Method of operating a terminal and terminal

Abstract

An operation method of a terminal and a terminal are provided. An operation method of a terminal configured to communicate with at least one cell of a plurality of cells includes generating first cell search training data for the plurality of cells, determining at least one training candidate cell based on the first cell search training data, updating a classification based on second cell search training data of the first cell search training data corresponding to the at least one training candidate cell and network information about the terminal to obtain an updated classification, and determining a valid cell of the plurality of cells based on the updated classification.

Description

[0001] This application claims the benefit of Korean Patent Application No. 10-2020-0035817, filed with the Korean Intellectual Property Office on March 24, 2020, the disclosure of which is incorporated herein by reference in its entirety. Technical Field

[0002] The example embodiment relates to an operating method and a terminal for improving cell detection performance. Background Technology

[0003] For network communication with the optimal or desired cell (or base station), a terminal can perform a cell search operation on multiple cells. The terminal calculates the correlation between the received signal and a reference signal using synchronization signals received from each of the multiple cells, and determines the effective cell from the multiple cells based on the correlation. However, when the above-described method for determining the effective cell is applied to a terminal, the accuracy of indicators such as correlation decreases in noisy environments with weak signal electric fields. In other words, when the correlation reference value used for determining the effective cell is set too high, the probability of incorrectly identifying a valid cell as an incorrect cell decreases, but the probability of missing a candidate cell that could be identified as a valid cell increases. On the other hand, when the correlation reference value used for determining the effective cell is set too low, the probability of incorrectly identifying a valid cell as an incorrect cell decreases, but the probability of identifying an incorrect cell as a valid cell increases. A reference design for effective cell determination that maintains improved performance regardless of the terminal's communication environment is desired. Summary of the Invention

[0004] An example embodiment provides a terminal and a method of operating the terminal, which can maintain improved cell search performance in various communication environments by determining references for effective cell design based on machine learning.

[0005] According to an example embodiment, an operating method is provided for a terminal configured to communicate with at least one of a plurality of cells. The operating method includes: generating first cell search training data for the plurality of cells; determining at least one training candidate cell based on the first cell search training data; updating a classification based on second cell search training data corresponding to the at least one training candidate cell in the first cell search training data and network information about the terminal to obtain an updated classification; and determining a valid cell among the plurality of cells based on the updated classification.

[0006] According to an example embodiment, an operating method is provided for a terminal configured to communicate with at least one of a plurality of cells. The operating method includes: generating first cell search training data for the plurality of cells; determining a plurality of training candidate cells based on the first cell search training data; classifying second cell search training data corresponding to the plurality of training candidate cells in the first cell search training data into a valid cell dataset and an erroneous cell dataset based on network information about the terminal; and updating the classification based on a linear separation between the valid cell dataset and the erroneous cell dataset to obtain an updated classification.

[0007] According to an example embodiment, a terminal configured to communicate with at least one of a plurality of cells is provided. The terminal includes: a plurality of antennas configured to receive high-frequency signals from each of the plurality of cells; a radio frequency integrated circuit configured to process the high-frequency signals into baseband signals; and a processing circuit configured to: detect a plurality of synchronization signals of the plurality of cells from the baseband signals; use the plurality of synchronization signals to generate first cell search training data; determine a training effective cell candidate group from the plurality of cells based on the first cell search training data; and update the classification based on second cell search training data corresponding to the training effective cell candidate group in the first cell search training data and network information about the terminal. Attached Figure Description

[0008] The exemplary embodiments will be more clearly understood from the following detailed description taken in conjunction with the accompanying drawings, in which:

[0009] Figure 1 This is a diagram illustrating a wireless communication system according to an example embodiment;

[0010] Figure 2 A block diagram illustrating a terminal according to an example embodiment is shown;

[0011] Figure 3 This is a block diagram illustrating a cell search controller according to an example embodiment;

[0012] Figure 4 This is a flowchart of a cell search method for a terminal according to an example embodiment;

[0013] Figure 5 It is shown in detail according to the example embodiment. Figure 4 The flowchart of operation S100 in the process;

[0014] Figure 6 It is shown in detail according to the example embodiment. Figure 4 The flowchart of operation S200 in the process;

[0015] Figure 7A and Figure 7B This is a detailed illustration based on an example embodiment. Figure 4 The flowchart of operation S300 in the process;

[0016] Figure 8A This is a graph used to explain the SVM-based classifier according to the example embodiment, and Figure 8B Based on reference Figure 8A The given description Figure 4 Detailed flowchart of the S300 operation;

[0017] Figure 9 According to the example embodiment Figure 4 Detailed flowchart of the S300 operation;

[0018] Figure 10A According to the example embodiment Figure 9 The detailed flowchart of operation S328 in the process, and Figure 10B It is used to explain in Figure 10A A graph of an example embodiment disclosed herein;

[0019] Figure 11A and 11B These are flowcharts, based on example embodiments, used to explain methods for storing effective cell datasets and erroneous cell datasets of a terminal, taking into account the limited storage capacity of the memory.

[0020] Figure 12 This is a diagram used to explain in detail the data storage method of a terminal according to an example embodiment;

[0021] Figure 13 This is a diagram used to explain the cell search method of a terminal according to an example embodiment; and

[0022] Figure 14 This is a block diagram of an electronic device according to an example embodiment. Detailed Implementation

[0023] A base station can communicate with a terminal and allocate communication network resources to the terminal. A base station may include at least one of a cell, a base station (BS), a node B (NB), an evolved Node B (eNB), a next-generation radio access network (NG RAN), a radio connectivity unit, a BS controller, and / or nodes on the network. In the following text, for ease of description, a BS may be referred to as a cell.

[0024] A terminal (or communication terminal) can communicate with a cell or another terminal. A terminal may be referred to as a node, user equipment (UE), next-generation UE (NG UE), mobile station (MS), mobile device (ME), apparatus, or terminal.

[0025] Additionally, the terminal may include at least one of the following: smartphone, tablet PC, mobile phone, video phone, e-book reader, desktop PC, laptop PC, netbook computer, personal digital assistant (PDA), portable multimedia player (PMP), MP3 player, medical device, camera, and wearable device. Furthermore, the terminal may include a television set, digital video disc (DVD) player, audio device, refrigerator, air conditioner, vacuum cleaner, oven, microwave oven, washing machine, air purifier, set-top box, home automation control panel, security control panel, media box (e.g., Samsung HomeSync). TM Apple TV™ or Google TV TM ), game consoles (e.g., Xbox) TM or PlayStation TM The terminal may include at least one of the following: electronic dictionary, electronic key, camera, and electronic photo frame. Additionally, the terminal may include at least one of the following: various medical devices (e.g., various portable medical measuring devices (blood glucose meters, heart rate monitors, blood pressure monitors, thermometers, etc.)), magnetic resonance angiography (MRA), magnetic resonance imaging (MRI), computed tomography (CT), camera or ultrasound device, navigation device, Global Navigation Satellite System (GNSS), event data recorder (EDR), flight data recorder (FDR), automotive infotainment device, marine electronic device (e.g., nautical navigation device, gyrocompass, etc.), avionics device, security device, vehicle head unit, industrial or domestic robot, drone, automated teller machine (ATM) (e.g., in financial institutions), point of sale (POS) (e.g., in stores), and / or Internet of Things (IoT) device (e.g., light bulbs, various sensors, sprinkler devices, fire alarms, thermostats, streetlights, toasters, exercise equipment, hot water tanks, heaters, boilers, etc.). Furthermore, the terminal may include at least one of the following: various types of multimedia systems capable of communication.

[0026] In the following, exemplary embodiments are described in detail with reference to the accompanying drawings.

[0027] Figure 1 This is a diagram of a wireless communication system 1 according to an example embodiment.

[0028] Reference Figure 1 The wireless communication system 1 may include a first cell 10 to a seventh cell 70 (e.g., a first cell 10, a second cell 20, a third cell 30, a fourth cell 40, a fifth cell 50, a sixth cell 60 and / or a seventh cell 70) and / or a terminal 100.

[0029] Terminal 100 can access wireless communication system 1 by transmitting signals to and receiving signals from first cells 10 to seventh cells 70. The wireless communication system 1 that terminal 100 can connect to may be referred to as a Radio Access Technology (RAT), and as a non-limiting example, may include wireless communication systems using cellular networks, such as fifth-generation (5G) communication systems, Long Term Evolution (LTE) communication systems, Evolved LTE(A) (LTE-A) communication systems, Code Division Multiple Access (CDMA) communication systems, Global System for Mobile Communications (GSM) communication systems, and / or Wireless Local Area Network (WLAN) communication systems or other arbitrary wireless communication systems.

[0030] The wireless communication network of wireless communication system 1 can support communication of multiple wireless communication devices, including terminal 100, by sharing available network resources. For example, in the wireless communication network, information can be transmitted using various multi-connection methods, such as CDMA, frequency division multiple access (FDMA), time division multiple access (TDMA), orthogonal frequency division multiple access (OFDMA), single-carrier frequency division multiple access (SC-FDMA), orthogonal frequency division multiplexing (OFDM)-FDMA, OFDM-TDMA, and OFDM-CDMA.

[0031] Cells 10 through 70 can generally be referred to as fixed stations communicating with terminal 100 and / or other cells, and can exchange data and control information through communication with terminal 100 and other cells. For example, each of cells 10 through 70 can be referred to as a base station, Node B, Evolved Node B (eNB), Next Generation Node B (gNB), sector, site, Base Transceiver System (BTS), Access Point (AP), Relay Node, Remote Radio Head (RRH), Radio Unit (RU), small cell, etc. In this specification, a cell can be generally interpreted as representing an area or function covered by a Base Station Controller (BSC) in CDMA, a Node B in Wideband CDMA (WCDMA), an eNB in ​​LTE, a sector, etc., and can cover all various coverage areas, such as megacells, macrocells, microcells, picocells, femtocells, relay nodes, RRHs, RUs, and / or small cell communication ranges.

[0032] Cells 10 through 70 can connect to terminal 100 via a wireless channel and provide various communication services. Cells 10 through 70 can serve all user services via a shared channel and can perform scheduling by collecting status information such as the buffer status, available transmit power status, and / or channel status of terminal 100. The wireless communication system 1 can support beamforming technology by using orthogonal frequency division multiplexing (OFDM) as the radio access technology. In addition, the wireless communication system 1 can support an adaptive modulation and coding (AMC) method for determining the modulation scheme and / or channel coding rate based on the channel status of terminal 100.

[0033] Cells 10 through 70 may be neighboring cells surrounding and / or adjacent to terminal 100, and may vary depending on the location of terminal 100. Terminal 100 may perform a cell search operation to connect to and communicate with any of cells 10 through 70. The cell search operation may include cell identification and cell measurement. For example, terminal 100 may detect and record cells 10 through 70 (e.g., cell identifiers (IDs) of each of cells 10 through 70) while performing cell identification, and may measure the signal power, etc., corresponding to each of cells 10 through 70 that has been identified while performing cell measurement.

[0034] The cell search operation of terminal 100 is described in detail below. Terminal 100 can generate cell search data using signals received from first cell 10 to seventh cell 70. For example, terminal 100 can receive synchronization signals from each of first cell 10 to seventh cell 70 and generate cell search data based on the detection results of the synchronization signals. Terminal 100 can determine a group of valid cell candidates from first cell 10 to seventh cell 70 based on the cell search data and a classifier. The classifier can classify cells from first cell 10 to seventh cell 70 that may be identified as valid cells as candidate cells. For example, the classifier can set a corresponding classification function, compare the output generated by applying the cell search data to the classification function with a reference value, and determine the group of valid cell candidates based on the comparison result. For example, when the output of the classification function generated by inputting cell search data corresponding to the first cell satisfies a first reference, terminal 100 can classify the first cell as a candidate cell, and when the output satisfies a second reference, the first cell can be classified as an incorrect cell. Cell search data may include the correlation between received signals from first cells 10 to seventh cells 70 and reference signals, the magnitude of the correlation, and the phase obtained from the correlation. The classifier may compare each of the outputs generated by applying cell search data including the aforementioned correlation data with the reference value and may determine a group of valid cell candidates based on the comparison results.

[0035] However, as mentioned above, since the accuracy of indices (such as correlation) varies in noisy environments with weak or strong signal electric fields, it may be desirable to design non-fixed classifiers (or classification functions and classification references) to accurately identify effective cells even in various noisy environments.

[0036] By continuously updating the classifier, the terminal 100 according to the example embodiment can maintain improved cell search performance even in varying noisy environments. As an example embodiment, the terminal 100 can generate first cell search training data for first cells 10 to seventh cells 70. The first cell search training data can update the classifier, and the terminal 100 can generate the first cell search training data periodically, separately from the cell search operation. Furthermore, the terminal 100 can generate the first cell search training data in parallel with the cell search operation (e.g., simultaneously or concurrently) to update the classifier, and the terminal 100 can also update the classifier using the cell search data generated during the cell search operation. On the other hand, in the example embodiment, the first cells 10 to seventh cells 70 may include cells that do not actually exist or erroneous cells, and the terminal 100 can generate first cell search training data for cells that do not actually exist or erroneous cells.

[0037] As an example embodiment, terminal 100 may determine a training effective cell candidate group including at least one training candidate cell based on first cell search training data. Training candidate cells may be defined as cells used to update the classifier. Terminal 100 may compare the magnitude of the correlation between the received signal and a reference signal of each cell included in the first cell search training data with a correlation reference value, and determine the training effective cell candidate group based on the comparison result. For example, the first cell search training data may include the correlation between the received signal and a reference signal from first cells 10 to seventh cells 70, the magnitude of the correlation, and / or the phase obtained from the correlation. Furthermore, the first cell search training data may also include at least one of the cell identifier (ID), signal timing, and / or received power of each of the first cells 10 to seventh cells 70.

[0038] Terminal 100 can update its classifier (e.g., update the classification function and / or classification) by selectively using second cell search training data (also referred to herein as second training cell search data, according to an example embodiment) corresponding to the training effective cell candidate group from the first cell search training data. As an example embodiment, terminal 100 can update the classifier based on the second cell search training data and network information about terminal 100. The network information about terminal 100 may include at least one of information about cells currently connected to terminal 100 and information about the current network connection state of terminal 100 (also referred to herein as current network state, according to an example embodiment). Details of the operation of updating the classifier of terminal 100 using the second cell search training data and network information will be described later. For example, terminal 100 may perform the update operation on the classifier based on various machine learning techniques.

[0039] For initialization prior to performing cell search operations, the terminal can use initial first cell search training data generated when a valid signal is applied and initial first cell search training data generated when an erroneous signal is applied, both within a previously anticipated noisy environment. The terminal can initially set up the classifier by performing initial learning using the initial first cell search training data. According to an example embodiment, the initial first cell search training data can be stored in the terminal's memory for classifier initialization.

[0040] According to the example embodiment, the terminal 100 can stably provide improved cell search performance in various noisy environments by continuously updating the classifier used to determine valid cells from the first cell 10 to the seventh cell 70 that are adjacent to each other. Furthermore, the terminal 100 can reduce excessive resource consumption by decreasing the frequency of connection operations to invalid cells.

[0041] Figure 2 This is a block diagram illustrating a terminal 100 according to an example embodiment.

[0042] Reference Figure 2 Terminal 100 may include multiple antennas 110, a radio frequency integrated circuit (RFIC) 120, a signal detector 130, a cell search controller 140, a classifier 150, a memory 160, and / or a processor 170. In an example embodiment, the multiple antennas 110 and RFIC 120 may be configured as a front-end module, and the signal detector 130, cell search controller 140, classifier 150, memory 160, and processor 170 may be configured as a back-end module. Furthermore, in Figure 2 The signal detector 130, cell search controller 140, classifier 150 and processor 170 are shown in a configuration divided by role, but are not limited thereto, and the signal detector 130, cell search controller 140, classifier 150 and / or processor 170 can be implemented with a single baseband processor.

[0043] Antenna 110 can transmit signals processed by RFIC 120 via a wireless channel and / or receive signals transmitted from the cell via a wireless channel. RFIC 120 can amplify the signals received from antenna 110 with low noise and perform down-conversion on the amplified signals to a baseband signal.

[0044] Signal detector 130 can detect synchronization signals received from cells adjacent to terminal 100. For example, the synchronization signals may include a primary synchronization signal (PSS) and a secondary synchronization signal (SSS), and signal detector 130 can sequentially detect the PSS and SSS, generating first cell search training data for neighboring cells. Signal detector 130 can detect the PSS of any neighboring cell, obtain specific timing information (e.g., 5 milliseconds) of the cell from the detected PSS, and obtain the location of the cell's SSS and the cell ID in the cell ID group. Next, signal detector 130 can detect the SSS, obtain the cell's frame timing from the detected SSS, and obtain the cell group ID to which the cell belongs and the reference signal corresponding to the cell. Signal detector 130 can output the first cell search training data, including the correlation generated by calculating the correlation between the signal received from the cell and the reference signal, the magnitude of the correlation, and / or the phase obtained from the correlation, to cell search controller 140.

[0045] Cell search controller 140 can determine a training effective cell candidate group that includes at least one training candidate cell by using first cell search training data. For example, cell search controller 140 can determine the training effective cell candidate group based on the magnitude of the correlation between the received signal and the reference signal of each cell included in the first cell search training data.

[0046] Cell search controller 140 can extract second cell search training data corresponding to the training effective cell candidate group from the first cell search training data. Cell search controller 140 can classify the second cell search training data into effective cell datasets and error cell datasets based on the network information of terminal 100. As an example embodiment, the network information of terminal 100 may include information about the cell currently connected to terminal 100 (e.g., the ID of the currently connected cell), and cell search controller 140 can classify the second training cell search data corresponding to training candidate cells with IDs matching the currently connected cell IDs as effective cell datasets, and classify the second training cell search data corresponding to training candidate cells with IDs inconsistent with the currently connected cell IDs as error cell datasets. Hereinafter, an error cell can be defined as a weak cell unsuitable for connection to terminal 100. Furthermore, error cells may include non-existent cells and may be referred to as ghost cells.

[0047] In an example embodiment, the cell search controller 140 may update the classifier 150 based on a linear separation between the valid cell dataset and the erroneous cell dataset. A linear separation may exist in the outputs when the valid cell dataset and the erroneous cell dataset are applied to the classification function (e.g., classification) corresponding to the classifier 150, and this will be described in detail later.

[0048] Because the effective cell dataset is classified based on information about currently connected cells, when the effective cell dataset is input into classifier 150 (e.g., applied to classification and / or input into a classification function), the output of classifier 150 is expected to satisfy a first reference. The first reference may be a preset or given reference used to classify at least one cell among multiple cells as a candidate cell or an effective cell during cell search. When the output of classifier 150 corresponding to the effective cell dataset does not satisfy the first reference, cell search controller 140 may appropriately update classifier 150. Furthermore, because the erroneous cell dataset is classified based on information about currently connected cells, when the erroneous cell dataset is input into classifier 150 (e.g., applied to classification and / or input into a classification function), the output of classifier 150 is expected to satisfy a second reference. The second reference may be a preset or given reference used to remove erroneous cells from multiple cells during cell search. When the output of classifier 150 corresponding to the erroneous cell dataset does not satisfy the second reference, cell search controller 140 may appropriately update classifier 150. According to the example embodiment, the first reference and / or the second reference may be design parameters determined through empirical studies.

[0049] In an example embodiment, the cell search controller 140's update operation on the classifier 150 may include adjusting at least one parameter of the classification function corresponding to the classifier 150 (e.g., adjusting the classification and / or classification criteria). As an example, the classification function may include at least one of slope and / or bias, and may correspond to a linear function. As a non-limiting example, the classification function may be a support vector machine (SVM) classification function. Furthermore, the cell search controller 140 may include a machine learning engine for continuously updating the classifier 150 in the manner described above.

[0050] In an example embodiment, the cell search controller 140 may selectively store the valid cell dataset and the erroneous cell dataset in memory 160, taking into account the storage capacity of memory 160, before inputting them into classifier 150. Memory 160 may include a first memory region and a second memory region, wherein the valid cell dataset may be stored in the first memory region and the erroneous cell dataset may be stored in the second memory region. On the other hand, because the first and second memory regions have limited storage capacity, and newly generated valid cell datasets and erroneous cell datasets periodically may not be stored, the cell search controller 140 may select data belonging to a specific valid range within the valid and erroneous cell datasets and adjacent to the classification reference of classifier 150, and store the selected data in each of the first and second memory regions. Details of this will be described later. The cell search controller 140 may perform a cell search operation using an updated classifier 150 and determine valid cells among a plurality of cells.

[0051] As an example embodiment, when cell search data is input from the cell search controller 140 while searching for cells, the classifier 150 can determine a group of valid cell candidates based on the input results. In the example embodiment, when cell search data corresponding to a group of valid cell candidates determined by the cell search controller 140 is input, the classifier 150 can remove erroneous cells included in the group of valid cell candidates based on the input results. In the example embodiment, when cell search data is input from the cell search controller 140, the classifier 150 can determine valid cells immediately or quickly based on the input results. The following description is primarily based on an example embodiment where an accurate group of valid cell candidates is determined using the classifier 150, but is not limited thereto. In the example embodiment, erroneous cells included in the group of valid cell candidates determined by the classifier 150 can be removed, or the operation of determining valid cells can be performed using cell search data. The processor 170 can control various operations for wireless communication with valid cells determined by the cell search controller 140.

[0052] Figure 3 This is a block diagram illustrating a cell search controller 140 according to an example embodiment. Figure 3 The configuration is based on the specific role of the cell search controller 140, but this is for ease of description and the example embodiment is not limited thereto. It should be clearly understood that the cell search controller 140 can be implemented in various ways (such as hardware or a combination of software and hardware).

[0053] Reference Figure 3 The cell search controller 140 may include a training data collector 142, a data preprocessor 144, a learner 146, and / or a classifier updater 148. The training data collector 142 may collect first cell search training data using signals received from a plurality of cells. In an example embodiment, the first cell search training data may include the correlation between the received signal and a reference signal for each of the plurality of cells, the magnitude of the correlation, and / or the phase obtained from the correlation, and may also include at least one of cell ID, signal timing, and / or received power. Alternatively, the magnitude of the correlation for each of the plurality of cells and / or the phase obtained from the correlation in the first cell search training data may correspond to variables applied to a classification function corresponding to a classifier described later. Details of this will be described later.

[0054] The training data collector 142 can determine a candidate group of valid training cells from the first cell search training data, and classify the second cell search training data into a valid cell dataset and an error cell dataset by using information about the cells currently connected to the terminal. The training data collector 142 can perform the classification operation for the error cell dataset only when the strength of the network electric field between the terminal and the currently connected cell is equal to or greater than the reference electric field strength. In other words, when the strength of the network electric field is less than the reference electric field strength, valid cells may be mistakenly identified as error cells, thus the reliability of the error cell dataset may decrease, and the training data collector 142 can perform the error cell dataset classification operation based on the above condition. The classification operation for the valid cell dataset can be performed regardless of the state of the network electric field. According to the example embodiment, the reference electric field may be a design parameter determined through empirical research.

[0055] Before updating the classifier using the effective cell dataset and the erroneous cell dataset, the data preprocessor 144 may perform preprocessing operations on the effective cell dataset and the erroneous cell dataset (e.g., second cell search training data) based on the terminal's current network state. For example, the terminal's current network state may include at least one of the terminal's network connection start state (e.g., the terminal's network connection state) and / or the terminal's handover state. The terminal's network connection start state may refer to the network state used to initiate a connection with a cell for the first time when the terminal is powered on. The terminal's handover state may refer to the network state used to switch the connection from one cell to another due to reasons such as the terminal's movement. The data preprocessor 144 may reflect the terminal's current network state to the classifier (e.g., update the classifier based on the terminal's current network state) by performing preprocessing on the effective cell dataset and the erroneous cell dataset. For example, the data preprocessor 144 may perform preprocessing operations by applying a specific rotation matrix to the effective cell dataset and the erroneous cell dataset. Details will be described later. On the other hand, in the example embodiment, the operations of the preprocessor 144 may be omitted. According to the example embodiments, the terms data preprocessor and preprocessing operation as used herein may have a meaning that does not reflect the timing or sequence of the preprocessing operation and / or the operation of the data preprocessor relative to other operations of terminal 100.

[0056] Learning machine 146 can input effective cell datasets and erroneous cell datasets into a classifier and perform machine learning (e.g., machine learning) on ​​a classification function corresponding to the classifier based on the output of the classifier. Learning machine 146 can learn a linear separation between the effective cell dataset and the erroneous cell dataset based on the classification function of the classifier and can derive the optimal or desired classification function capable of distinguishing between the effective cell dataset and the erroneous cell dataset. According to an example embodiment, learning machine 146 can determine the linear separation by inputting the effective cell dataset into the classifier and comparing the classifier's output with a first reference, and by inputting the erroneous cell dataset into the classifier and comparing the classifier's output with a second reference. In an example embodiment, the processing circuitry of terminal 100 can perform some operations (e.g., operations described herein by learning machine 146) via artificial intelligence and / or machine learning. As an example, the processing circuitry can implement an artificial neural network, wherein the artificial neural network is trained on a training dataset (e.g., the effective cell dataset and / or the erroneous cell dataset) via, for example, supervised, unsupervised, and / or reinforcement learning models, and wherein the processing circuitry can process feature vectors to provide output based on the training. Such artificial neural networks can utilize various artificial neural network organization and processing models, such as convolutional neural networks (CNNs), recurrent neural networks (RNNs) optionally including long short-term memory (LSTM) units and / or gated recurrent units (GRUs), stacked deep neural networks (S-DNNs), state-space dynamic neural networks (S-SDNNs), deconvolutional networks, deep belief networks (DBNs), and / or restricted Boltzmann machines (RBMs). Optionally or additionally, the processing circuitry may include other forms of artificial intelligence and / or machine learning, such as, for example, linear and / or logistic regression, statistical clustering, Bayesian classification, decision trees, dimensionality reduction (e.g., principal component analysis), and expert systems; and / or combinations thereof, including ensembles such as random forests. As used in this disclosure, the term "processing circuitry" may refer to, for example, hardware including logic circuitry; hardware / software combinations, such as a processor executing software; or combinations thereof. For example, the processing circuitry may more specifically include, but is not limited to, a central processing unit (CPU), an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field-programmable gate array (FPGA), a system-on-a-chip (SoC), a programmable logic unit, a microprocessor, an application-specific integrated circuit (ASIC), etc.

[0057] The classifier updater 148 can adjust at least one parameter of the classifier's classification function based on the learning results of the learning machine 146. The classification function can be implemented as a linear function, and the parameters of the classification function can include the slope and / or bias of the classification function.

[0058] Figure 4This is a flowchart of a cell search method for a terminal according to an example embodiment.

[0059] Reference Figure 4 Terminal 100 can generate first cell search training data for multiple cells (S100). For example, the terminal can detect synchronization signals received from multiple cells and generate first cell search training data, wherein the first cell search training data includes unique information containing cell IDs and the correlation between the received signal and a reference signal for each cell. The terminal can determine a training effective cell candidate group including at least one training candidate cell by using the first cell search training data (S200). For example, the terminal can compare the magnitude of the correlation between the received signal and the reference signal for each cell in the first cell search training data with a correlation reference value, and determine training candidate cells from multiple cells after checking the comparison results for each cell. The terminal can update the classifier based on second cell search training data corresponding to the training effective cell candidate group and the terminal's network information (S300). For example, the terminal can classify the second cell search training data into a valid cell dataset and an incorrect cell dataset based on information about the cells it is connected to in the network information. In addition, the terminal can reflect information about its current network connection state in the network information into the classifier. Because the appropriate classification function varies depending on the terminal's current network connection state, the terminal can indirectly adjust the classification function by performing preprocessing operations on the effective cell dataset and the erroneous cell dataset to be input into the classifier according to the current network connection state, so that at least one parameter of the classifier's classification function satisfies the terminal's current network connection state. The terminal can input the effective cell dataset and the erroneous cell dataset into the classifier and update the classifier based on the output results from the classifier. For example, when the effective cell dataset is applied to the classifier's classification function and the output does not satisfy a first reference, or when the erroneous cell dataset is applied to the classifier's classification function and the output does not satisfy a second reference, the terminal can update the classifier by adjusting at least one parameter of the classification function. The terminal can perform a cell search operation (S400) based on the updated classifier. For example, the terminal can generate cell search data for multiple cells during cell search and determine a group of effective cell candidates by applying the cell search data to the updated classifier. The terminal can determine the candidate cell with the maximum received power among the measured received power of each candidate cell in the group of effective cell candidates as the effective cell. For example, the received power may correspond to at least one of the following: reference signal received power (RSRP) associated with a cell-specific reference signal, synchronization received power (SCH_RP) associated with a synchronization signal, reference signal received quality (RSRQ), signal-to-interference and noise ratio (SINR), received signal strength indicator (RSSI), etc.

[0060] In an example embodiment of operation S400, the terminal may generate cell search data for the plurality of cells during the cell search operation and determine a valid cell candidate group based on the cell search data. The terminal may remove erroneous cells included in the valid cell candidate group by applying the cell search data corresponding to the valid cell candidate group to an updated classifier. In an example embodiment of operation S400, when the cell search data corresponding to the valid cell candidate group determined based on the cell search data is applied to an updated classifier, the terminal may immediately or rapidly determine valid cells based on the output of the classifier. According to the example embodiment, after operation S400, the terminal 100 may connect to valid cells included in the valid cell candidate group and perform communication with the valid cells (e.g., sending and / or receiving data).

[0061] Figure 5 According to the example embodiment Figure 4 The detailed flowchart of operation S100 in the process.

[0062] refer to Figure 5 The terminal can receive synchronization signals from multiple cells and detect the PSS in the synchronization signal for each cell (S110). The terminal can detect SSS for each cell by using the PSS detection result for each cell (S120). The terminal can generate first cell search training data by using the PSS detection result and SSS detection result for each cell (S130). For example, the first cell search training data can have a data format suitable for application to a classifier, and the first cell search training data can include unique information of multiple cells.

[0063] Figure 6 According to the example embodiment Figure 4 The detailed flowchart of operation S200 in the process.

[0064] Reference Figure 6 ,exist Figure 4 After operation S100, it can be determined whether the magnitude of the correlation between the received signal and the reference signal corresponding to the nth cell (where n is an integer of 1 or greater) among multiple cells is greater than the correlation reference value (S210). According to the example embodiment, during the execution Figure 6Before the operations described, 'n' is initialized to have the value '1'. When the result of operation S210 is "yes", the terminal can classify the nth cell as a training candidate cell (S220) and proceed to operation S230. Otherwise, when the result of operation S210 is "no", it can be determined whether "n" is equal to "m" corresponding to the total number of cells adjacent to the terminal (S230). When the result of operation S230 is "no", the terminal can count up (e.g., increment) "n" and perform (e.g., repeat) operation S210. Otherwise, when the result of operation S230 is "yes", the terminal can determine a group of training effective cell candidates that includes at least one previously classified training candidate cell (S250).

[0065] Figure 7A and Figure 7B According to the example embodiment Figure 4 The detailed flowchart of the operation S300.

[0066] Reference Figure 7A The terminal can classify the second training cell search data (e.g., the first part of the second training cell search data) containing training candidate cells with IDs that match the IDs of cells connected to the current terminal as a valid cell dataset (S310). The terminal can also classify the second training cell search data (e.g., the second part of the second training cell search data) containing training candidate cells with IDs that do not match the IDs of cells connected to the current terminal as an incorrect cell dataset (S320). The terminal can update the classifier based on the valid cell dataset and the incorrect cell dataset (S330).

[0067] refer to Figure 7B ,exist Figure 7A After operation S310, the terminal can determine whether the current network electric field strength is greater than the reference electric field strength (S315). When the result of operation S315 is "yes", the following can be followed. Figure 7A Operation S320 is performed within the given sequence. Otherwise, if the result of operation S315 is "no", it can be omitted. Figure 7A Operation S320 can be performed immediately or sequentially following operation S330. In other words, when the result of operation S315 is "no", the terminal can perform an update operation on the classifier using only the valid cell dataset other than (for example, no) erroneous cell dataset.

[0068] Figure 8A This is a graph used to explain the SVM-based classifier according to the example embodiment, and Figure 8B Based on reference Figure 8A The given description Figure 4A detailed flowchart of operation S300 is provided below. The classifier is described primarily based on an example embodiment implemented using SVM, but this is merely an example and is not limited thereto. Clearly, various classification techniques capable of separating valid cells corresponding to a valid cell dataset from erroneous cells corresponding to an erroneous cell dataset can be applied to the classifier. According to the example embodiment, the term "classifier" as used herein refers to processing circuitry configured to adjust (e.g., update) and / or manage a classification function (e.g., classification and / or classification criteria) and / or compute a classification using the classification function. According to the example embodiment, inputting data into the classifier may refer to using the input data as parameters in the classification function to compute a solution and / or classification. According to the example embodiment, the data output by the classifier may refer to the solution of the classification function and / or the classification computed using the classification function.

[0069] The classifier according to the example embodiment may be based on an SVM. An SVM can be a machine learning engine that provides a classifier capable of classifying a training dataset into any two groups. Assuming the existence of a classifier capable of linearly separating between the two groups, the margin between the two groups (e.g., between the effective cell dataset and the false cell dataset) can be increased by providing a solution (e.g., for an optimization problem). The classification function corresponding to the SVM-based classifier (or SVM classifier) ​​can be defined by Equation 1. According to the example embodiment, the classifier can determine and / or adjust the parameters of the classification function (e.g., the slope and / or bias of the classification function) to provide a maximum and / or determined linear separation (e.g., margin) between the effective cell dataset and the false cell dataset (e.g., the classifier can be updated based on the linear separation).

[0070] [Formula 1]

[0071] f(x i ) = w T ·x i +b

[0072] In Formula 1, w T It can be a symmetric matrix parameter related to the margin, serving as the slope of the classification function, and x i This can be used as a parameter input to the classification function, corresponding to any of the first cell search training data, the effective cell dataset, and / or the error cell dataset. b can be the bias of the classification function. i It can be defined by Formula 2.

[0073] [Formula 2]

[0074] x i =(θ i , ρ i ) T

[0075] ρ i θ can represent the magnitude of the correlation between the received signal and the reference signal corresponding to the i-th cell (where i is an integer of 1 or greater), and θ i The phase can be determined by the real and imaginary parts of the correlation with the i-th cell. It can be calculated by cell pair x. i The data is classified and input into a classification function. The terminal can input both valid and erroneous cell datasets into the classification function, and can adjust at least one of the slope and / or bias of the classification function by comparing the output from the classification function with a specific reference.

[0076] As in Figure 8A As shown, the effective cells corresponding to the effective cell dataset and the erroneous cells corresponding to the erroneous cell dataset can be linearly separated from each other, and at least one of the slope and bias corresponding to the SVM-based classifier can be adjusted such that the effective cells and erroneous cells can be classified at the optimal or desired interval.

[0077] In the example embodiment, the terminal can input the effective cell dataset into the classification function and determine whether the conditions of Formula 3 are met.

[0078] [Formula 3]

[0079] f(x k )>TH

[0080] When the effective cell dataset contains the k-th data x corresponding to the k-th training candidate cell (where k is an integer of 1 or greater), k When the input is given to the classification function and the output is greater than a specific reference value TH, the terminal can determine that the classification function is correct. On the other hand, when the output is equal to or less than TH, the terminal can determine that the classification function is incorrect and can adjust at least one of the slope and bias of the classification function. For example, the terminal can adjust the slope and / or bias of the classification function to be lower than before.

[0081] In the example embodiment, the terminal can input the erroneous cell dataset into the classification function and determine whether the conditions of Formula 4 are met.

[0082] [Formula 4]

[0083] f(x j ) <TH

[0084] When the j-th data x in the effective cell dataset corresponds to the j-th training candidate cell (where j is an integer of 1 or greater), jWhen the input is given to the classification function and the output is less than a specific reference value TH, the terminal can determine that the classification function is correct. On the other hand, when the output is equal to or greater than TH, the terminal can determine that the classification function is incorrect and adjust at least one of the slope and / or bias of the classification function. For example, the terminal can adjust the slope and / or bias of the classification function to be higher than before.

[0085] Further reference Figure 8B The terminal can input the k-th data corresponding to the k-th training candidate cell from the effective cell dataset into the classifier (S321a). According to the example embodiment, during execution... Figure 8B Before the operations described, "k" and "j" are initialized to have a value of "1". The terminal can determine whether the output of the classifier in response to the k-th data conforms to (e.g., satisfies) the first reference (S322a). When the result of operation S322a is "no", the terminal can adjust at least one parameter in the classification function (S323a) and proceed to operation S324a. Otherwise, when the result of operation S322a is "yes", the terminal can determine whether "p1", which is the total number of training candidate cells corresponding to the effective cell dataset, is the same as "k" (S324a). When the result of operation S324a is "no", the terminal can count up (e.g., increment) "k" (S325a) and then immediately perform (e.g., repeat) operation S321a. Otherwise, when the result of operation S324a is "yes", the terminal can input the j-th data corresponding to the j-th training candidate cell in the erroneous cell dataset into the classifier (S321b). The terminal can determine whether the classifier's output in response to the j-th data conforms to (e.g., satisfies) the second reference (S322b). When the result of operation S322b is "no", the terminal can adjust at least one parameter in the classification function (S323b) and proceed to operation S324b. Otherwise, when the result of operation S322b is "yes", the terminal can determine whether "p2", which is the total number of training candidate cells corresponding to the erroneous cell dataset, is the same as "j" (S324b). When the result of operation S324b is "no", the terminal can count up (e.g., increment) "j" (S325b) and then immediately perform (e.g., repeat) operation S321b. Otherwise, when the result of operation S324b is "yes", the terminal can complete the classifier update operation (S326).

[0086] Figure 9 According to the example embodiment Figure 4 The detailed flowchart of the operation S300.

[0087] Reference Figure 9The terminal can obtain the current network connection state (S327). For example, the terminal can identify whether the current network connection state is a network connection start state or a switching state. The terminal can reflect the current network connection state to the classifier (S328). For example, the terminal can reflect the current network connection state to the classifier by performing preprocessing operations on the valid cell dataset and the erroneous cell dataset according to the current network connection state.

[0088] Figure 10A According to the example embodiment Figure 9 The detailed flowchart of operation S328 in the process, and Figure 10B It is used to explain in Figure 10A A graph of an example embodiment disclosed herein.

[0089] refer to Figure 10A ,exist Figure 9 After operation S327, the terminal can determine whether the current network connection state is a network connection start state (S328a). When the result of operation S328a is "yes", the terminal can perform preprocessing operations on the valid cell dataset and the erroneous cell dataset to indirectly increase the slope of the classification function (S328b). Otherwise, when the result of operation S328a is "no", the terminal can determine whether the current network connection state is a handover state (S328c). When the result of operation S328c is "yes", the terminal can perform preprocessing operations on the valid cell dataset and the erroneous cell dataset to indirectly decrease the slope of the classification function (S328d). Otherwise, when the result of operation S328c is "no", the terminal can omit the preprocessing operation (S328e). The following can proceed... Figure 8B Operation S321a in the middle.

[0090] Further reference Figure 10B The terminal can indirectly adjust the slope of the classification function corresponding to the SVM classifier based on the current network connection status. For example, when the terminal is in the network connection start state (Case 1), it can perform a first preprocessing operation on the valid cell dataset and the erroneous cell dataset, thereby indirectly increasing the slope of the classification function corresponding to the SVM classifier. Conversely, when the terminal is in the handover state (Case 2), it can perform a second preprocessing operation on the valid cell dataset and the erroneous cell dataset, thereby indirectly decreasing the slope of the classification function corresponding to the SVM classifier.

[0091] Although an example embodiment has been described in which the terminal indirectly adjusts at least one parameter of the classifier's classification function by preprocessing a dataset of valid cells and a dataset of erroneous cells based on the current network connectivity state, this is merely an example embodiment and is not limited thereto. In the example embodiment, the terminal may directly adjust at least one parameter of the classification function based on the current network connectivity state.

[0092] Figure 11A and 11B These are flowcharts illustrating methods for storing valid cell datasets and erroneous cell datasets of a terminal, taking into account the limited storage capacity of the memory, according to an example embodiment.

[0093] Reference Figure 11A The terminal can set the valid range of the valid cell dataset (S500a). For example, as referred to above... Figure 8A As described, when the effective cell dataset includes the magnitude of the correlation between the received signal and the reference signal, and the phase determined from the real and imaginary parts of the correlation, the effective range can be set based on the phase. The terminal can divide the effective range into z regions (e.g., zones) (where z is an integer of 1 or greater) (S510a). The terminal can extract data corresponding to each region from the effective cell dataset (S520a). According to an example embodiment, the terminal can extract data including data detected in each region from the effective cell dataset that satisfies the reference value. The terminal can selectively store the extracted data in a first memory region divided for each region using a first method (S530a).

[0094] Reference Figure 11B The terminal can set the valid range of the erroneous cell dataset (S500b). The terminal can divide the valid range into z regions (S510b). However, when using... Figure 11A When processing the results of operations S500a and S510a, operations S500b and S510b can be omitted. The terminal can extract data corresponding to each region from the erroneous cell dataset (S520b). According to the example embodiment, the terminal can extract data including data detected in each region from the erroneous cell dataset that meets the reference value. The terminal can selectively store the extracted data in a second memory region divided for each region using a second method (S530b).

[0095] Available Figure 7A After operation S320, operations S500a to S530a and operations S500b to S530b are executed, and the terminal can perform operations using valid cell datasets and erroneous cell datasets stored in the first memory region and the second memory region, respectively. Figure 7A Operation S330 in the middle.

[0096] Further reference Figure 12 When the terminal stores the effective cell dataset VCDS in a first memory region MA1, the first memory region MA1 can be divided into z regions or from the first region R1 to the z-th region Rz, and the effective cell dataset VCDS can be stored as data cell data (that is, C_11a to C_1ta, C_21a to C_2ta, ..., C_z1a to C_zta), where each of the data cell data corresponds to each region (e.g., area). For example, each of the first region (e.g., area) R1 to the z-th region (e.g., area) Rz can have a limited space to store only "t" data (e.g., "t" bits, bytes, etc.) (where t is an integer of 2 or greater), by applying a specific priority to store data with high priority first, and deleting data with relatively low priority without saving it. For example, when the data in the effective cell dataset VCDS is input into a classification function, the terminal can estimate the output result as the output and... Figure 8A The proximity of the reference values ​​TH in Formula 3 is used to determine the storage priority of the data. For example, the terminal can set the storage priority according to the order of the small correlation size included in the data in each of the first zone to the z-th zone (that is, the first zone R1 to the z-th zone Rz), and store the effective cell dataset VCDS in the first memory area MA1.

[0097] The terminal can refer to the first index Index_11 to the i-th index Index_1t of the first memory region MA1 to access the valid cell dataset VCDS stored in the first memory region MA1, and use the accessed valid cell dataset VCDS to update the classifier in machine learning.

[0098] When the terminal stores the erroneous cell dataset FCDS in the second memory region MA2, the second memory region MA2 can be divided into z regions (also referred to here as areas) (that is, the first area R1 to the z-th area Rz), and each erroneous cell dataset FCDS can be stored as data cell data (that is, C_11b to C_1tb, C_21b to C_2tb, ..., C_z1b to C_ztb), where each of the data cell data corresponds to each area. For example, each of the first area R1 to the z-th area Rz can have a limited space to store only "t" data (e.g., "t" bits, bytes, etc.) (where t is an integer of 2 or greater), by applying a specific priority to store data with high priority first, and deleting data with relatively low priority without saving it. For example, when the data in the erroneous cell dataset FCDS is input into a classification function, the terminal can estimate the output result as the output and the... Figure 8AThe proximity of the reference values ​​TH in Formula 4 is used to determine the storage priority of the data. For example, the terminal can set the storage priority according to the order of the largest correlation size included in the data of each of the z zones (that is, the first zone R1 to the zth zone Rz), and store the effective cell dataset FCDS in the second memory area MA2.

[0099] The terminal can access the erroneous cell dataset FCDS stored in the second memory region MA2 by referring to the first index Index_21 to the t-th index Index_2t of the second memory region MA2, and use the accessed erroneous cell dataset FCDS to update the classifier in machine learning.

[0100] Figure 13 This is a diagram of a cell search method for a terminal according to an example embodiment.

[0101] Reference Figure 13 The terminal can generate cell search data for cell search (S600). For example, the terminal can receive synchronization signals from multiple cells and generate cell search data corresponding to each of the multiple cells. The terminal can refer to the application. Figures 1 to 12 The described example embodiment determines a valid cell candidate group based on an updated classifier (S610). As an example, the terminal can apply cell search data to the updated classifier and determine the cell candidate group by satisfying the criteria... Figure 8A Those cells that meet the conditions of Formula 3 are classified as candidate cells, and will satisfy the conditions of Formula 3. Figure 8A Cells classified as erroneous cells according to the conditions of Formula 4 can be identified as valid cell candidate groups. The terminal can determine a valid cell based on the received power corresponding to each candidate cell from the valid cell candidate group (S620).

[0102] In the example embodiment, the terminal can perform cell search operations including operations S600 to S620 in parallel with the classifier update operation, and can use the cell search data generated in operation S600 as the first cell search training data to update the classifier.

[0103] Figure 14 This is a block diagram of an electronic device 1000 according to an example embodiment.

[0104] Reference Figure 14 The electronic device 1000 may include a memory 1010, a processor unit 1020, an input / output controller 1040, a display unit 1050, an input device 1060, and / or a communication processing unit 1090. The memory 1010 may exist in multiple forms. The components are as follows.

[0105] The memory 1010 may include a program storage unit 1011 for storing programs for controlling the operation of the electronic device 1000 and a data storage unit 1012 for storing data generated during program execution. The data storage unit 1012 may store data for the operation of the application program 1013 and the effective cell differentiation program 1014. The program storage unit 1011 may include the application program 1013 and the effective cell differentiation program 1014. Here, the program included in the program storage unit 1011 may be an instruction set and may be represented as an instruction set.

[0106] Application 1013 may include an application operable in electronic device 1000. In other words, application 1013 may include application instructions executed by processor 1022. Effective cell classifier 1014 may control classifier update operations according to exemplary embodiments of this disclosure. In other words, electronic device 1000 may generate cell search training data using effective cell classifier 1014, and perform machine learning on a classifier capable of distinguishing optimal or desired effective cells even in various noisy environments using effective cell classifier 1014.

[0107] The peripheral device interface 1023 controls the connection of the base station's input / output peripheral devices to the processor 1022 and memory interface 1021. The processor 1022 can control the base station to provide applicable services using at least one software program. In this case, the processor 1022 can execute at least one program stored in the memory 1010 to provide services corresponding to the applicable program.

[0108] The input / output controller 1040 provides an interface between input / output devices (such as display unit 1050 and input device 1060) and peripheral device interface 1023. Display unit 1050 can display status information, input characters, moving images, still images, etc. For example, display unit 1050 can display information about an application program executed by processor 1022.

[0109] Input device 1060 can provide input data generated by selection by electronic device 1000 to processor unit 1020 via input / output controller 1040. In this case, input device 1060 may include a keyboard with at least one hardware button and a touchpad for sensing touch information. For example, input device 1060 can provide touch information (such as touch, touch movement, and touch release) already sensed by touchpad to processor 1022 via input / output controller 1040. Electronic device 1000 may include communication processing unit 1090 that performs communication functions for voice communication and data communication.

[0110] Conventional apparatuses and methods for communicating with multiple cells determine valid cells based on fixed correlation reference values. However, these fixed correlation reference values ​​are unreliable and inaccurate in various noisy environments (e.g., environments with enhanced noise from weak signal electric fields). Due to the lack of reliability and accuracy in these conventional apparatuses and methods, the number of attempts to connect to invalid cells increases, leading to excessive resource consumption (e.g., power, processor, memory, latency, etc.).

[0111] However, example embodiments provide an improved terminal and method for communicating with multiple cells using an updatable classifier to determine valid cells. The updatable classifier provides improved performance in various noisy environments, including enhanced noise environments with weak signal electric fields. Therefore, the improved terminal and method overcome the shortcomings of existing devices and methods, improving reliability and accuracy, and thus reducing excessive resource consumption (e.g., power, processor, memory, latency, etc.) compared to conventional devices and methods.

[0112] According to the example embodiment, the operations described herein, performed by the wireless communication system 1, the first cell 10 to the seventh cell 70, the terminal 100, the RFIC 120, the signal detector 130, the cell search controller 140, the classifier 150, the processor 170, the training data collector 142, the data preprocessor 144, the learning machine 146, the classifier updater 148, the electronic device 1000, the processor unit 1020, the input / output controller 1040, the communication processing unit 1090, the processor 1022, the peripheral device interface 1023 and / or the memory interface 1021, can be performed by processing circuitry.

[0113] According to the example embodiment, one or both of the classifier 150 and / or learner 146 can be implemented using processing circuitry separate from the processing circuitry used to implement terminal 100, RFIC 120, signal detector 130, cell search controller 140, processor 170, training data collector 142, data preprocessor 144, classifier updater 148, electronic device 1000, processor unit 1020, input / output controller 1040, communication processing unit 1090, processor 1022, peripheral device interface 1023, and / or memory interface 1021. For example, classifier 150 can be implemented using classifier processing circuitry, and / or learner 145 can be implemented using learner processing circuitry. According to the example embodiment, all of the terminal 100, RFIC 120, signal detector 130, cell search controller 140, classifier 150, processor 170, training data collector 142, data preprocessor 144, learning machine 146, classifier updater 148, electronic device 1000, processor unit 1020, input / output controller 1040, communication processing unit 1090, processor 1022, peripheral device interface 1023, and memory interface 1021 can be implemented by the same processing circuit (e.g., the processing circuit inside the terminal 100).

[0114] The various operations of the above-described methods can be performed by any suitable device capable of performing the operations (such as the processing circuits discussed above). For example, as mentioned above, the operations of the above-described methods can be performed by various hardware and / or software implemented in some form of hardware (e.g., processors, ASICs, etc.).

[0115] The software may include an ordered list of executable instructions for implementing logical functions and may be embodied in any processor-readable medium for use by or in conjunction with an instruction execution system, device, or apparatus (such as a single-core or multi-core processor or a system containing a processor).

[0116] The methods or algorithms and functionalities described in conjunction with the exemplary embodiments disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or a combination of both. If implemented in software, the functionality may be stored as one or more instructions or code on or transmitted through a tangible, non-transitory computer-readable medium. The software module may reside in random access memory (RAM), flash memory, read-only memory (ROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), registers, hard disks, removable disks, CD-ROMs, or any other form of storage medium known in the art.

[0117] The exemplary embodiments can be described with reference to the actions and symbolic representations of the operations implemented by the units and / or devices discussed in more detail below (e.g., in the form of flowcharts, flow diagrams, data flow diagrams, block diagrams, etc.). Although discussed in a particular manner, the functions or operations specified in a particular box may be performed differently from the processes specified in the flowcharts, flow diagrams, etc. For example, functions or operations shown as being performed serially in two consecutive boxes may actually be performed concurrently, simultaneously, synchronously, or in some cases, in the reverse order.

[0118] While exemplary embodiments have been specifically shown and described with reference to examples of exemplary embodiments, it should be understood that various changes in form and detail may be made therein without departing from the spirit and scope of the appended claims.

Claims

1. A method of operating a terminal configured to communicate with at least one of a plurality of cells, the method comprising: Generate first cell search training data for the plurality of cells; At least one training candidate cell is determined based on the training data searched in the first cell. The classifier is updated based on the second cell search training data corresponding to the at least one training candidate cell in the first cell search training data and network information about the terminal to obtain an updated classifier; as well as The updated classifier is used to determine the effective cells among the plurality of cells. The network information regarding the terminal includes the terminal's current network connection status, and The step of updating the classifier includes updating the classifier based on the current network connection status of the terminal.

2. The method according to claim 1, wherein, The training data for the first cell search includes: The correlation between the received signal and the reference signal in each of the plurality of cells. The magnitude of the correlation, and The phase obtained from the correlation.

3. The method according to claim 1, wherein, The steps for determining the at least one training candidate cell include: To obtain a determination result, the magnitude of the correlation between the received signal and a reference signal in each of the plurality of cells is determined to meet a reference value, wherein the correlation between the received signal and the reference signal in each of the plurality of cells is included in the first cell search training data; and Based on the determination result, at least one training candidate cell is determined from the plurality of cells.

4. The method according to claim 1, wherein, The network information about the terminal also includes information about the cells currently connected to the terminal among the plurality of cells, and The step of updating the classifier also includes: Based on information about the cells currently connected to the terminal, the second cell search training data is classified into valid cell datasets and erroneous cell datasets. The effective cell dataset and the erroneous cell dataset are applied to the classifier to obtain the application results. The classifier is updated based on the application results.

5. The method according to claim 4, wherein, The at least one training candidate cell includes multiple training candidate cells; as well as The steps to classify the second cell search training data into valid cell datasets and erroneous cell datasets include: The first portion of the second cell search training data is classified into the effective cell dataset, wherein the first portion of the second cell search training data corresponds to a first training candidate cell among the plurality of training candidate cells that has an ID matching the ID of the cell currently connected to the terminal, or The second part of the second cell search training data is classified into the error cell dataset, wherein the second part of the second cell search training data corresponds to a second training candidate cell among the plurality of training candidate cells that has an ID that does not match the ID of the cell currently connected to the terminal.

6. The method according to claim 4, wherein, The steps of applying the effective cell dataset and the erroneous cell dataset to the classifier include: The first output is generated by inputting the effective cell dataset into a classification function corresponding to the classifier; and A second output is generated by inputting the erroneous cell dataset into the classification function.

7. The method according to claim 6, wherein, The steps for updating the classifier based on the application results include: Adjust at least one parameter of the classification function based on whether the first output satisfies a first reference; and The classification function is adjusted based on whether the second output satisfies the second reference.

8. The method according to claim 7, wherein, The at least one parameter of the classification function includes at least one of the slope of the classification function and the bias of the classification function.

9. The method according to claim 4, wherein, The step of classifying the second cell search training data into a valid cell dataset and an incorrect cell dataset includes: classifying the second cell search training data into the incorrect cell dataset based on the fact that the strength of the network electric field between the terminal and the cell currently connected to the terminal is equal to or greater than the reference electric field strength.

10. The method of claim 4, further comprising: Based on the proximity to the classification reference of the classifier, the data in the effective cell dataset is stored in the first memory area of ​​the terminal; as well as Based on the proximity to the classification reference, the data in the erroneous cell dataset is stored in the second memory area of ​​the terminal.

11. The method according to claim 1, wherein, The steps for determining the valid cells among the plurality of cells include: Generate cell search data for the multiple cells; Based on the cell search data and the updated classifier, a valid cell candidate group is determined from the plurality of cells; and The effective cells are determined from the group of effective cell candidates.

12. The method according to claim 11, wherein, The cell search data is used as updated first cell search training data to update the classifier.

13. The method according to claim 11, wherein, The step of generating the first cell search training data is performed periodically, separately from the step of generating the cell search data.

14. The method according to claim 1, further comprising: The classifier is initialized based on the initial cell search training data stored in the terminal's memory.

15. A method of operating a terminal configured to communicate with at least one of a plurality of cells, the method comprising: Generate first cell search training data for the plurality of cells; Multiple training candidate cells are determined based on the training data from the first cell search. Based on network information about the terminal, the second cell search training data corresponding to the plurality of training candidate cells in the first cell search training data are classified into effective cell datasets and erroneous cell datasets. as well as The classifier is updated based on the linear separation between the effective cell dataset and the erroneous cell dataset to obtain an updated classifier. The network information regarding the terminal includes information about the cells currently connected to the terminal among the plurality of cells, as well as the terminal's current network connection status. The method further includes updating the classifier based on the current network connection status of the terminal.

16. The method according to claim 15, wherein, The steps to classify the second cell search training data into valid cell datasets and erroneous cell datasets include: The first portion of the second cell search training data is classified into the effective cell dataset, wherein the first portion of the second cell search training data corresponds to a first training candidate cell among the plurality of training candidate cells that has an ID matching the ID of the cell currently connected to the terminal, and The second part of the second cell search training data is classified into the error cell dataset, wherein the second part of the second cell search training data corresponds to a second training candidate cell among the plurality of training candidate cells that has an ID that does not match the ID of the cell currently connected to the terminal.

17. The method according to claim 15, wherein, The steps for updating the classifier include: The first output is generated by inputting the effective cell dataset into a classification function corresponding to the classifier. A second output is generated by inputting the erroneous cell dataset into the classification function; and The linear separation is determined by comparing each of the first and second outputs with the corresponding reference output.

18. The method according to claim 17, wherein, The step of updating the classifier also includes: The slope of the classification function or the bias of the classification function is adjusted based on the linear separation.

19. A terminal configured to communicate with at least one of a plurality of cells, wherein, The terminal includes: Multiple antennas are configured to receive high-frequency signals from each of the multiple cells; Radio frequency integrated circuit, configured to process the high-frequency signal into a baseband signal; and The processing circuit is configured as follows: Detect multiple synchronization signals from the multiple cells in the baseband signal. The multiple synchronization signals are used to generate training data for the first cell search. Based on the training data from the first cell search, a candidate group of effective training cells is determined from the plurality of cells, and The classifier is updated based on the second cell search training data corresponding to the effective candidate cell group in the first cell search training data and network information about the terminal. The network information regarding the terminal includes the terminal's current network connection status. The processing circuit is configured to update the classifier based on the current network connection status of the terminal.

Images