Method, apparatus, and computer program product for processing of traffic routing

Abstract

The application provides a processing method and device of service routing, electronic equipment and computer program product, and belongs to the field of mobile communication. The method comprises the following steps: collecting end-to-end network topology information of a mobile communication network, generating a first matrix and a second matrix of candidate service routes; obtaining subscription information and network performance data of a remote operation device, calculating network quality scores of each candidate service route according to the subscription information and network performance data, and generating a third matrix; obtaining a fusion matrix based on the second matrix and the third matrix; performing network configuration issuing based on the fusion matrix according to each candidate service route, performing standard action testing under the configured network, calculating force feedback deviation, obtaining action deviation scores of the candidate service routes, and calculating comprehensive scores of the candidate service routes according to the action deviation scores and corresponding network quality scores; and selecting a candidate service route with the highest comprehensive score as an optimal service route.

Description

Technical Field

[0001] This application relates to the field of mobile communications, and more particularly to a service routing processing method, apparatus, electronic device, and computer program product. Background Technology

[0002] Currently, master-slave teleoperation surgical robots, leveraging 5G technology with its core advantages of low latency, high bandwidth, and high reliability, make remote surgery feasible. Force feedback technology can sense the minute forces applied to tissues during surgery and transmit the tactile sensation to the surgeon, effectively reducing surgical trauma. Existing methods include force feedback-based surgical cutting training for surgical robots and master-slave force feedback-based bidirectional teleoperation control of robots. However, most of these methods rely on wired networks or short-range wireless LANs for force feedback data transmission, failing to consider the impact of packet loss, latency, and jitter generated by various service routing nodes in the 5G network during data packet forwarding in long-distance transmission scenarios. They also fail to address the problem of force feedback perception deviation caused by the accumulation of network node errors. Summary of the Invention

[0003] This application provides a service routing processing method, apparatus, electronic device, and computer program product, which can solve the problems of packet loss, latency, jitter impact feedback effect of multiple network nodes in long-distance transmission environments.

[0004] In a first aspect, embodiments of this application provide a service routing processing method, which includes the following steps: Collect end-to-end network topology information of the mobile communication network, and generate a first matrix and a second matrix of candidate service routes. The first matrix includes device identifiers, and the second matrix includes network addresses. Obtain the subscription information and network performance data of the remote operation device, calculate the network quality score of each candidate service route based on the subscription information and network performance data, and generate a third matrix; Based on the second and third matrices, the fusion matrix is ​​obtained; For each candidate service route, network configuration is issued based on the fusion matrix. Standard action tests are performed under the configured network, and force feedback deviation is calculated to obtain the action deviation score of the candidate service route. The comprehensive score is calculated based on the action deviation score of the candidate service route and the corresponding network quality score. The candidate service route with the highest comprehensive score is selected as the optimal service route.

[0005] Secondly, embodiments of this application provide a service routing processing apparatus, which includes the following: The information collection module is used to collect end-to-end network topology information of the mobile communication network and generate a first matrix and a second matrix of candidate service routes. The first matrix includes device identifiers and the second matrix includes network addresses. The quality scoring module is used to obtain the subscription information and network performance data of the remote operation device, calculate the network quality score of each candidate service route based on the subscription information and network performance data, and generate a third matrix. The fusion processing module is used to obtain a fusion matrix based on the second matrix and the third matrix; and, The routing module is used to execute network configuration distribution based on the fusion matrix for each candidate service route, perform standard action tests and calculate force feedback deviation under the configured network to obtain the action deviation score of the candidate service route, calculate its comprehensive score based on the action deviation score of the candidate service route and the corresponding network quality score, and select the candidate service route with the highest comprehensive score as the optimal service route.

[0006] Thirdly, embodiments of this application provide an electronic device including a processor, a memory, and a program or instructions stored in the memory and executable on the processor, wherein the program or instructions, when executed by the processor, implement the steps of the service routing processing method as described in the first aspect.

[0007] Fourthly, embodiments of this application provide a computer program product, the computer program product including a computer program stored on a non-transitory computer-readable storage medium, the computer program including program instructions, the program instructions being executed by a computer to implement the steps of the service routing processing method as described in the first aspect.

[0008] This application embodiment collects end-to-end network topology information, generates a routing matrix, and analyzes the status of each node by combining network performance data. This allows for the selection of the optimal route, avoiding packet loss, latency, jitter, and error accumulation in multi-node transmission, thus ensuring the stability of force feedback in long-distance mobile communication networks. By combining network quality scores with action deviation scores, dedicated high-quality routes are allocated to highly sensitive force feedback data, reducing signal transmission latency. Action deviation calculations correct execution deviations, ensuring consistency in force feedback perception between the master and slave ends. Maximizing the comprehensive score is used as the criterion to select the optimal route that simultaneously meets network and operational requirements, avoiding operational anomalies caused by substandard network quality and improving the security and reliability of remote operations. Attached Figure Description

[0009] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This is a flowchart illustrating a service routing processing method provided in an embodiment of this application; Figure 2 This is a flowchart illustrating another service routing processing method provided in an embodiment of this application; Figure 3 This is a schematic diagram of the structure of a service routing processing device provided in an embodiment of this application; Figure 4 This is a schematic diagram of a service routing selection platform for a master-slave teleoperated surgical robot provided in an embodiment of this application; Figure 5 This is a network connection diagram of a service routing selection platform for a master-slave teleoperated surgical robot provided in an embodiment of this application; Figure 6 This is a schematic diagram of the force feedback service routing selection process for a master-slave remote surgical robot provided in an embodiment of this application; Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0011] To enable those skilled in the art to better understand the technical solutions in this application, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this application.

[0012] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0013] Currently, master-slave teleoperated surgical robots, leveraging 5G technology with its core advantages of low latency, high bandwidth, and high reliability, make remote surgery feasible. Force feedback technology can sense the minute forces applied to tissues during surgery and transmit this tactile sensation to the surgeon. Statistics show that surgical robots using force feedback reduce the force applied to tissues by 43%, thus reducing surgical trauma. Currently, some products are widely used in various minimally invasive surgical procedures, and the application of force feedback technology can improve surgical precision and reduce complications.

[0014] For example, a force feedback-based surgical cutting training system and method for surgical robots. This method is applicable to surgical cutting training scenarios for surgical robots. It introduces a 3D virtual soft tissue model construction module, a force feedback module for loading model size, position, and rendering mode to implement force feedback operations, and a classification cutting module for implementing surface and volume cutting. However, this method is a solution for analyzing force feedback data in a short-distance or on-site environment and does not consider the impact of packet loss, latency, and jitter on the force feedback effect during data packet forwarding by various service routing nodes in long-distance data transmission scenarios.

[0015] For example, a two-way teleoperation control system and method for robots based on master-slave force feedback. This method is based on the DH method for constructing the robot arm motion model and solving the forward motion, the Jacobian inverse method for solving the inverse motion, and the Newtonian method... The method proposes an Euler iterative approach to solve for the dynamics of the master hand and an admittance control method based on end-effector force feedback. However, this method does not consider the analysis of force feedback under 5G networks or other long-distance transmission technologies, and the bidirectional compliant control method is a non-convergent control method, which is prone to error accumulation over long periods of execution, affecting the surgeon's perception of force feedback.

[0016] For example, there is a multi-functional drone system and control method based on AR remote control. This method involves controlling the drone's flight attitude and the precise operation of its robotic arm through a smart glove. The smart glove acquires data from the drone's sensors and then provides force feedback to the operator. However, this method does not consider the impact of accumulated errors at various network nodes on force feedback during long-distance transmission. Furthermore, the significant time delay that occurs during the conversion of electrical signals to force signals by micro-motors or electromagnetic brakes can lead to a deviation between the operator's perceived force feedback and the actual force feedback received.

[0017] In response, this application proposes a service routing processing method and apparatus to solve problems such as optimal service routing selection based on force feedback services, data fusion of network quality and force feedback judgment strategy, and mismatch between network quality and action accuracy.

[0018] The following is in conjunction with the appendix Figures 1 to 7This application provides a detailed description of a service routing processing method, apparatus, electronic device, and computer program product through specific embodiments and application scenarios.

[0019] Figure 1 This application illustrates a service routing processing method according to an embodiment of the present application. This method can be executed by an electronic device, which may include a server and / or terminal devices. In other words, the method can be executed by software or hardware installed on the server and / or terminal devices, and includes the following steps: Step 110: Collect end-to-end network topology information of the mobile communication network, and generate a first matrix and a second matrix of candidate service routes. The first matrix includes device identifiers, and the second matrix includes network addresses.

[0020] The mobile communication network can include fifth-generation mobile communication networks (5G), as well as enhanced versions of fifth-generation mobile communication networks (such as 5G-Advanced) or higher-level mobile communication networks (such as sixth-generation mobile communication networks 6G). The end-to-end network topology information can be the information of all network devices that the data packet passes through from the master control area to the slave operation area of ​​the remote operating equipment, including the identifier and network address of each device.

[0021] The candidate service routes can be combinations of multiple paths that a data packet might take from the main control area to the operating area. Each path is composed of access network devices, transmission network devices, and core network devices connected sequentially. The first matrix stores the device identification information of the candidate service routes. Each row corresponds to one candidate service route, and each column corresponds to a network node on that route. The matrix elements are device identifiers. The second matrix stores the network address information of the candidate service routes. It has the same number of rows as the first matrix, with the same index in each row corresponding to the same candidate service route. The matrix elements are network addresses, such as IP addresses.

[0022] In this step, a topology collection command is first sent to the network management system. The network management system then requests the device identifiers and network addresses of each network element from the access network, transmission network, and core network of the mobile communication network. Access network elements may include 5G base stations, transmission network elements may include SPN or PTN transmission equipment, and core network elements may include intermediate user plane network elements and anchor user plane network elements. After collecting the above information, it is organized into a first matrix and a second matrix. Each row of the first matrix represents an end-to-end candidate service route, and records the base station identifiers, transmission equipment identifiers, intermediate user plane network element identifiers, and anchor user plane network element identifiers sequentially along the route. The second matrix corresponds to the rows of the first matrix and records the service IP addresses of each network node on the same candidate service route.

[0023] Step 120: Obtain the subscription information and network performance data of the remote operation device, calculate the network quality score of each candidate service route based on the subscription information and network performance data, and generate a third matrix.

[0024] The remote operation devices can include master-slave teleoperation devices, which include a master control area and a slave operation area. The master control area is controlled by the operator, and the slave operation area performs specific actions, such as master-slave teleoperation surgical robots.

[0025] The subscription information can be the service configuration parameters that the remotely operated device subscribes to in the mobile communication network, including network slice identifier, data network name, data network access identifier, etc., used to identify the network transmission channels that the device can use. Network performance data can be data reflecting network transmission quality extracted from detailed data records (XDR) collected from the service signaling system. Specifically, it can include indicators such as traffic, round-trip time (RTT), jitter, and packet loss rate of each network node.

[0026] The third matrix stores the network quality score for each candidate service route. Each row corresponds to a candidate service route and includes the average fusion weight and network quality score of the route in the three dimensions of network slice, data network name, and data network access identifier.

[0027] In this step, the subscription information of the remote operation device is first obtained. Specifically, the user's permanent identifier for the remote operation device is obtained, and the network slice, data network name, and data network access identifier information subscribed to by the user's permanent identifier are queried from the unified data management network element of the core network.

[0028] Then, network performance data is acquired and network quality scores are calculated. Using the network addresses in the second matrix as indices, detailed data records are collected from the service signaling system. Network performance data for each node in each candidate service route is extracted across three dimensions: network slice, data network name, and data network access identifier. This includes traffic, round-trip latency, jitter, and packet loss rate. Traffic reflects the node's service load, while round-trip latency, jitter, and packet loss rate reflect the node's transmission quality. The extracted network performance data is standardized. For positive indicators such as traffic (i.e., higher values ​​indicate better network conditions), forward standardization is used. For negative indicators such as round-trip latency, jitter, and packet loss rate (i.e., lower values ​​indicate better network conditions), reverse standardization is used. The standardized values ​​for each indicator range from 0 to 1. A weighted column vector is introduced to weight and fuse the four standardized indicators, calculating the fusion weight for each node. Then, for each candidate service route, the fusion weights of all its nodes are averaged across the dimensions of network slice, data network name, and data network access identifier to obtain the average network fusion weight for that route in each dimension. The arithmetic mean of the average network fusion weights for each dimension is then used as the network quality score for that route. The average network fusion weights and network quality scores for each candidate service route can be organized into a third matrix.

[0029] Step 130: Based on the second matrix and the third matrix, obtain the fusion matrix.

[0030] The fusion matrix can be formed by merging the second and third matrices. Each row contains a network address sequence of a candidate service route and its corresponding average network fusion weight and network quality score, which are used for subsequent network configuration distribution and comprehensive score calculation.

[0031] In this step, the second matrix generated in step 110 and the third matrix generated in step 120 are merged to obtain a fusion matrix. The number of rows in the fusion matrix is ​​the same as that in the second and third matrices. Each row first contains the network address sequence of the candidate service routes corresponding to that row in the second matrix, and then appends the network quality score corresponding to that row in the third matrix. This provides a unified data source for network configuration distribution and comprehensive score calculation in subsequent steps.

[0032] Step 140: Based on each candidate service route, execute network configuration distribution based on the fusion matrix, perform standard action tests and calculate force feedback deviation under the configured network to obtain the action deviation score of the candidate service route, calculate its comprehensive score based on the action deviation score of the candidate service route and the corresponding network quality score, and select the candidate service route with the highest comprehensive score as the optimal service route.

[0033] The network configuration distribution can be based on the network slice, data network name, and data network access identifier information corresponding to the current candidate service route in the fusion matrix, generating configuration instructions and distributing them to the access network, transmission network, and core network of the mobile communication network, so that the route can carry the service traffic of remote operation equipment.

[0034] The standard motion test can involve sending pre-set standard motion commands to the main control area of ​​the remote operating device. These commands can be generated based on the principle of virtual displacement and the Denavit-Hartenberg (DH) model, and are used to drive the end effector of the robotic arm in the operating area to move along a planned trajectory. Force feedback deviation is the difference between the actual executed coordinates and the planned coordinates of the end effector of the robotic arm, and this difference reflects the impact of network transmission quality on operational accuracy.

[0035] The action deviation score can be a value obtained by normalizing the force feedback deviation, used to quantify the execution accuracy of remotely operated devices under the current network routing conditions. The comprehensive score can be obtained by adding the action deviation score to the corresponding network quality score in the fusion matrix, so as to comprehensively reflect the advantages and disadvantages of the current candidate service route in terms of network quality, operation accuracy, etc.

[0036] In this step, network configuration is first issued for each candidate service route. Specifically, the device identifier sequence of the current candidate service route is obtained from the first matrix. Configuration instructions are generated based on the network slice, data network name, and data network access identifier information corresponding to the route in the fusion matrix, and issued to the access network, transmission network, and core network of the mobile communication network to make the route available. Standard action tests are then performed under the configured network. Standard action instructions based on the virtual displacement principle and DH parameter model are sent to the main control area of ​​the remote operating device. The main control area forwards the instructions to the slave operating area via the network, driving the robotic arm end effector to move according to the planned trajectory. The slave operating area feeds back the actual execution coordinates of the robotic arm end effector to the main control area. Based on the deviation between the actual coordinates and the planned coordinates, the deviation norm of each link of the robotic arm end effector is calculated. The arithmetic mean of the deviation norms of all standard actions and all links is normalized to obtain the action deviation score of the candidate service route. The action deviation score is added to the network quality score corresponding to the route in the fusion matrix to obtain the comprehensive score of the route.

[0037] Repeat the above process for each candidate service route. After obtaining the comprehensive score of each route, select the candidate service route with the highest comprehensive score as the optimal service route, and re-deploy the configuration of the optimal route according to the network configuration distribution method to complete the activation of the optimal route.

[0038] In this embodiment, by collecting end-to-end network topology information, generating a routing matrix, and combining it with network performance data to analyze the status of each node, the optimal route can be selected. This avoids packet loss, latency, jitter, and error accumulation in multi-node transmission, ensuring the stability of force feedback in long-distance mobile communication networks. By combining network quality scores with action deviation scores, dedicated high-quality routes are allocated to highly sensitive force feedback data, reducing signal transmission latency. Combined with action deviation calculations to correct execution deviations, this ensures consistency in force feedback perception between the master and slave ends. Using the maximization of the comprehensive score as the criterion, the optimal route that simultaneously meets network and operational requirements is selected, avoiding operational anomalies caused by substandard network quality and improving the security and reliability of remote operations.

[0039] Figure 2 This diagram illustrates a flowchart of another service routing processing method provided by an embodiment of this application. This method can be executed by an electronic device, which may include a server and / or terminal devices. In other words, the method can be executed by software or hardware installed on the server and / or terminal devices, and includes the following steps: Step 210: Based on step 110 of the above embodiment, collect end-to-end network topology information of the mobile communication network, and generate a first matrix and a second matrix of candidate service routes. The method of this embodiment may further include the following specific steps: According to the topology collection instruction, the device identifiers and network addresses of each network element are collected from the access network, transmission network and core network of the mobile communication network; based on the collected device identifiers and network addresses, the first matrix and the second matrix are generated; wherein, each row of the first matrix corresponds to a sequence of device identifiers of a candidate service route, and the corresponding row in the second matrix corresponds to a sequence of network addresses of the candidate service route.

[0040] The topology collection command can be a control message used to trigger the collection of network topology information. It can be initiated by the network topology management function and used to instruct the network management system to perform network element information collection.

[0041] The access network can be the part of a mobile communication network that connects user equipment (UE) and the core network, including base station equipment (such as 5G NR base stations), which is responsible for transmitting, receiving, and processing radio signals. The transport network can be the part of a mobile communication network that connects the access network and the core network, including transport equipment (such as SPN equipment and PTN equipment), which is responsible for carrying and forwarding data packets between the base station and the core network. The core network can be the core part of a mobile communication network responsible for mobility management, session management, user data management, and data routing, including user plane network elements (such as intermediate user plane functions and anchor user plane functions) and control plane network elements (such as access and mobility management functions and session management functions).

[0042] The device identifier is used to uniquely identify the network device, such as base station ID, transmission equipment ID, user plane network element ID, etc. The network address is used to locate and address devices within the network, such as service IP addresses. A device identifier sequence can be a set of identifiers of network devices arranged sequentially on a candidate service route. A network address sequence can be a set of network addresses of the network devices corresponding to the device identifier sequence, used for subsequent data collection and configuration distribution.

[0043] In this step, the network topology information collection operation is first performed in response to the initiated topology collection command. This topology collection command can be generated by the network topology management function and is used to trigger the end-to-end collection of network element information for the entire mobile communication network. According to this topology collection command, the device identifiers and network addresses of each network element are collected from the access network, transmission network, and core network of the mobile communication network. Specifically, the device identifiers (e.g., base station IDs) and their service IP addresses of each base station are collected from the access network; the device identifiers and their service IP addresses of each transmission device (e.g., SPN devices, PTN devices) are collected from the transmission network; and the device identifiers and their service IP addresses of each user plane network element (e.g., intermediate user plane functions, anchor user plane functions) are collected from the core network. The collection operation can be executed by the network management system, which establishes a pre-established connection with each network element via an IP network, supporting the issuance of configuration commands and the reporting of status information. Then, a first matrix and a second matrix are generated based on the collected device identifiers and network addresses. The first matrix stores the device identification information of candidate service routes. Each row corresponds to an end-to-end candidate service route, and records the access network device identifiers, transmission network device identifiers, and core network user plane element identifiers sequentially along the route in the order of data packet transmission, forming the device identification sequence of the route. The second matrix has the same number of rows as the first matrix, and each row corresponds to the network address sequence of the candidate service route, that is, the service IP addresses of each network element are recorded in the same order.

[0044] Step 220: Obtain the subscription information and network performance data of the remote operation device, calculate the network quality score of each candidate service route based on the subscription information and network performance data, and generate a third matrix.

[0045] Step 230: Based on the second matrix and the third matrix, obtain the fusion matrix.

[0046] Step 240: Based on the fusion matrix, network configuration is issued for each candidate service route. Standard action tests are performed under the configured network, and force feedback deviation is calculated to obtain the action deviation score of the candidate service route. The comprehensive score is calculated based on the action deviation score of the candidate service route and the corresponding network quality score. The candidate service route with the highest comprehensive score is selected as the optimal service route.

[0047] Steps 220-240 can be found above. Figure 1 The specific descriptions of steps 120-140 in the illustrated embodiment are provided, and the same technical effects can be achieved. To avoid repetition, they will not be repeated here.

[0048] In this embodiment, the mobile communication network topology is transformed into a structured numerical matrix form using a first matrix and a second matrix. The first matrix provides device-level path information for each candidate route, used for subsequent network configuration distribution. The second matrix provides corresponding network address information, used for subsequent network performance data collection, such as collecting detailed data records indexed by IP address. The row correspondence between the two matrices ensures that the device identifier of each candidate route can be accurately associated with its network address.

[0049] In yet another exemplary embodiment, based on step 120 of the above embodiment, the subscription information and network performance data of the remote operation device are obtained, and the network quality score of each candidate service route is calculated according to the subscription information and network performance data to generate a third matrix. The method of this embodiment may also include the following specific steps: Obtain the user's permanent identifier for the remote operation device, and query the unified data management network element of the core network for the network slice, data network name, and data network access identifier subscribed to by the user's permanent identifier; using the network address in the second matrix as an index, collect detailed data records, and extract the network performance data of each node in each candidate service route based on the network slice, data network name, and data network access identifier; standardize the extracted network performance data, perform weighted fusion on the standardized network performance data, and calculate the fusion weight of each node; calculate the average network fusion weight of the candidate service route based on the fusion weight of all nodes on each candidate service route, obtain the network quality score of the candidate service route based on the average network fusion weight, and generate the third matrix.

[0050] The user's permanent identifier is a unique identifier for a remotely operated device in a mobile communication network, used to identify and authenticate the device within the network, such as the Subscription Permanent Identifier (SUPI) in a 5G network. The unified data management network element is the network element in the core network responsible for managing user subscription data. It stores the user's permanent identifier and various service configuration information they have subscribed to, such as the Unified Data Management Function (UDM).

[0051] Network slices are logically isolated virtual networks partitioned within a mobile communication network. Each network slice can provide differentiated network characteristics to meet specific service requirements, such as low-latency slices and high-bandwidth slices. The data network name identifies the data network to which the remotely operated device is connected, determining which data network services the device can access. The data network access identifier identifies the data network access point, determining the specific location where data packets are forwarded from the core network exit to the external network.

[0052] The detailed data record (XDR) is a data record collected from the service signaling system that records the transmission quality of each node in the network, including network performance data such as traffic, round-trip time, jitter, and packet loss rate. Network performance data can be extracted from the detailed data record and reflect the transmission quality of each network node, including but not limited to traffic, round-trip time, jitter, and packet loss rate.

[0053] The fusion weight is a weighted fusion of multiple network performance data for a single network node, reflecting the overall network quality of that node, with a value range of 0 to 1. The average network fusion weight can be the arithmetic mean of the fusion weights of all nodes on a candidate service route, reflecting the overall network quality of that route in a certain dimension (e.g., network slice, data network name, and data network access identifier). The network quality score is a comprehensive score obtained by further arithmetic averaging the average network fusion weights across multiple dimensions, used to quantify the overall network quality of each candidate service route.

[0054] Specifically, first, the permanent user identifier of the remote operating device is obtained. This identifier is stored in the user identity module of the remote operating device and can be obtained by sending a request message to the main control area of ​​the remote operating device. After obtaining the permanent user identifier, a subscription information query request is initiated to the unified data management network element of the core network. The unified data management network element, as the network element in the core network that uniformly stores user subscription data, records various service configuration parameters subscribed to by each user's permanent identifier. The queried subscription information may include: the network slice identifier that the remote operating device can use, the name of the data network that can be accessed, and the available data network access identifier.

[0055] Using the network address sequence corresponding to each candidate service route in the second matrix as an index, network performance data for each node in each candidate service route is extracted from the detailed data records, focusing on three dimensions: network slice, data network name, and data network access identifier. The extracted network performance data may include, but is not limited to: traffic, round-trip time (RTT), jitter, and packet loss rate. These four metrics reflect the transmission quality of network nodes from different dimensions: traffic reflects the node's load pressure, while RTT, jitter, and packet loss rate reflect the node's transmission stability and reliability.

[0056] The extracted network performance data is standardized. Since the metrics have different dimensions, and some are positive (higher values ​​indicate better network conditions, such as traffic) while others are negative (lower values ​​indicate better network conditions, such as round-trip latency, jitter, and packet loss rate), different standardization methods are required. For positive metrics, forward standardization is used, mapping the original values ​​to the 0-1 range; higher values ​​indicate better performance. For negative metrics, backward standardization is used, mapping the original values ​​to the 0-1 range; higher values ​​indicate better performance (i.e., the smaller the original value, the larger the standardized value). After standardization, all metrics fall within the 0-1 range, with higher values ​​indicating better network conditions, facilitating subsequent weighted fusion.

[0057] A weighted column vector is introduced to perform a weighted fusion of the four standardized metrics. The weighted column vector contains four weight coefficients, corresponding to the importance of traffic, round-trip latency, jitter, and packet loss rate, respectively. Each weight coefficient ranges from 0 to 1, and the sum of the four coefficients is 1. The specific values ​​of the weight coefficients can be dynamically adjusted according to business needs. For example, for latency-sensitive surgical procedures, the weight coefficient corresponding to round-trip latency can be increased, while for services with high reliability requirements, the weight coefficient corresponding to packet loss rate can be increased.

[0058] The weighted fusion calculation method is as follows: multiply the standardized traffic value by its weight coefficient, add the standardized round-trip latency value multiplied by its weight coefficient, add the standardized jitter value multiplied by its weight coefficient, and finally add the standardized packet loss rate multiplied by its weight coefficient to obtain the fusion weight of the node. This fusion weight can comprehensively reflect the overall network quality of the node in terms of traffic, latency, jitter, and packet loss rate.

[0059] For each candidate service route, the average network fusion weight is first calculated based on the fusion weights of all nodes on that route, for network slices, data network names, and data network access identifiers. Specifically, for network slices, the arithmetic mean of the fusion weights of each node on the route at the network slice dimension is taken to obtain the average network fusion weight of the route at the network slice dimension. Similarly, the average network fusion weights at the data network name dimension and the data network access identifier dimension are calculated respectively.

[0060] Then, the arithmetic mean of the average network fusion weights of the above three dimensions is taken to obtain the network quality score of the candidate service route. This score can comprehensively reflect the overall network quality of the route in terms of network slicing, data network name, and data network access identifier. The value range is 0 to 1, with higher values ​​indicating better network quality.

[0061] The network quality score of each candidate service route is organized into a third matrix. Each row of the third matrix corresponds to a candidate service route and includes, in order: the average network fusion weight under the network slice dimension, the average network fusion weight under the data network name dimension, the average network fusion weight under the data network access identifier dimension, and the network quality score. This matrix serves as the basis for generating the fusion matrix in subsequent steps.

[0062] In this embodiment, by using a user's permanent identifier to query subscription information from a unified data management network element, the accuracy and reliability of the obtained transmission channel parameters are ensured. Detailed data records are collected using network addresses as indexes, enabling precise location and extraction of network performance data for each node. Standardization processes unify indicators of different dimensions to the same scale, facilitating fusion and comparison. Weighted fusion supports flexible adjustment of the weights of each indicator according to business needs, adapting to the differentiated network performance requirements of different scenarios.

[0063] In yet another exemplary embodiment, the network performance data includes traffic, round-trip time, jitter, and packet loss rate.

[0064] The network performance data can be extracted from detailed call detail records (CDRs) of the mobile communication network and used to quantify the transmission quality of each network node. These metrics reflect the load status, transmission latency, latency variation, and transmission reliability of network nodes from different dimensions.

[0065] Traffic refers to the amount of data passing through a network node per unit of time, reflecting the node's service load. The unit of traffic is usually bits per second (bps) or megabits per second (Mbps). In network quality evaluation, traffic is a positive indicator. Higher traffic indicates a greater service load on the node and a higher importance in the network topology. However, excessively high traffic may also indicate node congestion, requiring a comprehensive assessment in conjunction with other indicators.

[0066] Round-trip time (RTT) refers to the total time required for a data packet to travel from the sender to the receiver and back immediately, reflecting the latency characteristics of network transmission. The unit of RTT is usually milliseconds (ms). In network quality evaluation, RTT is a negative indicator. Lower latency indicates faster network transmission, making it more suitable for remote operation services with high real-time requirements. Higher latency indicates slower network transmission, potentially leading to delays in operation commands and force feedback data.

[0067] Jitter refers to the variation in round-trip time, reflecting the stability of network transmission. The unit of jitter is usually milliseconds (ms). In network quality evaluation, jitter is a negative indicator. Lower jitter indicates more stable network transmission, with data packets arriving at relatively uniform time intervals. Higher jitter indicates more unstable network transmission, potentially leading to sudden arrivals or delays of data packets, affecting the continuity and smoothness of feedback.

[0068] Packet loss rate refers to the proportion of data packets lost during transmission, reflecting the reliability of network transmission. Packet loss rate is usually expressed as a percentage (%), ranging from 0% to 100%. In network quality evaluation, packet loss rate is a negative indicator. The lower the packet loss rate, the more reliable the network transmission, and the more complete the data packets will arrive. The higher the packet loss rate, the less reliable the network transmission, which may lead to the loss of control commands or missing force feedback data, seriously affecting the security of remote operations.

[0069] In yet another exemplary embodiment, based on step 120 of the above embodiment, the average network fusion weight of the candidate service route is calculated according to the fusion weight of all nodes on each candidate service route, and the network quality score of the candidate service route is obtained according to the average network fusion weight. The method of this embodiment may further include the following specific steps: Calculate the arithmetic mean of the fusion weights of all nodes on each candidate service route to obtain the average network fusion weight; based on the network slice, data network name, and data network access identifier, calculate the arithmetic mean of the average network fusion weights to obtain the network quality score.

[0070] The fusion weight can be a value obtained by standardizing and weighting multiple network performance data (such as traffic, round-trip latency, jitter, packet loss rate, etc.) of a single network node, reflecting the overall network quality of the node in a certain dimension, with a value range of 0 to 1.

[0071] The average network fusion weight is the arithmetic mean of the fusion weights of all nodes on a candidate service route. It reflects the overall network quality of the route across three dimensions: network slice, data network name, and data network access identifier. The network slice dimension is divided according to the network slice identifier; different network slices correspond to different resource isolation and service quality levels. The data network name dimension is divided according to the data network name; different data network names correspond to different data network access points. The data network access identifier dimension is divided according to the data network access identifier; different identifiers correspond to different exit points from user plane functions to the data network.

[0072] The network quality score is a comprehensive score obtained by taking the arithmetic mean of the average network fusion weights of the above three dimensions. It is used to quantify the overall network quality of each candidate service route. The value range is 0 to 1, and the higher the value, the better the network quality.

[0073] The calculated network quality score for each candidate service route is organized into a third matrix. The third matrix has the same number of rows as the first and second matrices, with each row corresponding to a candidate service route and containing that route's network quality score. This matrix is ​​used in subsequent steps to merge with the second matrix to generate a fusion matrix.

[0074] In this embodiment, the average network fusion weight is obtained by calculating the arithmetic mean of the fusion weights. This method is simple, computationally inexpensive, and suitable for remote operation scenarios with high real-time requirements. Calculating the average network fusion weight from three dimensions—network slice, data network name, and data network access identifier—covers key factors affecting service transmission quality in mobile communication networks. Furthermore, the arithmetic mean of the average network fusion weights from the three dimensions is used to obtain a network quality score. This integrates the multi-dimensional evaluation results into a single score, facilitating comprehensive comparison and route selection with the action deviation score in subsequent steps.

[0075] In yet another exemplary embodiment, based on step 130 of the above embodiment, a fusion matrix is ​​obtained based on the second matrix and the third matrix. The method of this embodiment may further include the following specific steps: The second and third matrices are merged to generate the fusion matrix; wherein each row of the fusion matrix includes a network address sequence of a candidate service route, as well as the average network fusion weight and network quality score of the candidate service route under the network slice dimension, data network name dimension, and data network access identifier dimension.

[0076] The second matrix stores the network address information of candidate service routes. Each row corresponds to a candidate service route, and the IP addresses of each network node traversed by the route are recorded sequentially according to the data packet transmission order, forming a network address sequence for the route. This network address sequence is used for subsequent network performance data collection (e.g., collecting detailed call detail records indexed by IP addresses) and network configuration distribution. The third matrix stores the network quality scores of candidate service routes. It has the same number of rows as the second matrix, with each row corresponding to a candidate service route. The scores include the average network fusion weight under the network slice dimension, the average network fusion weight under the data network name dimension, the average network fusion weight under the data network access identifier dimension, and the network quality score.

[0077] Specifically, the second and third matrices are merged to generate a fusion matrix. The merging operation is performed as follows: for the i-th candidate service route, the network address sequence in the i-th row of the second matrix is ​​combined sequentially with the values ​​in the i-th row of the third matrix to form the i-th row of the fusion matrix. All candidate service routes are processed in this manner until all rows are merged.

[0078] The generated fusion matrix has the following structural features: each row of the matrix corresponds to a candidate service route. The row first contains the network address sequence of the route (i.e., the service IP addresses of each network node through which the route passes, arranged in the order of data packet transmission), and then appends the average network fusion weight of the route in the network slice dimension, the average network fusion weight in the data network name dimension, the average network fusion weight in the data network access identifier dimension, and the network quality score.

[0079] Taking a typical candidate service route as an example, this route passes through five network nodes: the access network base station, the first transmission device, the second transmission device, intermediate user plane network elements, and anchor user plane network elements. The corresponding rows in the fusion matrix for this route would then contain, in sequence: the service IP address of the base station, the service IP address of the first transmission device, the service IP address of the second transmission device, the service IP address of the intermediate user plane network element, and the service IP address of the anchor user plane network element (forming a network address sequence). Following this are the route's average network fusion weight at the network slice dimension, the average network fusion weight at the data network name dimension, the average network fusion weight at the data network access identifier dimension, and the network quality score.

[0080] In this embodiment, a fusion matrix is ​​used to associate the network topology information stored in the second matrix with the network quality assessment results stored in the third matrix within the same data structure, providing a unified data source for subsequent network configuration deployment. Based on the network address sequence of the current route in the fusion matrix, the network nodes requiring configuration can be located. Based on the corresponding network quality assessment data, the network quality status of the route in each dimension can be determined. In subsequent steps, when calculating the comprehensive score for each route, the comprehensive network quality score can be directly read from the fusion matrix without the need for index matching between multiple matrices. The row correspondence of the fusion matrix is ​​clear, avoiding data inconsistencies caused by matrix index misalignment. The fusion matrix also retains independent scores and a comprehensive score for the three dimensions, satisfying both overall evaluation needs and supporting refined analysis for specific dimensions.

[0081] In yet another exemplary embodiment, based on step 140 of the above embodiment, network configuration is issued based on the fusion matrix according to each candidate service route. The method of this embodiment may further include the following specific steps: Obtain the device identifier sequence of the current candidate service route from the first matrix; generate a configuration instruction based on the network slice, data network name, and data network access identifier corresponding to the fusion matrix; and send the configuration instruction to the access network, transmission network, and core network of the mobile communication network.

[0082] The first matrix stores the device identifier information of candidate service routes. Each row corresponds to a candidate service route, and the device identifiers of each network node traversed by the route are recorded sequentially according to the data packet transmission order, forming a device identifier sequence for the route. The device identifier sequence is a set of device identifiers of each network node on a candidate service route arranged in sequence, used to uniquely identify each network element in the network, such as base station ID, transmission equipment ID, user plane network element ID, etc.

[0083] The fusion matrix is ​​formed by merging the second matrix (network address sequence) and the third matrix (network quality assessment data). Each row contains the network address sequence of a candidate service route, the average network fusion weight across three dimensions, and the network quality score. The network slice, data network name, and data network access identifier information are derived from the subscription information query results and are reflected in the calculation process of the network quality assessment data.

[0084] The configuration instructions can be control commands used to perform configuration operations on various network elements. Specifically, they can include configuring network slice identifiers, configuring data network names, configuring data network access identifiers, etc.

[0085] Specifically, the device identifier sequence of the current candidate service route is first obtained from the first matrix. For the m-th candidate service route to be configured, the device identifier sequence is read from the m-th row of the first matrix. This sequence may include, in sequence, the device identifiers of the access network base station, the device identifiers of each transmission network device, and the device identifiers of the core network user plane elements. These device identifiers are used to uniquely identify each network element that needs to be configured in the network.

[0086] Based on the network slice, data network name, and data network access identifier information corresponding to the current candidate service route in the fusion matrix, configuration instructions are generated. The network slice identifies the logically isolated virtual network resource combination to which the route belongs. The data network name identifies the data network to which the route accesses. The data network access identifier identifies the specific location from the core network egress to the external network. Based on these parameters, corresponding configuration instructions are generated.

[0087] The generated configuration commands are distributed to the access network, transmission network, and core network of the mobile communication network. The distribution order follows the data packet transmission path: first, the access network base stations are configured; then, the transmission network equipment is configured; and finally, the core network user plane elements are configured. The distribution method can utilize the standard configuration protocol interface between the network management system and each network element.

[0088] After receiving the configuration command, each network element executes the corresponding configuration operation and returns the execution result. Once all network elements have successfully completed the configuration, the current candidate service route becomes available and can be used to carry service traffic for remotely operated devices.

[0089] In this embodiment, the device identifier sequence is obtained from the first matrix to ensure that the configuration command can accurately locate the network elements that need to be configured. Configuration commands are generated based on the network slices, data network names, and data network access identifiers in the fusion matrix to correlate configuration parameters with network quality assessment results. Configurations are issued sequentially in the order of access network, transmission network, and core network, conforming to the path logic of data packet transmission, ensuring the integrity of end-to-end routing, and providing a usable network environment for subsequent standard action tests and force feedback deviation calculations, so that the test results can truly reflect the network quality of the current route.

[0090] In yet another exemplary embodiment, based on step 140 of the above embodiment, a standard action test is performed under the configured network and the force feedback deviation is calculated to obtain the action deviation score of the candidate service route. The method of this embodiment may further include the following specific steps: Send standard motion commands based on the principle of virtual displacement to the main control area of ​​the remote operating device; receive the actual coordinates of the end effector of the robotic arm fed back from the operating area of ​​the remote operating device; calculate the normalized mean motion deviation based on the deviation between the actual coordinates and the planned coordinates, and use it as the motion deviation score of the candidate service route.

[0091] The remote operating device can be a master-slave teleoperation device, including a master control area and a slave operating area. The master control area is operated by the operator and is used to send action commands. The slave operating area executes the specific actions and feeds back the execution results to the master control area. Specifically, the remote operating device includes a master-slave teleoperation surgical robot.

[0092] The principle of virtual displacement, also known as the principle of virtual work, is a fundamental principle of analytical mechanics. This principle states that for a system of particles with ideal constraints, the necessary and sufficient condition for equilibrium is that the sum of the virtual work done by all active forces in any virtual displacement is zero. In this application, the principle of virtual displacement is used to construct a force feedback model for a robotic arm, calculating the balance relationship between the joint torque and virtual work of the robotic arm's end effector under ideal constraints.

[0093] The standard motion commands refer to pre-defined, standardized sequences of robotic arm movements used to test network routing quality. These movements have defined planned trajectories and expected coordinates. By comparing the deviations between the actual executed coordinates and the planned coordinates, the impact of network transmission quality on operational accuracy can be quantified. The end effector of a robotic arm is generally an actuator installed at the end of the robotic arm that directly contacts the manipulated object (e.g., surgical tissue), such as needle holders, electrosurgical units, and clamps. Actual coordinates generally refer to the actual position coordinates reached by the end effector during the execution of the standard motion commands, collected and fed back from sensors in the operating area. Planned coordinates generally refer to the theoretically expected position coordinates of the end effector when executing the standard motion commands, given by the motion planning data in the standard motion commands.

[0094] The action deviation score is a value obtained by normalizing the deviation between the actual coordinates and the planned coordinates. It is used to quantify the execution accuracy of remotely operated devices under the current network routing conditions. The value range is from 0 to 1, with a larger value indicating higher execution accuracy (smaller deviation) and a smaller value indicating lower execution accuracy (larger deviation).

[0095] Specifically, after the network configuration for the current candidate service route is distributed, the route is now available. At this point, standard action commands based on the virtual displacement principle are sent to the main control area of ​​the remote operating device. These standard action commands can be pre-designed, standardized robotic arm action sequences used to test network routing quality. These action commands are generated based on the virtual displacement principle and the DH parameter model, and include the motion parameters of each joint of the robotic arm and the planned trajectory of the end effector.

[0096] During the execution of standard motion commands, the operator area uses its sensor system to collect the actual position coordinates of the robotic arm's end effector in real time. These sensors include joint angle sensors, position sensors, and force / torque sensors, which can record the actual position of the end effector in three-dimensional space with high accuracy. The operator area then feeds back the collected coordinate data to the main control area via a network. After receiving this data, the main control area compares it with the planned coordinates in the standard motion commands, providing a data basis for subsequent deviation calculations.

[0097] It should be noted that the deviation between the actual coordinates and the planned coordinates is affected by a variety of factors, among which network transmission quality (such as latency, jitter, and packet loss rate) is one of the main influencing factors. Under the same robotic arm and the same standard action commands, a larger deviation indicates a worse quality of the current network routing, and vice versa.

[0098] Based on the received actual coordinates and planned coordinates, a normalized mean motion deviation is calculated as the motion deviation score for the candidate service route. First, for each link (or joint) of the robotic arm's end effector, the deviation norm between the actual and planned coordinates is calculated for each standard motion. This deviation norm reflects the spatial distance difference between the actual execution position and the expected planned position. Second, considering the maximum executable coordinate range of the robotic arm's end effector, the above deviation norm is compared with the maximum coordinate norm to obtain a normalized deviation value. After normalization, the deviation value ranges from 0 to 1, where a larger value indicates that the actual coordinates are closer to the planned coordinates (smaller deviation), and a smaller value indicates a greater degree of deviation (larger deviation). Then, the arithmetic mean of the normalized deviation values ​​for all standard motions and all links is taken to obtain the motion deviation score for the candidate service route. This score comprehensively reflects the overall accuracy of the robotic arm's end effector in executing all standard motions under the current network routing conditions. The range of the action deviation score is 0 to 1. The closer the score is to 1, the higher the execution accuracy, meaning that the network transmission quality has a smaller impact on the operation accuracy. The closer the score is to 0, the lower the execution accuracy, meaning that the network transmission quality has a greater impact on the operation accuracy.

[0099] The calculated action deviation score will be fused with the network quality score in the fusion matrix to jointly evaluate the overall performance of the current candidate service route. The action deviation score reflects the impact of network transmission quality on actual operational accuracy and serves as an effective supplement to the network quality score. The network quality score evaluates route quality from the network side (traffic, latency, jitter, packet loss rate), while the action deviation score directly verifies route quality from the perspective of actual execution accuracy. The combination of the two can comprehensively and accurately reflect the applicability of candidate service routes to remote operation services.

[0100] In this embodiment, standard action commands are generated based on the principle of virtual displacement. These commands have clearly defined planned trajectories and expected coordinates, ensuring the comparability and repeatability of test results. Actual coordinates are collected from the sensor system in the operating area, enabling objective quantification of the robotic arm's execution accuracy. A normalization method is used to calculate the mean action deviation, quantifying the deviation into a standard score between 0 and 1, facilitating fusion calculation with network quality scores. The action deviation score can verify network routing quality from the perspective of service execution accuracy, complementing the network-side score and improving the accuracy of routing evaluation.

[0101] In yet another exemplary embodiment, the calculation of the mean motion deviation is based on the virtual displacement principle and the DH model. It is obtained by calculating the deviation norm between the actual coordinates and the planned coordinates of the end effector of the robotic arm, and then normalizing the deviation norms of all standard motions and all links and taking the arithmetic mean.

[0102] Corresponding to the service routing processing method provided in the above embodiments, based on the same technical concept, this application also provides a service routing processing apparatus. See [link to previous document]. Figure 3 The device 400 includes an information collection module 410, a quality scoring module 420, a fusion processing module 430, and a routing selection module 440.

[0103] The information collection module 410 is used to collect end-to-end network topology information of the mobile communication network and generate a first matrix and a second matrix of candidate service routes. The first matrix includes device identifiers, and the second matrix includes network addresses. The quality scoring module 420 is used to obtain the subscription information and network performance data of the remote operation device, calculate the network quality score of each candidate service route based on the subscription information and network performance data, and generate a third matrix. The fusion processing module 430 is used to obtain a fusion matrix based on the second matrix and the third matrix. The route selection module 440 is used to execute network configuration distribution based on the fusion matrix for each candidate service route, perform standard action tests and calculate force feedback deviation under the configured network to obtain the action deviation score of the candidate service route, calculate its comprehensive score based on the action deviation score and the corresponding network quality score of the candidate service route, and select the candidate service route with the highest comprehensive score as the optimal service route.

[0104] This application embodiment collects end-to-end network topology information, generates a routing matrix, and analyzes the status of each node by combining network performance data. This allows for the selection of the optimal route, avoiding packet loss, latency, jitter, and error accumulation in multi-node transmission, thus ensuring the stability of force feedback in long-distance mobile communication networks. By combining network quality scores with action deviation scores, dedicated high-quality routes are allocated to highly sensitive force feedback data, reducing signal transmission latency. Action deviation calculations correct execution deviations, ensuring consistency in force feedback perception between the master and slave ends. Maximizing the comprehensive score is used as the criterion to select the optimal route that simultaneously meets network and operational requirements, avoiding operational anomalies caused by substandard network quality and improving the security and reliability of remote operations.

[0105] It should be noted that the service routing processing apparatus and the service routing processing method provided in this application embodiment are based on the same application concept. Therefore, the specific implementation of this embodiment can refer to the implementation of the aforementioned service routing processing method, and repeated details will not be described again.

[0106] In yet another exemplary embodiment, a service routing selection platform for a master-slave teleoperated surgical robot is provided, including a device that simultaneously accesses a 5G network core network, transport network, and wireless network, as well as a control area and operating area for the teleoperated surgical robot. This device has the function of collecting, processing, and analyzing 5G topology and mobile internet service data. It selects the optimal service routing path by using the deviation between the planned and actual routing based on fusion force feedback and network quality as metrics.

[0107] See Figure 4 The service routing platform includes the master control area of ​​the master-slave teleoperated surgical robot, the 5G core network, the 5G transmission network, the 5G wireless network, the slave operation area of ​​the teleoperated surgical robot, the integrated network management system, the service signaling system, and the service routing system.

[0108] This service routing platform has the following communication interfaces: (1) A1 interface: It is the communication interface between the business routing platform and the main control area of ​​the master-slave teleoperation surgical robot, realizing the sending and feedback of robot operation commands, as well as data collection.

[0109] (2) A2 interface: It is the communication interface between the service routing platform and the integrated network management system, and realizes the collection of network configurations of wireless network, transmission network and core network and the issuance of configuration commands.

[0110] (3) A3 interface: It is the communication interface between the service routing platform and the service signaling system, and collects the full amount of XDR data for subsequent service routing model establishment and numerical calculation.

[0111] (4) A4 interface: It is the communication interface between the business routing platform and the master-slave teleoperated surgical robot in the operation area, and collects the actual coordinate information of the surgical robot end effector.

[0112] (5) B1 interface: This is the interface between the network topology management module and the service routing module of the platform. Its function is to allow the network topology management module to obtain network topology data from the service routing module.

[0113] (6) B2 interface: This is the interface between the routing weight module and the service routing module. Its function is to transmit various types of matrix data after calculation from the routing weight module to the service routing module.

[0114] (7) B3 interface: This is the interface between the fusion algorithm module and the service routing module. Its function is to output the calculation results of the routing decision algorithm from the fusion algorithm module to the service routing module.

[0115] (8) B4 interface: This is the interface between the motion orchestration module and the service routing module. Its function is to synchronize the information of the surgical robot operation actions from the motion orchestration module to the service routing module.

[0116] (9) B5 interface: This is the interface between the force feedback module and the service routing module. Its function is to push the transmission requirement parameters of the force feedback data to ensure that the service routing module allocates a dedicated optimal route for the highly sensitive force feedback data.

[0117] The service routing platform includes the following processing modules: (1) Network Topology Management Module: Initiates a 5G end-to-end network topology information collection command, driving the service routing module to obtain device IDs and service IP information of the wireless network, transmission network, and core network from the integrated network management system. It issues service routing configuration commands to the fusion algorithm module and receives optimal routing decisions from each route.

[0118] (2) Service routing module: forwards data collection and instruction issuance from other modules.

[0119] (3) Routing weight module: Collect XDR call detail records and extract network performance data such as traffic, RTT latency, jitter, and packet loss rate for each routing node. Construct tensor data for three channels: network slice, DNN, and DNAI, and complete the standardization and weighted fusion calculation to generate the routing weight matrix.

[0120] (4) Fusion Algorithm Module: Generates 5G network configuration instructions and distributes slice, DNN, and DNAI configurations to the 5G wireless network, 5G transmission network, and 5G core network. Sends standard robotic arm movement instructions to the robot's main control area to trigger teleoperation movement orchestration and force feedback testing. Receives actual and planned coordinate data of the robotic arm from the robot's main control area, calculates the normalized coordinate deviation value, and completes the fusion calculation by combining the data from the network weight matrix. Traverses all service routes, selects the optimal route with the highest fusion value, reissues the configuration, and completes the activation of the optimal route.

[0121] (5) Motion orchestration module: plans motion parameters and converts them into execution instructions for the robot in the operation area, ensuring the standardization and accuracy of motion execution.

[0122] (6) Force Feedback Calculation Module: Based on the principle of virtual displacement and the DH model, a force feedback model of the multi-rigid-body system of the robotic arm is constructed, ignoring friction to calculate the joint torque and virtual work balance relationship under ideal constraints. Combined with the force feedback data transmitted by the 5G network, the motion deviation of the robotic arm end effector is corrected to ensure the consistency of force feedback between the master and slave ends; at the same time, the coordinate data of the motion execution is recorded to provide a basis for subsequent deviation analysis.

[0123] Considering that message transmission in master-slave teleoperated surgical robots traverses a 5G network, involving 5G wireless network, transmission network, core network equipment, and the master control area and slave operation area of ​​the master-slave teleoperated surgical robot, this service routing platform can be deployed on a cloud-based ARM virtual machine based on the Ubuntu 16.04.1 LTSXenial Xerus and ROSKinetic Kame framework. The 5G wireless network, transmission network, core network equipment, integrated network management system, service signaling system, and service routing platform are all pre-configured through the IP network.

[0124] See Figure 5 The specific implementation methods of this service routing platform include the following: (1) The main control area of ​​the remote surgical robot is connected to multiple AUPF devices in the 5G core network via a dedicated line. Each AUPF and multiple IUPFs form a full connection through the transmission network. The IUPFs are connected to multiple 5G NR base stations through SPN / PTN transmission equipment. The 5G base stations form a wireless connection with the remote surgical robot's operating area via wireless. Through the above connections, the interactive service route of control message packets between the main control area and the operating area of ​​the remote surgical robot is formed.

[0125] (2) The 5G core network AMF, SMF, CHF and UDM devices are interconnected through the IP network, and the SMF is used to achieve session control with the UPF through the IP network.

[0126] (3) The integrated network management system connects to AUPF, IUPF, AMF, SMF, CHF, UDM, 5G transmission network and 5G wireless network through the IP network. This connection is used by the integrated network management system to issue configuration commands to the above network elements and devices.

[0127] (4) The service signaling system collects signaling information from the 5G transmission network, 5G wireless network and 5G core network through the IP network, and generates XDR files based on the signaling information for network topology generation of the service routing selection platform.

[0128] (5) The service routing selection platform generates a service routing matrix by collecting comprehensive network management and service signaling data, and integrates surgical robot force feedback deviation information to comprehensively measure and output the optimal service route.

[0129] See Figure 6 Based on this service routing platform, a master-slave remote surgical robot force feedback service routing process specifically includes the following steps: Step 311: Collect end-to-end network topology information. The network topology management module of the master-slave remote surgical robot force feedback system sends a command message to the service routing module to collect 5G network end-to-end network topology information. This message collects information from all 5G core network, 5G transmission network, and 5G wireless network devices for subsequent network topology information matrix construction. The service routing module forwards this message to the integrated network management system for 5G end-to-end network topology collection.

[0130] Step 312: Collect network topology information of the 5G wireless network. The integrated network management system collects network topology information from the 5G wireless network. Specific information includes the 5G base station ID and service IP address information. The 5G wireless network then feeds this information back to the integrated network management system.

[0131] Step 313: Collect network topology information of the 5G transmission network. The integrated network management system collects network topology information from the 5G transmission network. Specific information includes the 5G SPN device ID and service IP address information. The 5G transmission network then feeds this information back to the integrated network management system.

[0132] Step 314: Collect network topology information of the 5G core network. The integrated network management system collects network topology information from the 5G core network. Specific information includes the I-UPF device ID and its service IP address, and the A-UPF device ID and its service IP address. The 5G core network then feeds this information back to the integrated network management system.

[0133] Step 315: Collect and organize the received information to form 5G end-to-end network topology information. The integrated network management system organizes the collected 5G wireless network, transmission network, and core network service routing information to generate the following: matrix and ,matrix The format is as follows:

[0134] matrix The format is as follows:

[0135] For matrix A in Formula 1, taking the first row of the matrix as an example, the meaning of its elements is shown in the table below.

[0136]

[0137] For matrix A, it consists of m row vectors, each representing a complete combination of end-to-end service routes. Let's take the first row vector of matrix A as an example. For example, the meaning of its elements is as follows: This is the base station ID information. to This represents the 5G transmission device ID information. The 'n' is included because there may be multiple nodes for the transmission device; here it represents the information of the transmission devices from the 1st to the (n-1th). This indicates the ID information of the 5G core network I-UPF; Represents the ID information of A-UPF. Uses row vectors. The complete expression for the end-to-end service route in line m is:

[0138] For formula 2, which is matrix B, its elements represent the row vectors within matrix A. The business IP address information, i.e. express The IP address information corresponding to the element. Its format is:

[0139] The integrated network management system sends matrix A and B information to the service routing module, which then reports receipt of the message back to the integrated network management system. The service routing module then forwards the matrix A and B information to the routing weight module, which in turn reports receipt of the message back to the service routing module.

[0140] Step 316: Obtain the robot's SUPI identifier information within the 5G network. The routing weight module sends a request message to the robot control area (main) via the service routing module to collect the SUPI information from the SIM card used by the remotely operated surgical robot. The robot control area (main) then feeds back the SUPI card information used to operate the surgical robot via the service routing module.

[0141] Step 317: Initiate the acquisition of network slice, DNN, and DNAI information of the surgical robot within the 5G network. The routing weight module initiates the acquisition of network slice, DNN, and DNAI information of the surgical robot within the 5G network through the service routing module to the integrated network management system.

[0142] Step 318: Obtain the network slice, DNN, and DNAI information of the SUPI code number used by the robot from the core network UDM. The integrated network management system sends a query message to the core network UDM to obtain the network slice, DNN, and DNAI information of the SUPI code number used by the robot. The core network UDM then feeds back the queried SUPI code number-signed network slice, DNN, and DNAI information to the integrated network management system.

[0143] Step 319: Receive the network slice, DNN, and DNAI information signed up using the SUPI code number used by the robot. The integrated network management system feeds back the network slice, DNN, and DNAI information signed up using the SUPI code number used by the robot to the routing weight module through the service routing module.

[0144] Step 320: Collect and process XDR call detail records from the signaling system. The routing weight module uses the service IP addresses of the radio network, core network, and transmission network in the received matrix B as indexes to collect XDR call detail records from the signaling system through the service routing module. The signaling system then feeds back the aforementioned XDR call detail records to the routing weight module through the service routing module.

[0145] Obtain each from matrix B (The service IP node in Formula 4) refers to the traffic, RTT latency, jitter, and packet loss rate of the nth service node under the m-th service route in terms of network slicing, DNN, and DNAI. Here, RTT latency, jitter, and packet loss rate represent network performance data, traffic represents service volume data, and network slicing, DNN, and DNAI represent the direction of service routing.

[0146] There are three tensor channels: the p channel for network slice classification, the q channel for DNN classification, and the o channel for DNAI classification.

[0147] Taking network slice classification of p-channel as an example, let , , , All dimensions are The tensors, respectively, represent This provides information on the traffic, RTT latency, jitter, and packet loss rate of the p-th network slice carried by the n-th service node under the m-th service route. Similarly, for the DNN classification q-channel, let... , , , For dimensions are all The tensors, respectively, represent The traffic, RTT latency, jitter, and packet loss rate of the q-th DNN carried by the n-th service node under the m-th service route. , , , For dimensions are all Zhang Liang, respectively expressed The traffic, RTT latency, jitter, and packet loss rate of the o-th DANI carried by the n-th service node under the m-th service route. For the above 12 tensors, the units representing traffic, RTT latency, jitter, and packet loss rate are bits / second, milliseconds, milliseconds, and percentages, respectively, all of which are numerical data.

[0148] For the p-channel of a network slice, the corresponding three tensors, after flattening, are: , , , The matrix is ​​then standardized. Traffic is a positive metric, while RTT latency, jitter, and packet loss rate are negative metrics. The standardization formula is as follows:

[0149] For the formula introduced in formula 5 and subsequent formulas and The function, given a two-dimensional input matrix, searches for the maximum or minimum value among all elements and outputs a two-dimensional matrix with the same dimensions as the input, where all elements are equal to a maximum or minimum value. The largest (corresponding to) (function) or minimum value (corresponding to) (Minimum value). Therefore and They respectively indicate that for All The values ​​of each element representing the flow are compared, and the minimum value is obtained through the comparison (corresponding to...). ) or maximum value ( ) to form a new A 3D matrix, where all elements of the new matrix have the same value. The minimum value among all elements (corresponding to) ) or maximum value ( )composition. dimensional matrix It means the relative standardized size of the traffic of the nth service node in the mth service route. The larger the value, the larger the traffic and the greater the service weight it bears, and vice versa.

[0150]

[0151] For formula 6, matrix It means that the RTT latency of the nth service node in the mth service route is relatively standardized. The smaller the RTT latency, the better the service indicators and perception, and the greater the weight. Conversely, the larger the RTT latency, the greater the weight.

[0152]

[0153]

[0154] Formulas 7 and 8 are similar to Formula 6. The meaning of the formula is the relative standardized size of the jitter and packet loss rate of the nth service node of the mth service route. The smaller the value, the smaller the jitter and packet loss rate, the better the service indicators and perception, and the greater the weight. Conversely, the larger the value, the smaller the weight. , , and The dimensions are all The value range is 0 to 1.

[0155] Introducing fusion weight column vector For the matrix , , and Perform weight constraint processing.

[0156]

[0157]

[0158] For formulas 9 and 10, representing column vectors There are 4 elements in total, all of which are hyperparameters, corresponding to , , and The corresponding weighting coefficients for traffic, RRT latency, jitter, and packet loss rate, with the value range of all four elements being [missing information]. And the sum is 1.

[0159] Introduction The formula for calculating the fusion weight of the nth node in the mth service route is as follows:

[0160] Its corresponding matrix form is as follows:

[0161] For formula 11, , , , The range of all elements in the matrix is Column vector The range of the four elements is As for formula 12, All elements of the matrix are derived using linearly normalized weighted summation as described in Formula 11; therefore, the range of all elements is also [value missing]. The higher the corresponding value, the higher the fusion weight value, and the higher the network environment for the teleoperated surgical robot to perform surgical operations.

[0162] All the above calculations are performed under the network slice channel. To distinguish between DNN and DNAI channels, Formula 12 is rewritten as follows:

[0163] For DNN and DNAI channels, the same algorithm from Equations 5 to 12 is applied, resulting in Equations 14 and 15:

[0164]

[0165] Because the matrices in formulas 13, 14, and 15 contain the weighted values ​​of network slices, DNNs, and DNAI channels contained in each service node under different service routing paths that may carry remote surgical services, the following is generated. 3D matrix :

[0166] Formula 16 consists of four column vectors. The first column vector represents the fusion weight value of the m-th service route in matrix B under the network slice dimension. Similarly, the second and third column vectors represent the fusion weight values ​​of the m-th service route under the DNN and DNAI dimensions. The fourth column vector is the sum of the fusion weight values ​​of the DNN, DNAI, and network slice dimensions, reflecting the overall network quality supporting the operation of the remote surgical robot. The specific numerical calculation method is as follows:

[0167]

[0168]

[0169]

[0170] For formulas 17, 18, and 19 , , This is obtained by summing the row vector elements in formulas 13, 14, and 15, with a value range of 0 to 1. Formula 20... It is the arithmetic mean of formulas 17, 18, and 19, with a value range of 0 to 1. The routing weight module sends the information of matrices A, B, and C to the network topology management module. The network topology management module then sends feedback to the routing weight module confirming receipt of the message.

[0171] Step 321: Generate a fusion weight matrix based on the XDR call detail records and matrices A, B, and C. The network topology management module generates the fusion weight matrix based on the received matrices B and C. The fusion matrix D has the following format:

[0172]

[0173] For matrix D in Formula 18, it is formed by merging matrices B and C, and it represents the business routing and the corresponding network slices, DNN, DNAI fusion weight values, and the fusion weight values ​​of the three in a row vector manner.

[0174] Step 322: The network topology management module selects the m-th service route to enable. The network topology management module initiates the process of enabling the m-th service route to the fusion algorithm module, and at the same time sends matrices A, B, C, and D to the fusion algorithm module for subsequent calculations.

[0175] Step 323: Generate and distribute the m-th service route configuration. The fusion algorithm module generates the m-th network configuration instruction (corresponding to row m in the matrix) based on the IMPI number of the teleoperated surgical robot, according to the network element list of the 5G wireless network, core network, and transmission network in the routing topology of matrix A, and distributes it to the integrated network management system. This means executing the m-th network configuration to form the m-th service route carrying the surgical operation service of the teleoperated surgical robot. The integrated network management system sends a message to the fusion algorithm module confirming receipt of the distributed configuration instruction.

[0176] Step 324: Send execution instructions to the 5G wireless network. The integrated network management system generates instructions to configure the network slice (column n+1 of matrix D) based on the network slice corresponding to the m-th service route (row m of matrix D). These instructions are then sent to the 5G wireless network base station. The 5G wireless network base station returns a message to the integrated network management system confirming receipt and execution of the instructions.

[0177] Step 325: Send execution instructions to the 5G transmission network. The integrated network management system generates instructions to configure the network slice (column (n+1) of matrix D) based on the network slice corresponding to the m-th service route (row m of matrix D). These instructions are then sent to the transmission equipment of the 5G transmission network. The 5G transmission network equipment returns a message to the integrated network management system confirming receipt and execution of the instructions.

[0178] Step 326: Issue execution instructions to the 5G core network. Based on the network slice, DNN, and DNAI (columns n+1, n+2, and n+3 of matrix D) corresponding to the m-th service route in matrix D, the integrated network management system generates and issues configuration instructions for the corresponding network slice, DNN, and DNAI to the AMF, SMF, PCF, and UPF devices of the 5G core network. The successful execution instructions returned by the AMF, SMF, PCF, and UPF devices of the 5G core network are then fed back to the integrated network management system.

[0179] Step 327: The wireless network, transmission network, and core network report that the instructions have been executed. After receiving all the instructions from the 5G wireless network, 5G transmission network, and 5G core network that have been executed, the integrated network management system reports a message to the fusion algorithm module that all instructions have been executed.

[0180] Step 328: Initiate the surgical robot motion programming test process. The fusion algorithm module sends the standard motions executed by the surgical robot's end effector to the robot's main control area through the business routing module. When calculating the force feedback model of each end effector of the master-slave teleoperated mobile robot using the virtual displacement principle, friction can be ignored. The force feedback of the end effector is a multi-rigid-body system with ideal constraints, meaning that the sum of the virtual work done by the end effector in any virtual displacement is equal to 0. According to the DH model, the generalized active force and virtual displacement of the joints and links of the end effector are set as j and k (j=1, 2, ..., n; k=1, 2, ..., i), respectively. and , and The projections in the rectangular coordinate system are respectively , , as well as , , And the sum of virtual work is 0, the expression is:

[0181] Set the joint angle of the end effector of the teleoperated surgical robot arm to be... The corresponding small displacement is Then consider the active output torque of the end effector joint of the remotely operated surgical robot arm. and its gravitational torque The sum of virtual work generated If the value is 0, the formula is:

[0182] For formula 23, It can be obtained through the homogeneous transformation method of the DH model, let The column vector represents the centroid coordinates of the k-th link of the surgical robot's end effector relative to the base coordinate system. Let be the centroid coordinates of the k-th link of the surgical robot's end effector relative to the entanglement coordinate system. Let the homogeneous transformation matrix of the entrainment coordinate system of the surgical robot's end effector relative to the base coordinate system be given. Then, the homogeneous transformation matrix of the end effector relative to the base coordinate system can be obtained as follows:

[0183] Substituting formula 24 into formula 23, we get... Considering that teleoperated surgical robots typically exhibit some deviation when carrying feedback services on different network paths, specifically the network quality during the statistical period when network slicing, DNN, and DNAI carry out surgical operations, the operation of the surgical robot's end effector is affected.

[0184] Step 329: Collect information on the remote execution status of the teleoperated surgical robot. The robot feeds back the coordinate data of the actual execution of the surgical robot's end effector from the operating area to the robot's main control area.

[0185] Step 330: The fusion algorithm module performs fusion calculations based on the received data. The robot's main control area feeds back the actual execution coordinates of the surgical robot's robotic arm end effector, along with the corresponding planned coordinate data, to the fusion algorithm module for calculation through the business routing module.

[0186] The fusion algorithm module performs the following calculations: When executing the m-th service route, This represents the norm of the difference between the actual value of the k-th link and the planned coordinate when the end effector of the teleoperated surgical robot arm performs the h-th action. The formula for expressing this is as follows:

[0187] For formula 25, It is the maximum coordinate value that can be executed when the k-th link of the robotic arm's end effector performs the h-th action. These are the actual coordinate values ​​of the k-th link of the robotic arm's end effector when it performs the h-th action. These are the planned coordinate values ​​when the k-th link of the robotic arm's end effector performs the h-th action. This is a second-order norm function. A larger numerator indicates a smaller deviation between the actual and planned coordinates under force feedback conditions, and vice versa. Dividing by the denominator normalizes the data for easier fusion.

[0188] Based on formula 25, introduce The function represents the output of all standardized actions. The arithmetic mean of is calculated using the following formula:

[0189] Formula 26 means that, under the condition of executing the m-th business route, the arithmetic mean of the second norm of the normalized standard deviations of all actual and planned positions under the force feedback effect generated by the execution of all standard robotic arm end effector actions, with a value range between 0 and 1.

[0190] Because the network topology management module stores matrix D Elements (corresponding to matrix C) (element), introduced function pairs and The data is integrated.

[0191]

[0192] For formula 27, The range of the numerical value is from 0 to 2. It consists of two parts, namely and , representing the impact of the corresponding network slice, DNN, and DNAI network path combination on force feedback when executing the m-th service route. The larger the value, the higher the network quality, and the better the surgical robot master controller perceives the force feedback execution of the end effector of the surgical arm, and vice versa.

[0193] Step 331: Feedback of the routing result for the m-th service. The fusion algorithm module will calculate... The numerical values ​​are fed back to the network topology management module.

[0194] Step 332: The fusion algorithm module selects the (m+1)th service route based on matrix A, repeats steps 322 to 331, and calculates... .

[0195] Step 333, Selection of Fusion Algorithm Module The service route corresponding to the largest value is the optimal service route. The optimal service route is executed by issuing the optimal service route configuration according to steps 323 to 327.

[0196] This embodiment proposes a service routing platform based on a cloud-based ARM virtual machine deployment scheme using the Ubuntu 16.04.1 LTS + ROSKinetic Kame framework, and an IP network connection method between various 5G network elements and the platform. The platform synchronously accesses the 5G core network, 5G transmission network, wireless 5G network, network management system, service signaling system, and the master control area and slave operation area of ​​a master-slave teleoperation surgical robot. It includes 9 external interfaces and 6 modules to implement optimal service routing based on the force feedback of the teleoperation surgical robot. This platform represents a novel and innovative platform architecture.

[0197] This embodiment also proposes an algorithm for fusion decision-making on optimal service routing based on network quality and robot force feedback deviation. Three tensor channels—network slice, DNN, and DNAI parameters—are established by collecting XDR call detail records from the service signaling system. A routing weight matrix is ​​generated by standardizing and weighting four indicators: traffic, RTT latency, jitter, and packet loss rate, and the normalized values ​​are obtained. Then, using the virtual displacement principle and the modulus of the DH model, the arithmetic mean of the second norm of the normalized standard deviations of all actual and planned coordinates under the force feedback effect generated by the force feedback of all standard robotic arm end effector actions is calculated for the m-th service route. The sum of these two values ​​is then used to evaluate the service route, and the service route corresponding to the highest value is the optimal service route. This algorithm is a novel fusion algorithm and is innovative.

[0198] This embodiment also proposes a master-slave remote surgical robot force feedback service routing selection process. This process enables the matrix-based collection and organization of 5G end-to-end network topology information, generating a topology matrix. Subsequently, it uses SUPI-based network switching, DNN, and DNAI subscription information queries, and combines this with UDM equipment to obtain subscription data, and uses XDR call detail records to generate service routes. Through round-robin routing configuration, force feedback motion testing of the remote surgical robot, calculation of the deviation between the actual and planned paths of the robotic arm end effector, and consideration of traffic, RTT latency, jitter, and packet loss rate, an end-to-end closed-loop process is established to select the optimal route. This process is a novel and innovative service routing selection process.

[0199] This embodiment also proposes a multi-dimensional fusion calculation model for network quality and action accuracy. This algorithm model achieves collaborative judgment of network performance and teleoperation action accuracy through a fusion algorithm, specifically including: establishing a fusion calculation logic based on the comprehensive network performance of three tensor channels (network slices, DNN, and DNAI) and the average action deviation of the teleoperation surgical robot. A weighted summation fusion method is used for service routing evaluation; the stronger the overall capability of the corresponding route to carry teleoperation surgery, the better. The optimal route determination criterion is as follows: using the maximization of the fusion value as the core criterion, all candidate service routes are traversed, and the route with the highest fusion value is selected as the optimal route. This criterion directly correlates network transmission quality with surgical operation accuracy, ensuring that the selected route simultaneously meets the network requirements of low latency and low packet loss, and the surgical operation requirements of high precision and strong consistency. This calculation model is a novel and innovative model.

[0200] The service routing processing method and apparatus provided in this application can solve the problem of force feedback service caused by network node influence during long-distance transmission. Existing technologies only implement force feedback data transmission based on wireless LANs, failing to consider the impact of packet loss, latency, and jitter of multiple network nodes on the force feedback effect during long-distance transmission, and also failing to address the problem of node error accumulation. This application's embodiments design a master-slave teleoperated surgical robot service routing selection platform, simultaneously accessing the 5G core network, transmission network, and wireless network. By collecting end-to-end network topology information, generating a routing matrix, and combining it with XDR call detail records to analyze the network performance of each node, it can select the optimal route, avoid error accumulation in multi-node transmission, and ensure the stability of force feedback under long-distance 5G transmission.

[0201] The service routing processing method and apparatus provided in this application can solve the problem of large latency in transmitting force feedback information in surgical robots. Existing technologies mostly use short-range transmission or designated service routing for packet forwarding under 5G networks, resulting in significant latency and causing deviations between force feedback perception and actual operation. This application's embodiments utilize a fusion decision algorithm based on network quality and force feedback deviation. It standardizes and weights indicators such as network slicing, DNN, DNAI traffic, RTT latency, jitter, and packet loss rate, and combines this with the deviation values ​​between the actual and planned coordinates of the robotic arm's end effector calculated based on the virtual displacement principle and the DH model to comprehensively evaluate the routing. Simultaneously, it allocates dedicated transmission channels for highly sensitive force feedback data through optimal routing, reducing signal transmission latency and working with the force feedback calculation module to correct action deviations, ensuring consistency in force feedback between the master and slave ends.

[0202] The service routing processing method and apparatus provided in this application can solve the problem of difficulty in matching network quality and action accuracy. Existing technologies do not link network quality and action accuracy; quality anomalies in dimensions such as network slicing, DNN, and DNAI can directly lead to force feedback operation anomalies. This application constructs a multi-dimensional fusion calculation model for network quality and action accuracy. Using the maximization of the fusion value as the criterion, it traverses all candidate routes and selects the optimal route after judging the fusion strategy that simultaneously meets the network requirements of low latency and low packet loss, as well as the deviation between the actual and planned force feedback coordinate values. This achieves dynamic matching of network quality and action accuracy, avoiding surgical operation anomalies caused by substandard network quality.

[0203] Furthermore, the service routing processing method and apparatus provided in this application embodiment possess adaptive and scalable capabilities. Implemented based on mobile communication networks (e.g., 5G), it can be extended to other communication network architectures. By traversing candidate routes and dynamically selecting the optimal path, it can adapt to changes in network status, exhibiting strong scenario adaptability and commercial application value. Specifically, based on a 5G focus, it can be extended to an NGN (5G + satellite) converged architecture. Deployed from a cloud-based ARM virtual machine, it extends to a three-tiered cloud-edge-device architecture, pushing force feedback calculation and motion orchestration modules down to the surgical operation area to reduce latency.

[0204] This application establishes a dual-mode acquisition mechanism of active detection and passive collection to avoid the risk of incomplete topology information. A new emergency handling submodule is added; for example, when the packet loss rate > 20% and latency > 100ms, it automatically switches to a backup route, simplifies the fusion algorithm, and notifies the force feedback module to activate a conservative operation mode, thus avoiding the risk of poor adaptability in extreme network environments. A dual closed-loop correction mechanism is constructed, establishing a friction force database. Before the operation, acquisition deviations are calibrated and parameters are corrected using standard procedures; during the operation, the model is fine-tuned in real time to ensure consistency between theory and practice, thus avoiding the risk of force feedback model deviation.

[0205] Corresponding to the service routing processing method provided in the above embodiments, based on the same technical concept, this application also provides an electronic device for executing the above method. Figure 7 To illustrate the structure of an electronic device according to various embodiments of this application, as shown in the following diagrams... Figure 7 As shown. Electronic device 500 can vary considerably due to differences in configuration or performance, and may include one or more processors 510 and memory 520. Memory 520 may store one or more application programs or data. Memory 520 may be temporary or persistent storage. The application programs stored in memory 520 may include one or more modules (not shown), each module may include a series of computer-executable instructions for the electronic device. Furthermore, processor 510 may be configured to communicate with memory 520 and execute the series of computer-executable instructions stored in memory 520 on the electronic device.

[0206] This application also provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium. The computer program includes program instructions, which, when executed by a computer, implement the steps of the service routing processing method described above.

[0207] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, apparatus, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0208] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems, devices), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0209] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0210] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0211] In a typical configuration, an electronic device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0212] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0213] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0214] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0215] Those skilled in the art will understand that embodiments of this application can be provided as methods, apparatus, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0216] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

[0217] It should be understood that the training and prediction processes of the artificial intelligence (AI) models involved in the various embodiments of this specification all adhere to multiple legal and compliant principles, including legal data sources, compliant data content, compliant data governance, compliant training objectives and schemes, compliant training processes, compliant training environments and tools, and compliant ethical verification of training results, and comply with the requirements of Article 5 of the Patent Law. Among them: Data source legitimacy: All datasets used for AI model training were obtained through legal means, covering three categories: publicly authorized data, data authorized by partners, and self-collected compliant data. Publicly authorized data comes from compliant data sources following open-source licenses such as Apache 2.0, with complete copyright attribution and authorization scope clearly marked, and no unauthorized open-source code or data reuse. Data authorized by partners has been subject to formal data usage agreements, clearly defining the scope, duration, and confidentiality obligations, and possessing a complete authorization chain. For self-collected data involving personal information, strict informed consent procedures have been followed, and anonymization processes (including but not limited to field masking, feature anonymization, and differential privacy technology applications) have been implemented to remove personally identifiable information, fully complying with the requirements of relevant laws and regulations such as the "Interim Measures for the Administration of Generative Artificial Intelligence Services" and the "Personal Information Protection Law."

[0218] Data content compliance: The AI ​​model's dataset undergoes multiple screenings and cleaning processes to remove all content that may violate social morality or harm public interests. It contains no obscene, pornographic, violent, discriminatory, or information that endangers national or public safety, nor does it involve the illegal acquisition or use of genetic resources. For data in sensitive fields (such as healthcare and finance), an additional privacy-preserving computation module (including federated learning and secure multi-party computation technologies) ensures that the data is "usable but not visible," avoiding compliance risks during the original data transmission process and ensuring that the data application scenarios and uses comply with public order and good morals and industry regulatory requirements.

[0219] Data governance norms: A complete data traceability system is established during the AI ​​model training process to automatically record the source, collection time, annotation process, cleaning rules, and permission allocation of training data, generating traceable compliance reports to ensure that the data is verifiable throughout its entire lifecycle. The dataset annotation process for AI models is completed by a professional human R&D team, clearly defining the proportion of human creative contributions and avoiding reliance on AI-generated data that has not undergone substantial human modification, thus meeting the examination requirements for "human main contributions" in AI patent applications.

[0220] Training objectives and plans are compliant: The training objective of the AI ​​model focuses on [specific technical scenarios that can be supplemented, such as intelligent driving decision optimization, multimodal information interaction, etc., and replaced based on specific content]. The training scheme and the final output results do not violate any mandatory provisions of laws and administrative regulations, do not harm the public interest or the legitimate rights and interests of others, and do not pose any potential risks of being used for illegal activities, infringing on privacy, or disrupting public safety. The model strictly adheres to the ethical principle of "intelligent for good".

[0221] Training process compliance: A closed-loop training framework is adopted to ensure compliance and controllability of the training process. The specific process is as follows: First, training samples are obtained through compliant data sources. After the aforementioned data cleaning and desensitization, they are input into the neural network model to generate preliminary training results. Second, an expert system is introduced to verify the preliminary results. Based on preset rules and human expert experience, the feasibility of the results is evaluated, and outputs that may pose ethical risks or compliance hazards are corrected (such as removing decision-making logic that violates public order and good morals, and adjusting model parameters that do not comply with safety regulations). Finally, the loss function weights are dynamically optimized based on expert system feedback to strengthen the model's learning of compliant results, avoid overfitting errors or non-compliant labels, and form a closed-loop control of "data input - model training - expert verification - parameter optimization - result feedback" to ensure that the entire training process complies with A5 ethical review requirements.

[0222] Training environment and tool compliance: AI model training is implemented using nationally licensed chips and a compliant training platform. All open-source frameworks and components used in the training process have obtained their corresponding licenses, and copyright statements and patent citation information are fully retained, with no instances of infringement or reuse. The training environment is built using virtual devices (containers / virtual machines) with fixed random seeds and initial parameter configurations to ensure the reproducibility of the training process. Furthermore, through access control and operation log recording, risks such as data leakage and parameter tampering during training are prevented, ensuring the security and compliance of the training process.

[0223] Training results ethical verification compliance: After the model is trained, it undergoes additional third-party ethical compliance assessment and algorithm filing review to verify that the model output does not violate social morality or harm public interests. For potentially sensitive scenarios (such as public services and intelligent decision-making), a special result verification mechanism is established to ensure that the model always complies with Article 5 of the Patent Law and relevant laws and regulations in practical applications.

[0224] In summary, the data and training process used in the AI ​​model of this specification strictly comply with the relevant provisions of Article 5 of the Patent Law and the Patent Examination Guidelines (2023 Edition), and there are no violations of laws, social ethics, public interests, or illegal use of genetic resources. It fully meets the compliance requirements for patent authorization.

[0225] Definitions of terms and abbreviations

Claims

1. A method for processing service routing, characterized in that, The method includes the following steps: Collect end-to-end network topology information of the mobile communication network, and generate a first matrix and a second matrix of candidate service routes. The first matrix includes device identifiers, and the second matrix includes network addresses. Obtain the subscription information and network performance data of the remote operation device, calculate the network quality score of each candidate service route based on the subscription information and network performance data, and generate a third matrix; Based on the second and third matrices, the fusion matrix is ​​obtained; For each candidate service route, network configuration is issued according to the fusion matrix. Standard action tests are performed under the configured network and force feedback deviation is calculated to obtain the action deviation score of the candidate service route. The comprehensive score is calculated based on the action deviation score of the candidate service route and the corresponding network quality score. The candidate service route with the highest overall score is selected as the optimal service route.

2. The method according to claim 1, characterized in that, The process of collecting end-to-end network topology information of the mobile communication network and generating a first matrix and a second matrix of candidate service routes includes the following steps: According to the topology collection instructions, collect the device identifiers and network addresses of each network element from the access network, transmission network and core network of the mobile communication network; Based on the collected device identifiers and network addresses, generate the first matrix and the second matrix; In this matrix, each row of the first matrix corresponds to a device identifier sequence of a candidate service route, and the corresponding row in the second matrix corresponds to the network address sequence of the candidate service route.

3. The method according to claim 1, characterized in that, The steps of obtaining the subscription information and network performance data of the remote operation device, calculating the network quality score of each candidate service route based on the subscription information and network performance data, and generating a third matrix include the following: Obtain the user's permanent identifier for the remote operation device, and query the unified data management network element of the core network for the network slice, data network name, and data network access identifier subscribed to by the user's permanent identifier; Using the network addresses in the second matrix as indexes, collect detailed data records and extract network performance data of each node in each candidate service route based on the network slice, data network name, and data network access identifier; The extracted network performance data is standardized, and the standardized network performance data is then weighted and fused to calculate the fusion weight for each node. Based on the fusion weights of all nodes on each candidate service route, the average network fusion weight of the candidate service route is calculated, and the network quality score of the candidate service route is obtained based on the average network fusion weight, thus generating the third matrix.

4. The method according to claim 3, characterized in that, The step of calculating the average network fusion weight of a candidate service route based on the fusion weights of all nodes on each candidate service route, and obtaining the network quality score of the candidate service route based on the average network fusion weight, includes the following steps: The average network fusion weight is obtained by calculating the arithmetic mean of the fusion weights of all nodes on each candidate service route. Based on the network slice, data network name, and data network access identifier, the arithmetic mean of the average network fusion weight is calculated to obtain the network quality score.

5. The method according to claim 1, characterized in that, The process of obtaining the fusion matrix based on the second and third matrices further includes the following steps: The second matrix and the third matrix are merged to generate the fusion matrix; Each row of the fusion matrix includes a network address sequence of a candidate service route, as well as the average network fusion weight and network quality score of the candidate service route under the dimensions of network slice, data network name, and data network access identifier.

6. The method according to claim 1, characterized in that, The step of issuing network configuration based on the fusion matrix for each candidate service route includes the following steps: Obtain the device identifier sequence of the current candidate service route from the first matrix; Based on the network slice, data network name, and data network access identifier corresponding to the fusion matrix, a configuration instruction is generated; The configuration instructions are sent to the access network, transmission network, and core network of the mobile communication network.

7. The method according to claim 1, characterized in that, The step of performing standard action tests and calculating force feedback deviations under the configured network to obtain the action deviation score for the candidate service route includes the following steps: Send standard action commands based on the principle of virtual displacement to the main control area of ​​the remote operation device; Receive the actual coordinates of the robotic arm end effector fed back from the operating area by the remote operating device; Based on the deviation between the actual coordinates and the planned coordinates, the normalized mean action deviation is calculated and used as the action deviation score for the candidate service route.

8. A service routing processing apparatus, characterized in that, The device includes the following: The information collection module is used to collect end-to-end network topology information of the mobile communication network and generate a first matrix and a second matrix of candidate service routes. The first matrix includes device identifiers and the second matrix includes network addresses. The quality scoring module is used to obtain the subscription information and network performance data of the remote operation device, calculate the network quality score of each candidate service route based on the subscription information and network performance data, and generate a third matrix. The fusion processing module is used to obtain a fusion matrix based on the second matrix and the third matrix; as well as, The routing module is used to execute network configuration distribution based on the fusion matrix for each candidate service route, perform standard action tests and calculate force feedback deviation under the configured network to obtain the action deviation score of the candidate service route, and calculate its comprehensive score based on the action deviation score of the candidate service route and the corresponding network quality score. The candidate service route with the highest overall score is selected as the optimal service route.

9. An electronic device, characterized in that, It includes a processor, a memory, a program or instructions stored in the memory and executable on the processor, wherein the program or instructions, when executed by the processor, implement the steps of the method as described in any one of claims 1 to 7.

10. A computer program product, characterized in that, The computer program product includes a computer program stored on a non-transitory computer-readable storage medium, the computer program including program instructions that, when executed by a computer, implement the steps of the method as described in any one of claims 1 to 7.

Images