.

For the first part, first set the movement speed of different positions and different step lengths of the UAV, and set the event strategy parameters of different compensations, and input the event strategy parameters of different positions of the UAV with different step lengths and different step lengths into the network simulation software , get the network signal quality parameters.

For the second part, event strategy parameters with different step lengths and corresponding network signal quality parameters are used as training data sets. Specifically, refer to the above-mentioned acquisition part of the optimal training data set, which will not be repeated here. Then normalize the training data set.

Use the normalized processing method to obtain the processed data set, realize the optimization of the DNN model parameters, obtain the trained DNN model, and input the test data set into the trained DNN model to obtain the predicted value, and calculate the predicted value and The error between the actual values. When the error is less than or equal to the error threshold, the test passes and the trained DNN model is obtained; otherwise, the test fails.

For the third part, the UAV reports the flight status information measured by the actual flight to the network equipment. The flight status information can include: signal quality parameters, flight direction, flight speed and distance from adjacent cells, and normalized For processing, input the normalized data into the trained DNN model, output the prediction vector, and denormalize the prediction vector to obtain the event switching strategy.

Among them, in the handover algorithm correction and location deployment, Fig. 7 is a schematic diagram of the hardware environment of a handover method provided by the embodiment of the present invention. As shown in Fig. 7, after the algorithm model is trained, it belongs to the offline model and does not need to occupy a lot of memory , during each actual flight, the flight measurement data of the UAV is stored in the edge cloud server and added to the training data set. When the cloud is idle, the DNN model is re-updated to increase the accuracy of the model inversion. In order to ensure the delay requirements of the real-time switching algorithm, the switching algorithm is deployed on the edge cloud server using Mobile Edge Computing (MEC, Mobile Edge Computing) technology to reduce the switching delay.

Through the above example, the cell intelligent switching method can intelligently give the optimal cell switching strategy in real time according to the flight position of the UAV and the network situation, and combine the propagation model, the flying speed and height of the UAV in the air, and the relative position with the base station Taking into account the influence parameters, combined with the deep learning DNN model, the optimal decision strategy can be given in real time according to the flight environment. The switching algorithm can be deployed offline, and the algorithm can be deployed on the edge server, and the algorithm can be updated when the server is idle. Guaranteed low latency for real-time parameter selection and switching of algorithms.

A handover method provided by the present invention, the method includes: receiving the current flight state information of the drone, wherein the flight state information includes: flight direction, flight speed, signal quality parameters and distance from adjacent cells, and the current The flight state information is input into the trained deep neural network model to determine the parameter set of the switching event of the UAV, and the parameter set of the switching event of the UAV is sent to the base station, wherein the switching event of the UAV The parameter set is used by the base station to instruct the UAV to perform cell handover; that is, in the embodiment of the present invention, through the trained deep neural network model, the parameters of the handover event corresponding to the current flight state of the UAV can be determined In this way, the parameter set of the switching event determined by using the trained deep neural network model combined with the current flight status information of the UAV is more suitable for the flight environment of the UAV. Then, based on the determined parameter set of the switching event Set to switch cells, so that the UAV can ensure the communication quality of the UAV when it performs cell switching, thereby improving the communication quality of the UAV when it performs cell switching.

Embodiment two

Based on the same inventive concept, an embodiment of the present invention also provides a switching device. FIG. 8 is a schematic structural diagram of an optional switching device provided by an embodiment of the present invention. As shown in FIG. 8, the switching device may include:

The receiving module 81 is used to receive the current flight state information of the drone; wherein, the flight state information includes: flight direction, flight speed, signal quality parameters and distance from adjacent cells;

Determining module 82, is used for inputting current flight status information into the trained deep neural network model, to determine the parameter set of the switching event of unmanned aerial vehicle;

The handover module 83 is configured to send the parameter set of the handover event of the UAV to the base station; wherein, the parameter set of the handover event of the UAV is used by the base station to instruct the UAV to perform cell handover.

In an optional embodiment, the above-mentioned device is also used for:

Obtain a training data set from the collected sample data set; wherein, the sample data set is: the flight state information and the parameter set of the switching event corresponding to the flight state information;

Input the training data set into the preset deep neural network model for training, and obtain the trained deep neural network model;

According to the trained deep neural network model, a trained deep neural network model is determined.

In an optional embodiment, the above device determines the trained deep neural network model according to the trained deep neural network model, including:

Obtain a test dataset from the sample dataset;

Input the flight state information in the test data set into the trained deep neural network model to obtain the parameter set of the switching event;

The error between the parameter set of the calculated switching event and the parameter set of the switching event corresponding to the flight state information in the test data set;

When the error satisfies the second preset condition, the trained deep neural network model is determined as the trained deep neural network model;

When the error does not meet the second preset condition, return to the execution of obtaining the training data set from the collected sample data set.

In an optional embodiment, the above-mentioned device is also used for:

When the error is less than or equal to a preset error threshold, it is determined that the error satisfies a second preset condition;

When the error is greater than the preset error threshold, it is determined that the error does not satisfy the second preset condition.

In an optional embodiment, the above-mentioned device is also used for:

Based on the preset flight direction, the data set of the flight direction is determined according to the step size of the preset flight direction;

Based on the preset flight position, determine the data set of the distance between the preset flight position and the adjacent cell;

Based on the preset flight speed, according to the preset flight speed step size, determine the data set of the flight speed;

According to the ratio of 1:1:1, the data set of flight direction, the data set of flight speed and the data set of distance are respectively used to form a three-dimensional vector;

The corresponding relationship between each vector in the three-dimensional vector and each set of parameter sets in the preset switching event parameter set is formed into an array;

From the array, select the dataset for the sample.

In an optional embodiment, the above device selects a data set of samples from the array, including:

According to each array in the array, invoke the LOS path propagation model to obtain the signal quality parameters corresponding to each array;

From the signal quality parameters corresponding to each array, select the flight state information and Correspondence between parameter sets of switching events;

The selected flight state information and the parameter set of the switching event corresponding to the selected flight state information are determined as a sample data set.

In an optional embodiment, the above-mentioned device is also used for:

When the handover failure rate in the signal quality parameter is less than or equal to the preset failure rate threshold, and/or, when the channel bit error rate in the signal quality parameter is less than or equal to the preset bit error rate threshold, determine the handover rate and the channel error rate in the channel quality parameter The code rate meets the first preset condition;

When the handover failure rate in the signal quality parameter is greater than the preset failure rate threshold, and the channel bit error rate in the signal quality parameter is greater than the preset bit error rate threshold, it is determined that the handover rate and the channel bit error rate in the channel quality parameter do not meet the first threshold. a preset condition.

In practical applications, the receiving module 81, the determining module 82 and the switching module 83 can be implemented by a processor located on the network device, specifically a central processing unit (Central Processing Unit, CPU), a microprocessor (Microprocessor Unit, MPU), Implementations such as Digital Signal Processing (DSP) or Field Programmable Gate Array (Field Programmable Gate Array, FPGA).

FIG. 9 is a schematic structural diagram of an optional network device provided by an embodiment of the present invention. As shown in FIG. 9, an embodiment of the present invention provides a network device 900, including:

A processor 91 and a storage medium 92 storing instructions executable by the processor 91, the storage medium 92 relies on the processor 91 to perform operations through a communication bus 93, when the instructions are executed by the processor 91, Execute the handover method described in Embodiment 1 above.

It should be noted that, in actual application, various components in the terminal are coupled together through the communication bus 93 . It can be understood that the communication bus 93 is used to realize connection and communication between these components. In addition to the data bus, the communication bus 93 also includes a power bus, a control bus and a status signal bus. However, the various buses are labeled as communication bus 93 in FIG. 9 for clarity of illustration.

An embodiment of the present invention provides a computer storage medium, storing executable instructions, and when the executable instructions are executed by one or more processors, the processors execute the switching method described in Embodiment 1.

Wherein, the computer-readable storage medium may be a magnetic random access memory (ferromagnetic random access memory, FRAM), a read-only memory (Read Only Memory, ROM), a programmable read-only memory (Programmable Read-Only Memory, PROM), an erasable In addition to programmable read-only memory (Erasable Programmable Read-Only Memory, EPROM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, EEPROM), flash memory (Flash Memory), magnetic surface memory, optical disc, Or CD-ROM (Compact Disc Read-Only Memory, CD-ROM) and other memory.

Those skilled in the art should understand that the embodiments of the present invention may be provided as methods, systems, or computer program products. Accordingly, the present invention can take the form of a hardware embodiment, a software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage and optical storage, etc.) having computer-usable program code embodied therein.

The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It should be understood that each procedure and/or block in the flowchart and/or block diagram, and a combination of procedures and/or blocks in the flowchart and/or block diagram can be realized by computer program instructions. These computer program instructions may be provided to a general purpose computer, special purpose computer, embedded processor, or processor of other programmable data processing equipment to produce a machine such that the instructions executed by the processor of the computer or other programmable data processing equipment produce a An apparatus for realizing the functions specified in one or more procedures of the flowchart and/or one or more blocks of the block diagram.

These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing apparatus to operate in a specific manner, such that the instructions stored in the computer-readable memory produce an article of manufacture comprising instruction means, the instructions The device realizes the function specified in one or more procedures of the flowchart and/or one or more blocks of the block diagram.

These computer program instructions can also be loaded onto a computer or other programmable data processing device, causing a series of operational steps to be performed on the computer or other programmable device to produce a computer-implemented process, thereby The instructions provide steps for implementing the functions specified in the flow chart or blocks of the flowchart and/or the block or blocks of the block diagrams.

The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the protection scope of the present invention.

Claims (10)

1. A method of handover, comprising:

receiving current flight state information of the unmanned aerial vehicle; wherein the flight status information includes: flight direction, flight speed, signal quality parameters and distance to neighboring cells;

Inputting the current flight state information into a trained deep neural network model to determine a parameter set of a switching event of the unmanned aerial vehicle;

issuing a parameter set of the switching event of the unmanned aerial vehicle to a base station; the parameter set of the switching event of the unmanned aerial vehicle is used for the base station to instruct the unmanned aerial vehicle to conduct cell switching.

2. The method according to claim 1, wherein the method further comprises:

acquiring a training data set from the acquired sample data set; wherein the sample dataset is: a parameter set of a switching event, wherein the flight state information corresponds to the flight state information;

inputting the training data set into a preset deep neural network model for training to obtain a trained deep neural network model;

and determining the trained deep neural network model according to the trained deep neural network model.

3. The method of claim 2, wherein the determining the trained deep neural network model from the trained deep neural network model comprises:

obtaining a test dataset from the sample dataset;

Inputting the flight state information in the test data set into the trained deep neural network model to obtain a parameter set of a switching event;

calculating the error between the obtained parameter set of the switching event and the parameter set of the switching event corresponding to the flight state information in the test data set;

when the error meets a second preset condition, determining the trained deep neural network model as the trained deep neural network model;

and when the error does not meet the second preset condition, returning to execute the acquisition of the training data set from the acquired sample data set.

4. A method according to claim 3, characterized in that the method further comprises:

when the error is smaller than or equal to a preset error threshold value, determining that the error meets the second preset condition;

and when the error is larger than the preset error threshold value, determining that the error does not meet the second preset condition.

5. The method according to claim 2, wherein the method further comprises:

determining a data set of the flight direction according to the step length of the preset flight direction based on the preset flight direction;

Determining a data set of the distance between the preset flight position and the adjacent cell based on the preset flight position;

determining a data set of the flying speed according to a preset flying speed step length based on the preset flying speed;

according to 1:1:1, respectively utilizing the data sets of the flying directions, wherein the data sets of the flying speeds and the data sets of the distances form three-dimensional vectors;

forming an array by the corresponding relation formed by each vector in the three-dimensional vector and each group of parameter sets in the preset switching event parameter sets;

and selecting a data set of the sample from the array.

6. The method of claim 5, wherein the selecting the data set of the sample from the array comprises:

according to each array in the arrays, calling an LOS path propagation model to obtain signal quality parameters corresponding to each array;

selecting a corresponding relation between the flight state information and a parameter set of a switching event when the switching failure rate in the signal quality parameters and the channel error rate in the signal quality parameters meet a first preset condition aiming at each flight direction from the plurality of groups of signal quality parameters corresponding to each group of groups;

And determining the parameter set of the switching event corresponding to the selected flight state information and the selected flight state information as the sample data set.

7. The method of claim 6, wherein the method further comprises:

when the switching failure rate in the signal quality parameter is smaller than or equal to a preset failure rate threshold value, and/or the channel error rate in the signal quality parameter is smaller than or equal to a preset error rate threshold value, determining that the switching rate and the channel error rate in the channel quality parameter meet the first preset condition;

and when the switching failure rate in the signal quality parameter is larger than a preset failure rate threshold value and the channel error rate in the signal quality parameter is larger than a preset error rate threshold value, determining that the switching rate and the channel error rate in the channel quality parameter do not meet the first preset condition.

8. A switching device, comprising:

the receiving module is used for receiving the current flight state information of the unmanned aerial vehicle; wherein the flight status information includes: flight direction, flight speed, signal quality parameters and distance to neighboring cells;

the determining module is used for inputting the current flight state information into a trained deep neural network model so as to determine a parameter set of a switching event of the unmanned aerial vehicle;

The switching module is used for issuing the parameter set of the switching event of the unmanned aerial vehicle to the base station; the parameter set of the switching event of the unmanned aerial vehicle is used for the base station to instruct the unmanned aerial vehicle to conduct cell switching.

9. A network device, the network device comprising:

a processor and a storage medium storing instructions executable by the processor, the storage medium performing operations in dependence on the processor through a communication bus, the instructions, when executed by the processor, performing the handover method of any one of claims 1 to 7.

10. A computer storage medium storing executable instructions which, when executed by one or more processors, perform the handover method of any one of claims 1 to 7.

Images