Method and system for switching between NSA and LTE

Abstract

The invention discloses a method and system for switching between NSA and LTE, and relates to the field of mobile phone terminal mobile communication, and the method comprises the steps: obtaining a threshold t of a timer through the prediction of a neural network; the UE monitors whether the TAU access replied by the communication network is received or not within the threshold time t; if the UE does not receive the TAU accept replied by the network in the t, reporting a limited service state to the NSA, notifying the UI to display no signal in a signal bar by the UE, and meanwhile, recording the TAU failure and recording the transmitting power P1 at the time by the UE; after a period of time greater than the threshold time t, when the UE triggers the switching between the TLE and the NSA again, the Modem of the UE checks whether the TAU request receives the TAU access last time or not, and if the TAU request does not receive the TAU access last time, the Tx transmitting power of the UE is increased from P1 to P2; the UE judges whether the increased Tx transmitting power P2 exceeds the maximum allowable transmitting power Pmax of the UE or not, and if P2 is smaller than or equal to Pmax, the UE sends TAU request to the network through the P2. Aiming at the low switching success rate between the NSA and the LTE in the prior art, the invention improves the switching success rate.

Description

Technical Field

[0001] The present application relates to the field of mobile communications for mobile terminals, and more specifically, to a method and system for switching between NSA and LTE. Background Art

[0002] With the gradual commercial deployment of 5G networks, NSA (non-standalone networking) plays an important role as a transition solution for 5G in the evolution to SA (standalone networking). NSA networks rely on the existing 4G LTE core network and achieve synergy between 4G and 5G through dual connection (EN-DC) technology to provide users with an enhanced mobile broadband experience. However, in the scenario of frequent switching between NSA and LTE, terminal devices face the challenge of optimizing the TAU (tracking area update) process.

[0003] At present, major mobile phone manufacturers mainly follow the 3GPP protocol and their own designs, and follow the network instructions to take the next action. However, for real-time network abnormality scenarios, the existing solutions do not provide a dynamic optimization processing mechanism. As a result, during the switching process, once an abnormal situation such as TAU request failure or network response timeout occurs, the terminal device will enter the limited service state and display no signal on the signal bar, which greatly affects the user experience. Therefore, a technical solution for design optimization on the mobile terminal side is urgently needed to avoid such problems and improve the reliability and user perception of NSA / LTE switching.

[0004] In the existing NSA / LTE switching process, the terminal device needs to perform the TAU process to update the tracking area information. Specifically, the terminal sends a TAU request to the network, and the NAS layer starts a timer. If the network replies with a TAU accept message to the terminal before the timer expires, it means that the TAU process is successfully completed. On the contrary, if the network's TAU accept response is not received after the timer expires, it means that the TAU process has failed, and the terminal will enter the limited service state.

[0005] The above unoptimized solution has the following shortcomings: When NSA switches to LTE, if the terminal TAU request is not successfully sent to the network side, or if the TAU accept is not replied to the terminal within the validity period of the timer due to network anomalies, the terminal will be triggered to enter the limited service state. At this time, the user interface signal bar icon will be displayed as no signal, which seriously affects the user experience and perception. Summary of the invention

[0006] 1. Technical issues to be solved

[0007] In view of the problem of low switching success rate between NSA and LTE in the prior art, the present application provides a method and system for switching between NSA and LTE, which improves the switching success rate and shortens the out-of-service time by introducing a neural network model to dynamically predict the TAU timer threshold and combining it with a mechanism of gradually increasing the transmission power when TAU fails.

[0008] 2. Technical solution

[0009] The purpose of this application is achieved through the following technical solutions.

[0010] One aspect of an embodiment of the present application provides a method for switching between NSA and LTE, including: when a user equipment UE triggers a switch between LTE and NSA, the user equipment UE sends a TAU request to the network with a first transmission power value P1; the NAS of the UE starts a limited buffer timer, and uses the network connection status of the user's mobile phone, the 4G or 5G network speed and the device performance as input parameters, and obtains the threshold t of the timer through a neural network prediction; the UE monitors whether a TAU accept reply from the communication network is received within the threshold time t; if the UE does not receive the TAU accept reply from the network within the threshold time t, the timer times out, the UE's modem reports the limited service status to the NSA, the UE notifies the UI to display the signal bar as no signal, and the UE records the TAU failure and the transmission power P1 here; after a period of time greater than the threshold time t, when the UE triggers the switch between LTE and NSA again, the UE's modem checks whether the last TAU ​​request received the TAU accept, if not received, increase the UE's Tx transmit power from P1 to P2, where P2 is greater than P1; the UE determines whether the increased Tx transmit power P2 exceeds the UE's maximum allowed transmit power Pmax. If P2 is less than or equal to Pmax, the UE sends a TAU request to the network with a second transmit power value P2, and goes to the UE's NAS to restart the limited buffer timer step.

[0011] Specifically, the present application proposes an intelligent method for optimizing switching between NSA (Non-Standalone) and LTE (Long Term Evolution) networks. The method significantly improves the switching success rate and shortens the service interruption time by dynamically adjusting key parameters such as transmit power and timer thresholds, thereby improving the user's communication experience.

[0012] When the user equipment (UE) triggers the switch between LTE and NSA, the UE first sends a TAU (Tracking Area Update) request message to the network with a preset first transmit power value P1. P1 here can be an empirical value, or a better initial value obtained based on the historical switching data statistics of the UE. After sending the TAU request, the NAS (Non-Access Stratum) of the UE immediately starts a timer called limited buffer. Unlike the traditional solution, the present application uses a neural network model to predict the timeout threshold t of the timer. Specifically, the input parameters of the model include: the current network connection status of the UE, such as signal strength, channel quality, etc.; the data transmission rate of the 4G or 5G network currently accessed by the UE; the hardware performance indicators of the UE itself, such as the processor main frequency, memory size, etc. By training and using this neural network model, the UE can predict an optimal timer threshold t in real time according to its actual situation and network conditions, replacing the fixed empirical value in the traditional solution. After sending the TAU request and starting the timer, the UE enters a waiting state to monitor whether a TAU accept message is received from the network side within the threshold time t. Here, t is an ideal value predicted by the UE based on the current environment, representing how long the UE hopes the network will take to complete the TAU process. If the UE receives the TAU accept message as expected within time t, the switching process is successfully completed and the UE has successfully registered with the target network. However, if the UE does not receive the TAU accept message until the timer expires, it means that the TAU attempt has failed. To deal with this situation, the UE will take a series of failure processing and optimization measures: the UE's Modem reports the limited service status to the higher-level NAS protocol to inform that the current network service is limited; the UE notifies the user interface (UI) to set the signal strength icon on the screen to a no signal state to intuitively inform the user; the UE records the TAU failure event internally and saves the currently used transmission power value P1 to prepare for optimizing subsequent retry processes.

[0013] After a period of time greater than the timer threshold t, if the UE triggers the switch between LTE and NSA again due to reasons such as mobility, the UE's Modem will first check the result of the last TAU ​​request. If the last TAU ​​was successfully completed, the UE can continue to use the transmit power P1 to perform a new round of TAU procedures. However, if the last TAU ​​has failed, the UE believes that the power value of P1 may be too low and needs to be adjusted appropriately. The strategy adopted in this application is to increase the transmit power from P1 to a new value P2. Here, the difference between P2 and P1 can be a preset fixed step size, or it can be calculated in real time through a certain adaptive algorithm. To ensure that the adjusted transmit power is legal and compliant, the UE needs to compare the new power value P2 with its own maximum allowed transmit power Pmax: if P2 does not exceed Pmax, the UE believes that P2 is an acceptable legal value, and P2 can be used instead of P1 to perform a new round of TAU procedures and return to step 2 to restart the timer; but if P2 exceeds Pmax, the UE cannot use P2. To avoid interference problems caused by excessive power, the UE will abandon power adjustment at this time and continue to use the original P1 to perform TAU retries. The above process of the present application is continuously cycled until the UE successfully obtains network acceptance in a round of TAU retries and completes the switch between LTE and NSA, or reaches the preset maximum number of retries, and actively terminates the process. By introducing mechanisms such as neural network prediction timer and adaptive transmit power adjustment, the present application greatly improves the success rate of UE in the LTE and NSA switching process, and reduces the delay and resource waste caused by repeated retries. In addition, since key parameters can be dynamically optimized in real time according to the actual state of the UE, the switching performance is no longer limited to the traditional static configuration, but can adapt to different network environments and user needs.

[0014] Further, if the UE determines that a TAU accept reply from the network is received within the threshold time t, the NSA determines that the TAU process is successful and controls the UE to exit the present switching method.

[0015] Furthermore, the neural network includes a fitting network of N neurons. Specifically, the fitting network refers to a type of feedforward neural network specifically used for function fitting and regression analysis. It approximates and simulates the inherent nonlinear relationship of the unknown system by learning a large number of input-output data pairs. The input of the fitting network is various factors that affect the performance of the TAU, such as the network connection status of the UE, the data transmission rate, the hardware performance, etc.; and the output is the predicted value of the timer threshold t. By continuously training and optimizing the fitting network, it can automatically adjust and generate the optimal threshold prediction according to the changes in the input parameters. The fitting network usually adopts a multi-layer perceptron (MLP) structure, including an input layer, a hidden layer and an output layer. Among them, the hidden layer is composed of several neurons, which are responsible for feature extraction and nonlinear transformation. The fitting network in this application includes N neurons, where N can be flexibly set according to the complexity and accuracy requirements of the actual system. In order to further improve the prediction performance and save computing resources, this application introduces a transfer learning mechanism based on the fitting network. Specifically, after each use of the fitting network prediction timer threshold t, the UE will compare the actual response time of the TAU accept message received with the predicted value: if the deviation between the two is less than a preset threshold, it means that the current fitting network already has a good prediction capability and does not need to be frequently updated, and the UE can continue to use it; but if the deviation exceeds the threshold, it means that the prediction accuracy of the fitting network needs to be improved. At this time, the UE will trigger an incremental learning process and use the new actual response time sample to fine-tune the network.

[0016] Furthermore, the predicted value of the timer threshold t predicted by the neural network is compared with the obtained response time value of the UE receiving TAU accept to obtain a prediction deviation; it is determined whether the prediction deviation is greater than a preset deviation threshold, and when it is greater than the deviation threshold, some layers of the neural network are frozen, and the some layers are at least one hidden layer in the neural network except the input layer and the output layer; after freezing some layers, a new response time value of the UE receiving TAU accept is obtained as new training data; the unfrozen neural network layers are recompiled and trained using the new training data to obtain an updated neural network model.

[0017] Specifically, unlike traditional incremental learning, this application adopts a selective fine-tuning strategy, that is, only retraining part of the layers of the fitting network instead of the entire network. Freeze the underlying feature extraction layer to retain its learned common features, avoid repeated learning, and speed up convergence; only adjust the top regression layer to adapt it to the new data distribution, improve prediction accuracy, and reduce computational overhead. In terms of technical implementation, freezing part of the layers of the neural network can be achieved by setting the corresponding learning rate. The learning rate determines the update amplitude of the weights of each layer during the back propagation process. We can set the learning rate of the layer that needs to be frozen to 0 so that its weight remains unchanged during training; and set the learning rate of the layer that needs to be fine-tuned to a smaller positive value, allowing its weight to be adjusted to a limited extent. More specifically, if our fitting network contains 1 input layer, 3 hidden layers, and 1 output layer. When incremental learning is triggered, we can: set the learning rate of the input layer and the first two hidden layers to 0, freeze their weights; set the learning rate of the third hidden layer and the output layer to 0.01, allowing fine-tuning; then, use the new actual response time samples to retrain the network for several rounds, update the weights of the last two layers, and obtain a fitted network adapted to the new data. In this way, through transfer learning and partial layer freezing, we maximize the use of existing knowledge while achieving continuous evolution and optimization of the fitting network's prediction capabilities through small incremental learning, which not only saves computing resources, but also improves the real-time and accuracy of timer threshold prediction. It is a very effective strategy.

[0018] Among them, in this application, we introduce a preset threshold to determine whether the deviation between the neural network prediction value and the actual value is acceptable. When the deviation exceeds this threshold, we believe that the current neural network prediction performance is not ideal and needs to be optimized and adjusted through incremental learning. General principles for threshold setting: According to the hardware performance and power level of the UE, a benchmark threshold is set to meet the needs of incremental learning without causing performance and power consumption problems; on this basis, the benchmark threshold is fine-tuned according to the needs of different services to obtain an adaptive and personalized threshold; during the operation of the UE, the changes in network conditions are monitored in real time, and the threshold is dynamically optimized based on this so that it can adapt to different external environments. Specifically, a piecewise function is set to dynamically calculate the threshold: when the network signal strength (RSRP) is higher than -80dBm, the benchmark threshold is multiplied by 1.0 to appropriately relax the requirements; when the RSRP is between -80dBm and -100dBm, the benchmark threshold is multiplied by 0.8 to slightly tighten the requirements; when the RSRP is lower than -100dBm, the benchmark threshold is multiplied by 0.6 to further increase the requirements. At the same time, we can also set a maximum number of retries. When the number of consecutive incremental learning attempts exceeds this upper limit, learning will be suspended until the network conditions improve significantly and then resumed to avoid excessive consumption of computing resources.

[0019] Further, if the Modem of the UE checks that the last TAU ​​request receives TAU accept within a preset time period, the UE is controlled to send the TAU request to the network at the first transmission power value P1.

[0020] Further, if the UE determines that the increased Tx transmit power P2 exceeds the maximum allowed transmit power Pmax of the UE, the UE is controlled to send a TAU request to the network at the first transmit power value P1.

[0021] Furthermore, TAU request is Tracking Area Update request; TAU accept is Tracking Area Update acceptance; Among them, Tracking Area Update (TAU), called Tracking Area Update in Chinese, is an important process for mobile management. Its basic principle is: the communication network divides its coverage area into multiple Tracking Areas (TA) and assigns a unique identification code (TAI) to each TA; when the UE moves in the network, if it detects that the TAI of the current resident cell has changed, it will initiate a TAU request to the network to report its new TA; after receiving the TAU request, the network will update the location information of the UE and reply with a TAU reception message to notify the UE that the update is successful. Compared with the traditional Routing Area Update (RAU), TAU allows a wider range of UE movement, reduces the frequency of location updates, and saves signaling overhead; the network can flexibly adjust the size of the TA according to the UE's mobility mode, and balance the signaling overhead and paging resource consumption; during the TAU process, the network can simultaneously complete multiple tasks such as the negotiation of security parameters and the reconfiguration of IP addresses, thereby improving process efficiency.

[0022] Furthermore, the maximum allowed transmit power Pmax of the UE is a preset constant value.

[0023] Furthermore, after the UE records the TAU failure, it also includes the step of the UE increasing the number of TAU retries. In order to further improve the TAU success rate, the present application also introduces a TAU retry mechanism based on the neural network prediction and transmit power adjustment. Whenever the UE records a TAU failure, we not only store the transmit power at the time of failure, but also increase the value of a TAU retry counter. The function of this retry counter is to: count the cumulative number of TAU failures over a period of time to reflect the severity of the current network environment; when the number of retries reaches a certain preset upper limit, trigger some special remedial measures, such as switching to other available base stations, adjusting mobility management parameters, etc.; jointly decide with the predicted value of the neural network whether to continue to increase the transmit power, or give up the power adjustment and restore the initial value. Set an initial retry limit N, which is determined based on factors such as UE equipment performance and service level; each time a TAU failure is recorded, add 1 to the current retry counter and compare it with the upper limit N: if the upper limit is not reached, continue the neural network prediction and transmit power adjustment process; if the upper limit is reached or exceeded, trigger remedial measures and reset the retry counter to 0; at the same time, set a retry interval T to control the minimum time interval between two consecutive TAU retries to avoid network congestion caused by too frequent retries; in addition, an avoidance algorithm can be designed to dynamically extend the retry interval T and slow down the retry frequency according to the increase in the number of retries. For example, you can set: the initial retry limit N = 5, that is, the remedial measures are triggered after 5 consecutive TAU failures; the initial retry interval T = 1 second, that is, at least 1 second between two TAU retries; after each TAU failure, multiply the interval T of the next retry by 2, showing an exponential backoff growth. In this way, when the UE fails in the first five TAU attempts, it will initiate a retry after 1 second, 2 seconds, 4 seconds, 8 seconds, and 16 seconds respectively. After the sixth failure, it will trigger remedial operations such as switching base stations and restart a new round of TAU procedures. Through retry control and interval backoff, combined with neural network prediction and adaptive adjustment of transmit power, this application can maximize the TAU success rate, optimize UE energy consumption and network resource utilization, and is a very flexible and efficient fault-tolerant solution.

[0024] Another aspect of the present application also provides a system for switching between NSA and LTE, which is used to execute a method for switching between NSA and LTE of the present application.

[0025] 3. Beneficial effects

[0026] Compared with the prior art, the advantages of this application are:

[0027] The introduction of a neural network model to dynamically predict the TAU timer threshold can adjust the timer timeout in real time based on factors such as the user's mobile phone network connection, 4G / 5G network speed, and device performance. Compared with the traditional fixed threshold solution, this method can more flexibly and intelligently adapt to different network environments and device states, thereby shortening the response waiting time of the TAU process, reducing the signal interruption duration during the switching process, and improving user perception.

[0028] Combined with the mechanism of gradually increasing the transmit power when TAU fails, the success rate of TAU request transmission can be effectively improved. When the TAU request at the initial transmit power fails to receive a network response within the predicted threshold time, the terminal will record the current transmit power and appropriately increase the transmit power when the next handover is triggered. This adaptive transmit power control scheme can better cope with abnormal situations such as network signal attenuation and interference, reduce the probability of handover failure due to TAU request transmission failure, and thus reduce the frequency and duration of limited service status.

[0029] By avoiding being in the limited service state for a long time, the method of the present application can minimize the display time of the no signal icon and effectively improve the user's visual perception and usage experience. In traditional solutions, once the limited service state is entered, the signal icon will immediately change to no signal, and the user will mistakenly think that the phone has a problem or the network is completely disconnected. However, the present application maintains the connection with the network as much as possible through multiple adaptive retries of TAU requests, delaying or avoiding the appearance of the no signal icon, allowing users to switch between NSA / LTE networks more seamlessly and smoothly.

[0030] The neural network prediction model of this application uses a transfer learning method for online fine-tuning, which can continuously improve the prediction accuracy of the TAU timer threshold. By comparing the predicted threshold with the actual TAU response time, the situation with large prediction deviation is identified, and some network layers are frozen accordingly, and the model is retrained using new measured data. This adaptive optimization mechanism enables the prediction model to be continuously improved over time and changes in the environment, providing a more accurate and reliable timer threshold for the switching process, thereby improving the success rate and stability of the entire switching process. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 This is a diagram of the neural network structure of this application;

[0032] Figure 2 This is a diagram of the transfer learning process for this application;

[0033] Figure 3 This is a technical route flow chart for this application. DETAILED DESCRIPTION

[0034] The present application is described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0035] The technical solution of this application is mainly applied to the following scenarios: the mobile phone used by the user supports communication technology based on the 4G core network, such as LTE or 5GNSA (non-independent networking); the user's mobile phone switches between LTE and 5GNSA; after the terminal sends a TAU request, the network does not reply to the terminal with a TAU accept within the validity period of the timer. When the switch between LTE and NSA is triggered, the UE (user equipment) sends a TAU request to the network with an initial transmission power value of P1. At the same time, the NAS (non-access layer) of the UE starts a timer called limited buffer.

[0036] Figure 1 This is the neural network structure diagram of this application, such as Figure 1 As shown, in order to set the timeout threshold of the timer more intelligently and dynamically, this application introduces a prediction model based on a neural network. Specifically, we use the Keras deep learning framework to build a fitting network containing 10 neurons. The input parameters of the network include: the network connection status of the user's mobile phone; the current 4G and 5G network speeds; the performance parameters of the mobile device, etc.

[0037] By training the neural network model, we can predict an optimal timer threshold t based on the actual environment and status of the current mobile phone. This threshold t will be used to set the timeout of the limited buffer timer. Compared with the traditional fixed threshold solution, this application uses a neural network to adaptively predict and dynamically adjust the timer threshold, which can better adapt to different network conditions and device performance, thereby improving the response speed and success rate of the TAU process. At the same time, this solution also provides an optimized starting point for subsequent TAU ​​retries and transmit power control, which helps to further improve the overall performance of NSA / LTE switching.

[0038] Considering that the actual use of the user may be different from the network prediction results, we added the transfer learning part based on the existing neural network. Specifically, we compare the timer threshold predicted by the neural network with the actual TAU response time. If the deviation between the predicted value and the actual value is large and exceeds the preset tolerance range, it means that the prediction performance of the current network model needs to be improved.

[0039] At this point, we take the following optimization measures: First, freeze some layers of the existing neural network and keep their weight parameters unchanged. Then, use the newly obtained actual TAU response time as additional training data to build a custom layer and insert it into the original network structure. Finally, we recompile and train the entire neural network to make full use of the new measured data and fine-tune and update the model parameters.

[0040] Through this transfer learning method, our neural network model can continuously perform adaptive optimization according to actual usage, gradually improving the accuracy and reliability of timer threshold prediction. At the same time, since we only partially adjust the network structure and use existing parameters as initial values, the convergence speed of the training process will also be greatly accelerated, achieving a balance between model performance and training efficiency.

[0041] Figure 2 This is the transfer learning process diagram for this application. Figure 2 It shows the structure of the neural network after adding transfer learning. Figure 3 This is the technical route flow chart of this application. Figure 3 The overall process of the technical solution of this application is described as follows: Step 1: Triggering the handover between LTE and NSA. When the user's mobile device moves at the edge of the LTE and NSA network coverage areas, the mobile phone's wireless access network and core network will decide whether to trigger the handover between LTE and NSA based on factors such as network signal strength and service quality. Once the handover conditions are met, a handover instruction will be sent to the UE to start the handover process.

[0042] Step 2: The UE sends a TAU request to the network with an initial transmit power of P1. After receiving the handover instruction, the UE first sends a TAU (Tracking Area Update) request message to the target network with a preset initial transmit power of P1. The message contains the UE's identity information, security parameters, etc., and is used to register with the new network and update the location information.

[0043] Step 3: NAS starts the limited buffer timer and sets the timeout period according to the threshold t predicted by the neural network. After sending the TAU request, the NAS (Non-Access Stratum) layer of the UE starts a timer called limited buffer. The timeout threshold t of this timer is predicted in real time by the neural network model. The model comprehensively considers factors such as the UE's network connection status, signal quality, and historical response time, and gives a dynamically optimized timer threshold.

[0044] Step 4: Determine whether the TAU accept message from the network is received within time t. After sending the TAU request, the UE enters a waiting state and listens for response messages from the network side. If the TAU accept message returned by the network is successfully received within the timer threshold t predicted by the neural network, it means that the switching process is successfully completed and the UE has successfully registered with the new network. At this time, the UE can jump out of the optimization process of this application and directly start using the services provided by the target network. On the contrary, if the UE still has not received the TAU accept message after a period of time of t, it means that the TAU request may have failed due to poor signal quality, network congestion and other reasons. At this time, the timer has timed out, and the UE needs to further execute step 5 and subsequent optimization measures.

[0045] Step 5: Modem reports limited service status to the upper layer. Once the TAU process fails, the UE's modem will send a limited service status report to the upper layer of the protocol stack (such as RRC, NAS, etc.) to inform that the current network service is limited. This will trigger the UE to take a series of remedial and optimization measures, such as transmit power control, TAU retry, etc.

[0046] Step 6: The signal bar on the UI interface displays a no signal icon, and the UE's user interface (UI) is immediately updated, with the signal strength icon at the top of the screen set to no signal. This change intuitively informs the user that there is a problem with the current mobile network connection, which may temporarily affect the normal use of communication services.

[0047] Step 7: The UE records the TAU failure and the current transmit power P1. In order to optimize the subsequent TAU ​​retry process, the UE will internally record the relevant information of this TAU failure event, especially the current transmit power P1. This power value will provide a reference for the next step of transmit power adaptive adjustment.

[0048] Step 8: After a period of time, the handover between LTE and NSA is triggered again. Due to the user's mobility and the dynamic changes in network conditions, the UE may face the need to handover between LTE and NSA networks again after a period of time. At this time, the handover triggering condition described in step 1 is met again, and the UE needs to try to perform a new round of TAU process.

[0049] Step 9: The modem checks whether the last TAU ​​request received TAU accept as expected. Before starting a new TAU process, the UE's modem will first check the result of the last TAU ​​request. If it is found that the last TAU ​​was successful, that is, the TAU accept message was received within the time t, it means that the currently used transmission power P1 is appropriate and can continue to be used. At this time, the UE directly jumps to step 2 and re-initiates the TAU request with power P1. However, if the check finds that the last TAU ​​request failed due to timeout, it means that the transmission power needs to be appropriately adjusted in order to improve the success rate of the next TAU. At this time, the UE enters step 10 and starts executing the transmission power adaptation algorithm.

[0050] Step 10: Increase the transmit power from P1 to P2. According to the transmit power P1 recorded in step 7 when the last TAU ​​failed, the UE adaptively adjusts the transmit power by increasing it by a preset step size to obtain a new transmit power value P2. The size of this step size can be flexibly set according to the actual application scenario and device capabilities. It should be large enough to quickly improve the TAU success rate, but should not be too large to cause interference or power consumption problems.

[0051] Step 11: Determine whether the current transmit power exceeds the maximum allowed value Pmax. In order to ensure that the transmit power is adjusted within a reasonable range, the UE needs to compare the new power value P2 obtained in step 10 with the preset maximum allowed transmit power Pmax. If P2 exceeds Pmax, it means that continuing to increase the power may violate the RF specifications or cause other problems. Therefore, the UE should not use P2, but restore to the initial power P1, and then re-execute step 2 to initiate a new round of TAU requests with P1. However, if P2 does not exceed Pmax, it means that this new transmit power value is acceptable, and the UE can use it to try to re-TAU in order to obtain a higher success rate. At this time, the UE enters step 12 and executes the subsequent TAU ​​process with P2.

[0052] Step 12: Send TAU request to the network with new transmit power P2. The UE resends the TAU request message to the target network using the new transmit power value P2 obtained after the judgment in step 11. This time, due to the use of a higher transmit power, the transmission reliability of the message on the air interface is expected to be improved, thereby increasing the probability that the TAU request is successfully received by the network.

[0053] Step 13: Go to step 3, restart the timer and wait for the network to reply. After sending a new TAU request, the UE re-enters the waiting state, but at this time it is necessary to start the limited buffer timer again, and use the neural network model to dynamically predict and optimize the timing threshold t. The process here is basically the same as steps 3 and 4, except that the updated transmission power and timing threshold are used. In this way, the UE will repeat the closed loop formed by steps 3 to 13 until one of the following two situations occurs: the TAU process is successfully completed, the UE receives the network's TAU accept message as expected, and successfully switches to the target network; or, the number of TAU retries reaches the preset maximum limit. To avoid an infinite loop, the UE actively terminates the process and reports that the TAU has completely failed. Through the above intelligent optimization process, the UE can adaptively adjust key parameters such as transmission power and timer threshold according to the actual network conditions, continuously try to improve the TAU success rate, minimize the switching gap time, and improve the user communication experience. From the above process, it can be seen that this application introduces multiple optimization measures such as neural network prediction, adaptive transmission power control, and transfer learning on the basis of the traditional TAU mechanism, which can significantly improve the signaling transmission success rate in NSA / LTE switching scenarios and shorten the duration of the no-service state, thereby maximizing the user's network experience. This technical solution has distinct intelligent and adaptive characteristics, and represents an important development direction for future mobile communication systems.

[0054] The invention of the present application and its implementation methods are described schematically above. The description is not restrictive. Without departing from the spirit or basic features of the present application, the present application can be implemented in other specific forms. What is shown in the accompanying drawings is only one of the implementation methods of the invention of the present application. The actual structure is not limited to this. Any figure mark in the claims should not limit the claims involved. Therefore, if a person of ordinary skill in the art is inspired by it, without departing from the purpose of the present invention, a structural method and an embodiment similar to the technical solution are designed without creativity, which should all fall within the scope of protection of this patent. In addition, the word "including" does not exclude other elements or steps, and the word "one" before the element does not exclude the inclusion of "multiple" elements. The multiple elements stated in the product claim can also be implemented by one element through software or hardware. The words first, second, etc. are used to indicate names, and do not indicate any specific order.

Claims

1. A method for switching between NSA and LTE, comprising: When triggering a handover between LTE and NSA, the user equipment UE sends a TAU request to the network with a first transmit power value P1; The NAS of the UE starts the limited buffer timer and obtains the threshold value t of the timer through neural network prediction based on the input parameters; wherein the input parameters include network information and device performance parameters; the network information includes the uplink and downlink data transmission rates of the UE in the LTE network or NSA network; the device performance parameters include the processor computing power and memory capacity of the UE; The UE monitors whether it receives a TAU accept reply from the communication network within the threshold time t; If the UE does not receive a TAU accept from the network within the threshold time t, the timer times out, the UE's modem reports the limited service status to the NSA, and the UE notifies the UI to display the signal bar as no signal. At the same time, the UE records the TAU failure and the transmit power P1 here; After a period of time greater than the threshold time t, when the UE triggers the switch between TLE and NSA again, the UE's Modem checks whether the last TAU ​​request has received TAU accept. If not, the UE's Tx transmit power is increased from P1 to P2, where P2 is greater than P1. The UE determines whether the increased Tx transmit power P2 exceeds the maximum allowed transmit power Pmax of the UE. If P2 is less than or equal to Pmax, the UE sends a TAU request to the network with the second transmit power value P2, and goes to the step of restarting the limited buffer timer of the UE's NAS.

2. The method for switching between NSA and LTE according to claim 1, characterized in that: If the UE determines that it has received a TAU accept reply from the network within the threshold time t, the NSA determines that the TAU process is successful and controls the UE to exit the present switching method.

3. The method for switching between NSA and LTE according to claim 1, characterized in that: A neural network comprises a fitted network of N neurons.

4. The method for switching between NSA and LTE according to claim 3, characterized in that: Compare the predicted value of the timer threshold t predicted by the neural network with the obtained response time value of the UE receiving TAU accept to obtain the prediction deviation; Determine whether the prediction deviation is greater than a preset deviation threshold. When the deviation is greater than the deviation threshold, freeze some layers of the neural network, where the some layers are at least one hidden layer in the neural network except the input layer and the output layer. After freezing some layers, obtain a new UE response time value for receiving TAU accept as new training data; The unfrozen neural network layers are recompiled and trained using new training data to obtain an updated neural network model.

5. The method for switching between NSA and LTE according to claim 4, characterized in that: If the modem of the UE checks that the last TAU ​​request receives a TAU accept within a preset time period, the UE is controlled to send a TAU request to the network at a first transmission power value P1.

6. The method for switching between NSA and LTE according to claim 5, characterized in that: If the UE determines that the increased Tx transmit power P2 exceeds the maximum allowed transmit power Pmax of the UE, the UE is controlled to send a TAU request to the network at the first transmit power value P1.

7. The method for switching between NSA and LTE according to claim 6, characterized in that: TAU request is Tracking Area Update request; TAU accept is accepted by Tracking Area Update.

8. The method for switching between NSA and LTE according to claim 7, characterized in that: The maximum allowed transmit power Pmax of the UE is a preset constant value.

9. The method for switching between NSA and LTE according to claim 8, characterized in that: After the UE records the TAU failure, the step of the UE increasing the number of TAU retries is also included.

10. A system for switching between NSA and LTE, characterized in that: include: At least one processing unit; used to execute instructions to implement the method for switching between NSA and LTE as described in any one of claims 1 to 9.

Images