A video transmission and hierarchical early warning method based on 5G and Wi-Fi hybrid networking

Abstract

The application discloses a video transmission and hierarchical early warning method based on 5G and Wi-Fi hybrid networking, and relates to the technical field of network transmission and safety monitoring. The application comprises the following steps: firstly, deploying a fusion anchor point containing three types of core hardware modules and determining the number; secondly, extracting the service time delay and bandwidth requirement grading under the XR scene, establishing the QoS parameter mapping relationship of 5G and Wi-Fi, and dynamically allocating bandwidth by reading the key parameters of the fusion anchor point; thirdly, determining the sampling interval and marking the sampling frame based on the parameter matrix and the weight matrix; and finally, evaluating the threat parameters of the sampling frame and executing the hierarchical response. The application improves the accuracy of threat evaluation while improving the synergy of hybrid networking, and guarantees the continuity and safety of video transmission under the XR scene.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of network transmission and security monitoring, in particular to a video transmission and hierarchical early warning method based on 5G and Wi-Fi hybrid networking. BACKGROUND

[0002] In XR and other scenarios with high requirements for real-time performance, bandwidth and transmission security, 5G and Wi-Fi hybrid networking gradually becomes an important solution for video transmission due to its advantages of wide coverage and high bandwidth. However, the existing video transmission and hierarchical early warning method based on this hybrid networking still has the following defects:

[0003] Firstly, in the existing solution, 5G and Wi-Fi are mostly deployed independently, lacking integrated hardware carriers and unified resource scheduling mechanisms, and when the terminal switches between the two links, it is prone to high protocol conversion delay and resource sharing problems;

[0004] Secondly, the existing method does not establish a precise mapping relationship between the QoS parameters of 5G and Wi-Fi, making it difficult to match the resource scheduling priorities and delay protection capabilities of the two links. After switching links for high real-time and high-bandwidth services, the priority is likely to be downgraded, affecting transmission reliability;

[0005] Thirdly, the existing bandwidth allocation is mostly based on fixed strategies and does not dynamically adjust based on core parameters such as service delay requirements, link interference levels, signal strengths and terminal position deviations, resulting in low resource utilization;

[0006] Fourthly, the existing threat assessment mostly relies on fixed threshold judgments and does not build dynamic baselines based on historical normal data, resulting in high misjudgment rates. Moreover, the hierarchical response is not coordinated with link resources, making it difficult to deal with different levels of threats and unable to effectively guarantee the security and continuity of video transmission;

[0007] Therefore, there is an urgent need for a video transmission and hierarchical early warning method based on 5G and Wi-Fi hybrid networking. SUMMARY

[0008] To overcome the shortcomings of the prior art, the present application provides a video transmission and hierarchical early warning method based on 5G and Wi-Fi hybrid networking, which solves the problems of poor coordination of hybrid networking, insufficient bandwidth allocation and threat disposal.

[0009] To achieve the above purpose, the present application realizes the following technical scheme: a video transmission and hierarchical early warning method based on 5G and Wi-Fi hybrid networking, comprising:

[0010] Step 1: Deploy fusion anchor points in the XR scene coverage area, each fusion anchor point is built-in with three types of core hardware modules, and the number of anchor points is determined according to the anchor point spacing;

[0011] Step 2, extract the end-to-end delay T and peak bandwidth demand B of the service under the XR scene, divide them into two grades respectively, and establish the QoS parameter mapping relationship of 5G and Wi-Fi;

[0012] Step 3, read the service delay demand, interference degree, signal strength and position deviation in the XR scene in real time through the fusion anchor point, and dynamically allocate bandwidth;

[0013] Step 4, determine the sampling interval K based on the core parameter matrix M and the weight matrix W, and mark the sampling frame, the dimension of M is 3*4, wherein three rows represent three consecutive time windows, and four columns represent four types of core parameters, the dimension of W is 1*4, and the sum of all elements is 1;

[0014] Step 5, threat parameter evaluation is performed on the received sampling frame, and hierarchical response is executed based on the evaluation result.

[0015] As a further scheme of the application, the three types of core hardware modules include a 5G NR baseband module, a Wi-Fi 6E / 7 radio frequency module and a protocol conversion core chip.

[0016] As a further scheme of the application, the specific steps for determining the number of anchor points according to the anchor point spacing are as follows:

[0017] By means of a laser range finder or a CAD scene drawing, the key vertex coordinates of the XR scene boundary are collected to generate a minimum circumscribed polygon of the scene;

[0018] The minimum circumscribed polygon is divided into uniform square grids, the grid side length is Rmin / 2, and each grid is marked as an effective grid or an invalid grid, and the total number of effective grids Ngrid is counted, wherein the effective grid represents that the space occupied by the grid is completely within the scene, and the invalid grid represents that the space occupied by the grid is partially or completely outside the scene;

[0019] The measured 5G coverage radius of the fusion anchor point in the current scene is R1, the Wi-Fi coverage radius is R2, and the single-anchor dual-link coverage radius is Rmin, and the single-anchor physical coverage area is calculated as , wherein Rmin=min(R1,R2);

[0020] The coverage ranges of adjacent anchor points need to overlap, and the overlap rate is set as , and the effective coverage area considering the overlap rate is calculated as ;

[0021] The effective area of the scene is calculated as , and the preliminary number of anchor points is obtained as , wherein represents rounding up;

[0022] The triangular grid layout is adopted, that is, adjacent anchor points are connected to form a triangle, Ninitial anchor points are evenly placed in the minimum outer polygon, and the distance between adjacent anchor points is The position and size of the fixed obstacle in the scene are collected, and are marked as an occlusion area.

[0023] The ray method is used to detect the blind area: for each effective grid center, four rays are emitted, including up, down, left and right, if at least one ray can reach a certain anchor point without occlusion, that is, the ray does not pass through the occlusion area, then the grid has no blind area; if all rays are blocked, the grid is marked as a blind area grid;

[0024] If there is a blind area grid, an anchor point is added near the blind area, and the blind area needs to be detected again for each added anchor point, until all effective grids have no blind area, and finally the number of anchor points Nfinal is determined.

[0025] As a further scheme of the application, the specific operation of dividing T and B into two grades is as follows:

[0026] If , it is divided into a high real-time grade; if , it is divided into a normal real-time grade.

[0027] If , it is divided into a high bandwidth grade; if , it is divided into a normal bandwidth grade.

[0028] Wherein, Tth and Bth are the delay threshold and bandwidth threshold respectively.

[0029] As a further scheme of the application, the specific mapping rule of the QoS parameters of 5G and Wi-Fi is as follows:

[0030] If it is a high real-time grade + a high bandwidth grade, it is mapped to 5QI=5+AC_VO; if it is other demand combinations, it is mapped to 5QI=9+AC_VI.

[0031] The other demand combinations include a high real-time grade + a normal bandwidth grade, a normal real-time grade + a high bandwidth grade, and a normal real-time grade + a normal bandwidth grade.

[0032] As a further scheme of the application, the specific steps of dynamically allocating bandwidth are as follows:

[0033] Four key parameters and value ranges are determined: delay requirement Td∈[Tmin,Tmax], interference degree Gr∈[0,Gmax], signal strength Sg∈[Smin,Smax], and position deviation Xq∈[0,Xmax], and an ideal reference sequence is constructed ;

[0034] Based on the terminal actual parameters and ideal sequence, the absolute difference of each terminal parameter is calculated , the minimum value and the maximum value of all are counted , the resolution coefficient is set , the correlation coefficient of each terminal four parameters is calculated according to the formula ;

[0035] The four correlation coefficients of the terminal are used as the characteristics to construct a matrix, each terminal is initially set as an independent cluster, and the Euclidean distance between clusters is calculated, and the clusters with the minimum distance are repeatedly merged until three clusters are formed, and the average correlation coefficient of each cluster is calculated ;

[0036] For each cluster, the correlation entropy H(Td), H(Gr), H(Sg), H(Xq) of the four parameters in the cluster is calculated according to the formula , and the entropy weight w(Td), w(Gr), w(Sg), w(Xq) of each parameter is calculated according to the formula ;

[0037] Then, the resource coefficient of the three clusters is calculated according to the formula ;

[0038] The total sum of the resource coefficients of the three clusters is counted, the initial bandwidth of each cluster is allocated according to the formula group bandwidth = total bandwidth × (group resource coefficient / resource coefficient total sum), and the secondary bandwidth allocation of the terminals in each cluster is completed according to the formula terminal bandwidth = (group bandwidth / group terminal number) × (terminal / cluster );

[0039] Every 5 minutes, the above steps are re-executed to collect new parameters, update correlation coefficients, correlation groups, resource coefficients and bandwidth allocation, and realize dynamic adjustment.

[0040] As a further scheme of the application, each element in M is the average value of the parameters in the window, and each element in M is standardized according to , wherein is the maximum value of the jth column parameter in the three windows.

[0041] As a further scheme of the application, the comprehensive decision value of each time window is calculated , and the average value of the three windows is taken as the final decision value Davg: if Davg≥Dth1, K takes 3; if Dth2≤Davg

[0042] ​​As a further scheme of the present application, for the sampling frame collected per K frames, the value index is calculated according to the formula If , the corresponding sampling frame is marked as a high-value frame and is returned through a 5G link, otherwise, it is marked as a normal frame and is returned through a Wi-Fi link, wherein is a service priority weight, is a normalized service priority of a time window t to which the current sampling frame belongs, is a real-time complexity of the current sampling frame, is a value index threshold.

[0043] As a further scheme of the present application, the specific steps for threat parameter evaluation of the sampling frame are as follows:

[0044] Kiu frames of normal sampling frames in the same XR scene in the past Miu days are selected, and the transmission dimension and the content dimension are focused, wherein Miu≥30 and Kiu≥ ;

[0045] In the transmission dimension, the content complexity Chist of each frame is calculated, and the specific formula is: Chist=number of changed pixels / total number of pixels; the transmission time Thist of each frame is calculated; and a linear regression algorithm is used to fit the correlation curve of the two, and the specific formula is: Tfit=k×Chist+b, wherein Tfit is the predicted transmission time, k and b are fitting coefficients; and the confidence interval C is marked.

[0046] In the content dimension, two adjacent frames are taken in time sequence, the SHA-256 check value Hamming distance of each pair of adjacent frames is calculated, and then divided by the total number of bits to obtain a series of check value difference rates Vhist; the Vhist sequence is processed by sliding with a window size of 100 frames, and the 2.5% quantile and the 97.5% quantile of Vhist in each window are calculated; the minimum value of the 2.5% quantiles of all windows is taken as Vlow, and the maximum value of the 97.5% quantiles of all windows is taken as Vhigh, to form a baseline band [Vlow, Vhigh];

[0047] The transmission time Treal, the content complexity Creal, and the check value variation Vreal are collected in real time.

[0048] Creal is substituted into the above correlation curve to obtain Tfit, the transmission deviation rate Rdev=(Treal-Tfit) / Tfit is calculated, and if |Rdev|>C%, it is determined that there is a transmission abnormality suspicion;

[0049] When Vreal [Vlow, Vhigh], it is determined that there is a content abnormality suspicion.

[0050] If neither of the two dimensions is marked, it is classified as no threat; if one of the dimensions is marked, it is classified as moderate threat; and if both dimensions are marked, it is classified as severe threat.

[0051] The application provides a video transmission and hierarchical early warning method based on 5G and Wi-Fi hybrid networking.

[0052] (1) The fusion anchor point with the built-in 5G NR baseband, Wi-Fi 6E / 7 radio frequency and protocol conversion core chip is deployed, the physical fragmentation of the independent deployment of 5G and Wi-Fi is broken, the real-time conversion of the two-link protocols and the unified scheduling of resources are realized, the protocol delay and resource fault problem when the terminal switches the link are avoided, and the fluency and continuity of video transmission in the XR scenario are effectively guaranteed.

[0053] (2) The application divides the delay and bandwidth requirements of the business in the XR scenario into grades, establishes the QoS parameter equivalent mapping relationship of 5G and Wi-Fi, ensures that the business priority is not downgraded after the link switching, and combines the dynamic calculation of the resource coefficient to allocate the bandwidth according to the multi-core parameter, thereby improving the network resource utilization efficiency.

[0054] (3) The application constructs the dynamic baseline of the transmission and content dimensions based on the historical normal data, and performs differential response according to the threat level linkage of the link resources, so that the transmission abnormality and content tampering and other problems can be targetedly disposed, and the safety and business stability of the video transmission in the XR scenario are effectively guaranteed. BRIEF DESCRIPTION OF DRAWINGS

[0055] Figure 1 The step flowchart is for the application;

[0056] Figure 2 The step flowchart is for determining the number of anchor points;

[0057] Figure 3 The step flowchart is for dynamically allocating bandwidth. DETAILED DESCRIPTION

[0058] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of the application.

[0059] Embodiment 1

[0060] As Figure 1 , the application provides a video transmission and hierarchical early warning method based on 5G and Wi-Fi hybrid networking, which comprises:

[0061] Step 1, deploy fusion anchors in the XR scene coverage area, each anchor has three types of core hardware modules built-in, and determine the number of anchors according to the anchor spacing;

[0062] In traditional XR scenarios, 5G base stations and Wi-Fi are deployed independently: 5G is responsible for wide coverage but low bandwidth scheduling flexibility, Wi-Fi is responsible for high bandwidth but limited coverage, for terminals, either connect to 5G or Wi-Fi, cannot share resources across links;

[0063] The core role of the fusion anchor is to break this "physical split", integrate 5G and Wi-Fi as a collaborative link through integrated hardware, allowing terminals to seamlessly switch between the two networks and unified resource scheduling;

[0064] The three types of core hardware modules include 5G NR baseband module, Wi-Fi 6E / 7 radio frequency module, and protocol conversion core chip:

[0065] 5G NR baseband module, responsible for accessing 5G network, supports Sub-6GHz or millimeter wave frequency selection;

[0066] Sub-6GHz frequency band has strong diffraction ability and low wall penetration loss, suitable for multi-obstruction scenarios such as XR operating rooms with lead shielding doors, ensuring stable 5G signal coverage;

[0067] Millimeter wave frequency band has large bandwidth but high penetration loss, suitable for open scenarios such as e-sports halls and large exhibition halls, meeting the high bandwidth requirements of XR image quality;

[0068] Wi-Fi 6E / 7 radio frequency module, including 60GHz millimeter wave sub-module, supporting 2.4GHz / 5GHz / 60GHz three frequency band switching, responsible for accessing Wi-Fi network;

[0069] Three frequency band switching is to balance coverage and bandwidth: 2.4GHz frequency band has wide coverage but low bandwidth, suitable for edge area terminals; 5GHz frequency band has medium bandwidth, suitable for ordinary areas; 60GHz millimeter wave sub-module has high bandwidth but small coverage radius, suitable for high bandwidth demand in core interaction areas such as e-sports battle stations and main operating areas of surgery;

[0070] Protocol conversion core chip, responsible for real-time conversion of 5G and Wi-Fi protocols, high main frequency ensures conversion latency ≤10ms, if using low-performance chips, conversion latency may reach 50ms, causing XR image lag;

[0071] 5G and Wi-Fi have significant differences in coverage characteristics: 5G has a larger coverage radius R1, but the bandwidth is greatly affected by the number of users; Wi-Fi has a smaller coverage radius R2, but the single link bandwidth is extremely high, suitable for small range high-density terminals;

[0072] If only R1 is deployed, there will be a large number of coverage blind spots for Wi-Fi. For example, if R1 is 50 meters and R2 is 15 meters, the Wi-Fi signal in the middle area of the two anchor points will be weak when deployed at a distance of 50 meters. If only R2 is deployed, it will result in too many anchor points and waste of 5G resources.

[0073] Step 2, extract the end-to-end delay T and peak bandwidth demand B of the XR service, and classify them into two grades respectively, and establish the QoS parameter mapping relationship between 5G and Wi-Fi;

[0074] The specific reasons for preferentially selecting end-to-end delay T and peak bandwidth demand B as core indicators, rather than other indicators such as packet loss rate and jitter, are as follows:

[0075] The core of XR service is real-time interaction and spatial immersion. A delay of more than 30 ms will cause a perceptual fault in the human eye and motion, and the delay indicator directly determines whether the service can be implemented. For example, a delay of more than 20 ms in XR remote surgery may result in operational errors.

[0076] Most XR pictures are 3D modeling or ultra-high-definition videos, and the data volume of a single frame is 3-5 times that of traditional videos. Insufficient bandwidth will cause excessive compression of pictures and texture blurring.

[0077] Combining delay and bandwidth can cover most XR scenario requirements, and other indicators can be indirectly guaranteed through these two core indicators. For example, high-priority delay / bandwidth requirements will trigger "packet loss retransmission optimization" at the network bottom layer, without the need for additional separate definitions.

[0078] Two grades are adopted, and the core is to balance the coverage breadth and conversion complexity: multiple grades will increase the calculation pressure of terminal reporting and anchor point conversion, and two grades can cover most scenarios while reserving an extension interface. The specific grading conditions are as follows:

[0079] If , it is divided into a high real-time grade, such as XR remote surgery and XR multi-person e-sports.

[0080] If , it is divided into a normal real-time grade, such as XR virtual exhibition roaming and XR live goods delivery.

[0081] If , it is divided into a high bandwidth grade, such as XR 8K surgery picture transmission and XR virtual exhibition roaming.

[0082] If , it is divided into a normal bandwidth grade, such as XR standard-definition remote guidance and XR lightweight virtual tour.

[0083] Among them, Tth and Bth are the delay threshold and bandwidth threshold respectively, which need to be set according to actual conditions.

[0084] The XR terminal reads the T and B values of the current scene from the service configuration file at the start of the service, and sends them to the fusion anchor point through a lightweight reporting protocol (such as MQTT) with a reporting delay ≤5ms to avoid affecting the start of the service;

[0085] After the anchor point receives the indicators, it compares them with the preset scene-indicator threshold table. If the terminal reporting value exceeds the reasonable range, it is automatically corrected to the default value for that scene to avoid conversion errors caused by abnormal terminal reporting;

[0086] The core of establishing the QoS parameter mapping relationship between 5G and Wi-Fi is the equalization of cross-link QoS guarantee strength, i.e., the QoS parameters (5QI) of 5G and the AC level of Wi-Fi need to be consistent in resource scheduling priority, delay guarantee capability, and bandwidth allocation weight to avoid high-priority services becoming low-priority after terminal link switching. The specific basis is as follows:

[0087] 5G QoS parameters are pre-set to 5QI=5 and 5QI=9:

[0088] If 5QI=5, it is defined as "guaranteed bit rate GBR service with low latency and high reliability", resource scheduling weight ≥70%, end-to-end latency commitment ≤30ms, and it adapts to the high real-time and high-bandwidth needs of XR;

[0089] If 5QI=9, it is defined as "GBR service with medium latency and medium reliability", resource scheduling weight is 30%-70%, end-to-end latency commitment ≤100ms, and it adapts to the non-highest priority needs of XR;

[0090] Wi-Fi AC level is pre-set to AC_VO and AC_VI:

[0091] If it is AC_VO, it represents the highest priority queue, scheduling weight ≥70%, priority transmission of voice and video data, end-to-end latency ≤30ms, and it is completely equal to the guarantee strength of 5QI=5;

[0092] If it is AC_VI, it represents a medium-high priority queue, scheduling weight is 30%-70%, mainly transmits video data, end-to-end latency ≤100ms, and it is equal to the guarantee strength of 5QI=9;

[0093] According to the T and B, the specific mapping rules are set as follows:

[0094] If it is high real-time + high bandwidth, it is mapped to 5QI=5+AC_VO, and the main application scenarios include XR remote surgery, XR industrial robot control, etc. This kind of scene needs the highest level of resource guarantee, and the GBR service of 5QI=5 can lock the bandwidth to avoid being occupied by other services, and the AC_VO queue can ensure priority transmission in the Wi-Fi link, even if multiple devices are concurrent, high demand terminals can still preempt resources;

[0095] If it is other demand combination, it is mapped to 5QI=9+AC_VI;

[0096] The other demand combination includes three types of scenes, which are all mapped to the same mapping, and the core reason is "no need to occupy the highest resource, medium and high priority can meet the needs", as follows:

[0097] If it is high real-time + ordinary bandwidth, such as XR remote text guidance, although low latency is required, the bandwidth demand is low, and the medium and high priority of 5QI=9 can avoid occupying the high bandwidth resource of 5QI=5 on the basis of guaranteeing the delay;

[0098] If it is ordinary real-time + high bandwidth, such as XR 4K on-demand, high bandwidth demand can be locked by GBR service of 5QI=9, but it does not need the highest scheduling priority (low delay requirement), avoiding affecting high real-time scenarios;

[0099] If it is ordinary real-time + ordinary bandwidth, such as XR 2K education on-demand, medium and high priority can fully meet the needs, without occupying higher resources, saving network overhead;

[0100] When the XR service demand changes dynamically, the terminal reports the new T and B value to the anchor point in real time, and the anchor point re-matches the mapping relationship according to the new index, such as a scene from "ordinary real-time + ordinary bandwidth" to "ordinary real-time + high bandwidth", still maintaining 5QI=9+AC_VI, without switching to 5QI=5, if the service changes from "ordinary real-time" to "high real-time", it is automatically upgraded to 5QI=5+AC_VO.

[0101] Step 3, read four key parameters in the XR scene in real time through the fusion anchor point, and dynamically allocate bandwidth;

[0102] The four key parameters specifically include business delay demand, interference degree, signal strength, and position deviation.

[0103] Step 4, determine the sampling frequency based on the core parameter matrix M and the weight matrix W, and mark the sampling frame;

[0104] The dimension of the core parameter matrix M is 3x4, wherein the three rows represent three consecutive time windows, and the four columns represent four types of core parameters;

[0105] The time window needs to be combined with the frame frequency characteristics of the XR service. The length of a single time window can be set as Tw in, and the value interval is [1s, 2s], and 1.5s is preferred. Three consecutive windows are selected mainly to filter abnormal values through multi-window comparison. At the same time, consecutive windows can reflect the time trend of the parameters.

[0106] Each element M[i,j] (i-th window, j-th parameter) in the matrix M is the average value of the parameters in the window. Since each time window contains multiple frames of data, the parameter value will change with the frame. Taking the average value can reflect the true resource status in the window and avoid misjudgment caused by extreme values of a single frame.

[0107] The four types of core parameters include terminal allocation bandwidth, video frame complexity, service priority, and link real-time quality.

[0108] After constructing the core parameter matrix M, each element in the matrix M needs to be standardized to eliminate the dimension effect. Since the above four types of core parameters are positive parameters, the specific standardization formula is: wherein, is the maximum value of the j-th column parameter in the three windows.

[0109] In order to adapt to the dimension of the core parameter matrix M, a weight matrix W is constructed , which satisfies ;

[0110] The initial weight can be set as , , , :

[0111] Corresponding to the terminal allocation bandwidth, the weight is the highest. The core constraint of the sampling frequency is "whether the bandwidth is supported". Even if the frame complexity is high and the service priority is high, if the bandwidth is insufficient, high sampling frequency will cause link congestion.

[0112] Corresponding to the video frame complexity, the weight is the second highest. Because the complexity is directly related to the "information value" of the frame, if the high complexity frame is not sampled enough, key details will be lost, affecting the decision.

[0113] Corresponding to the service priority, the weight is medium. Because the priority determines the "tilt direction of sampling resources", under the same bandwidth, high priority is given priority to high sampling frequency.

[0114] Corresponding to the link real-time quality, the weight is relatively low. Because the link quality more affects the "transmission reliability after sampling", rather than the sampling frequency itself, when the link is poor, it can be solved by link switching, rather than adjusting the sampling frequency preferentially.

[0115] Calculate the comprehensive decision value of each time window , and take the average of the three windows as the final decision value Davg;

[0116] Determine the sampling interval K according to Davg: if Davg≥Dth1, then K takes 3; if Dth2≤Davg<Dth1, then K takes 4; if Davg<Dth2, then K takes 5; wherein Dth1 and Dth2 are the upper and lower limits of the final decision value, respectively;

[0117] For the sampling frame collected every K frames, calculate its value index ;

[0118] Wherein, is the service priority weight, and the service priority is the core guide of the frame value. Even if the frame complexity is slightly lower, as long as the priority is high, it should be classified as a high-value frame to avoid misjudging key service frames as ordinary frames; is the normalized service priority of the time window t to which the current sampling frame belongs; is the real-time complexity of the current sampling frame;

[0119] If , the corresponding sampling frame is marked as a high-value frame and is returned through the 5G link, otherwise, it is marked as an ordinary frame and is returned through the Wi-Fi link.

[0120] Step 5, perform threat parameter evaluation on the received sampling frame, and execute a hierarchical response based on the evaluation result;

[0121] The core value of XR service lies in the "digital restoration of real scene", and the sampling frame is the core carrier of service interaction. If it is maliciously tampered with, such as replacing the picture content and implanting false information, it may directly lead to service errors or even safety accidents. Threat parameter evaluation can accurately identify tampering behavior by checking the matching of data volume and transmission efficiency and the continuity of frame content;

[0122] The specific steps for threat parameter evaluation of the sampling frame are as follows:

[0123] Select Kiu frames of normal sampling frames within Miu days of the same XR scene, wherein Miu≥30 and Kiu≥ ;

[0124] Focus on two core correlation dimensions: transmission dimension, reflecting the "matching of data volume and transmission efficiency"; content dimension, reflecting the "continuity of frame content";

[0125] In the transmission dimension, the content complexity Chist of each frame is calculated, which can be represented by the "intra-frame pixel change rate": Chist=number of changed pixels / total number of pixels; the transmission time Thist of each frame (the time difference between the terminal sending and the server receiving) is calculated; the correlation curve between the two is fitted by a linear regression algorithm, and the specific formula is: Tfit=k x Chist+b, where Tfit is the predicted transmission time, k and b are fitting coefficients, and the goodness of fit is required to ensure the reliability of the correlation , and the C% confidence interval is marked, with a value of 95% being preferred;

[0126] In the content dimension, the SHA-256 checksum Hamming distance of each pair of adjacent frames is calculated, and then divided by the total number of bits to obtain a series of checksum difference rates Vhist; a window size of 100 frames is used to slide the Vhist sequence, and the 2.5% quantile and the 97.5% quantile of Vhist in each window are calculated; the minimum value of the 2.5% quantile of all windows is taken as Vlow, and the maximum value of the 97.5% quantile of all windows is taken as Vhigh, forming the baseline band [Vlow, Vhigh];

[0127] Real-time collection of transmission time Treal, content complexity Creal, and checksum variation Vreal;

[0128] Substitute Creal into the above correlation curve to obtain Tfit, calculate the transmission deviation rate Rdev=(Treal-Tfit) / Tfit, and if |Rdev|>C%, it is determined to be a transmission abnormality suspect;

[0129] When Vreal [Vlow, Vhigh], it is determined to be a content abnormality suspect;

[0130] If neither dimension is marked, it is classified as non-threatening and no additional operations are performed;

[0131] If one of the dimensions is marked, it is classified as a moderate threat and the following measures are taken: ① Merge anchor points to trigger link switching, such as switching from Wi-Fi to 5G or from 5G to Wi-Fi, and select the link with better current quality; ② The server sends a frame retransmission request to the terminal to obtain a backup copy of the abnormal frame;

[0132] If both dimensions are marked, it is classified as a severe threat and the following measures are taken: ① The server sends a threat tracing report to the terminal management, including abnormal frame screenshots, terminal location, Rdev / Vreal values, and link logs; ② Preserve the abnormal frame evidence and store it in an encrypted database for subsequent security analysis.

[0133] Embodiment 2

[0134] This embodiment continues to disclose a method for determining the number of anchor points based on embodiment 1, as shown, the specific content includes: Figure 2

[0135] Collect the key vertex coordinates of the XR scene boundary through a laser range finder or a CAD scene drawing, and generate the minimum circumscribed polygon of the scene. The minimum circumscribed polygon is the simplest polygon that can completely wrap the scene, so as to avoid missing edge areas.

[0136] Divide the minimum circumscribed polygon into a uniform square grid, the grid side length is Rmin / 2, and label each grid as a valid grid or an invalid grid, and count the total number of valid grids Ngrid.

[0137] The valid grid represents that the space occupied by the grid is completely within the scene, and the invalid grid represents that the space occupied by the grid is partially or completely outside the scene.

[0138] The actual measurement fusion anchor point has a 5G coverage radius R1 and a Wi-Fi coverage radius R2 in the current scene, obtains a single anchor point dual-link coverage radius Rmin, and calculates a single anchor point physical coverage area , wherein Rmin=min(R1, R2).

[0139] To avoid blind areas, the coverage ranges of adjacent anchor points need to overlap, and the overlap rate is set to , and the value interval is generally [20%, 30%].

[0140] The effective coverage area with the overlap rate is calculated as .

[0141] The scene effective area is calculated as , and the preliminary anchor point number is obtained as , wherein represents rounding up to avoid coverage deficiency due to decimals.

[0142] A regular triangle grid layout (the connecting lines of adjacent anchor points form a regular triangle, and the coverage is most uniform) is adopted, Ninitial anchor points are uniformly placed in the minimum circumscribed polygon, and the distance between adjacent anchor points is ensured to be .

[0143] The positions and sizes of fixed obstacles in the scene are collected and marked as shielding areas.

[0144] The ray method is used to detect blind areas: for each valid grid center, four rays (up, down, left, and right) are emitted. If at least one ray can reach a certain anchor point without being blocked (i.e., the ray does not pass through the shielding area), the grid has no blind area. If all rays are blocked, the grid is marked as a blind area grid.

[0145] ​If there is a blind area grid, then supplement the anchor point near the blind area, every supplement a station, need to re-detect blind area, until all the effective grid without blind area, the final determination of anchor point number Nfinal.

[0146] Embodiment 3

[0147] This embodiment continues to disclose a method for dynamically allocating bandwidth based on embodiment 1 and embodiment 2, as shown in the figure, the specific process is as follows: Figure 3

[0148] Determine four key parameters and value range: time delay demand Td∈[Tmin, Tmax], interference degree Gr∈[0, Gmax], signal strength Sg∈[Smin, Smax], position deviation Xq∈[0, Xmax], and construct ideal reference sequence

[0149] Based on the actual parameters of the terminal and the ideal sequence, calculate the absolute difference of each terminal parameter , statistics of all The minimum value and the maximum value , set the resolution coefficient , calculate the correlation coefficient of each terminal four parameters according to the formula ;

[0150] The value interval of the correlation coefficient is [0, 1], and 0.5 is preferred to balance the resolution ability and stability. If it is too small, it will lead to too large difference in correlation degree, which is easy to be affected by abnormal value. If it is too large, it will lead to too small difference in correlation degree, which cannot distinguish the priority of terminal; Take the four correlation coefficients of the terminal as the characteristics to construct the matrix, initially set each terminal as an independent cluster, and repeatedly merge the cluster with the smallest distance by calculating the Euclidean distance between clusters, until three clusters are formed, and the average correlation coefficient of each cluster is calculated

[0151] ;

[0152] For each cluster, calculate the correlation entropy H(Td), H(Gr), H(Sg), H(Xq) of the four parameters respectively according to the formula , and calculate the entropy weight w(Td), w(Gr), w(Sg), w(Xq) of each parameter according to the formula ;

[0153] Correlation entropy is used to calculate the dispersion degree of "certain parameter correlation coefficient" in a cluster. The smaller the dispersion degree (the smaller the entropy), the more consistent the correlation degree of the parameter of the terminal in the cluster, and the higher the reference value of the parameter to resource allocation (the greater the weight);

[0154] According to the formula​​​ The resource coefficients of the three clusters are calculated;

[0155] If the resources are allocated only according to the problem of "intra-group parameter confusion" is ignored, for example, the high correlation cluster of 0.9, but ∈[0.4, 0.9), if the high resources are allocated only according to , some terminals will still be stuck due to the weak Sg, and the present scheme reduces the weight of Sg through entropy weight, which will make the resource coefficient appropriately down-regulate to avoid waste;

[0156] The total sum of the resource coefficients of the three clusters is counted, the initial bandwidth of each cluster is allocated according to group bandwidth = total bandwidth × (group resource coefficient / resource coefficient total sum), and in each cluster, the secondary bandwidth allocation of the terminals in the group is completed according to terminal bandwidth = (group bandwidth / group terminal number) × (terminal / cluster );

[0157] Every 5 minutes, the above steps are re-executed to collect new parameters, update the correlation coefficient, the correlation group, the resource coefficient and the bandwidth allocation, and realize dynamic adjustment.

[0158] Some data in the above formula are dimensionless numerical calculations, and the contents not described in detail in the specification all belong to the prior art known to those skilled in the art.

[0159] The above examples are only used to illustrate the technical method of the present application and are not limiting, and although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present application can be modified or replaced equivalently without departing from the spirit and scope of the technical method of the present application.

Claims

1. A method for video transmission and hierarchical early warning based on 5G and Wi-Fi hybrid networking, characterized in that, The application comprises the following steps: Step 1, deploying fusion anchor points in the XR scene coverage area, each fusion anchor point is internally provided with three types of core hardware modules, and the number of anchor points is determined according to the anchor point spacing; Step 2, extracting the end-to-end delay T and peak bandwidth demand B of the service in the XR scene, and dividing them into two grades respectively, and establishing the QoS parameter mapping relationship between 5G and Wi-Fi; Step 3, real-time reading the service delay demand, interference degree, signal strength and position deviation in the XR scene through the fusion anchor point, and dynamically allocating bandwidth; Step 4, determining the sampling interval K based on the core parameter matrix M and the weight matrix W, and marking the sampling frame, the dimension of M is 3x4, wherein the three rows represent three consecutive time windows, and the four columns represent four types of core parameters, the dimension of W is 1x4, the sum of all elements is 1, and the four types of core parameters include terminal allocated bandwidth, video frame complexity, service priority and link real-time quality; Step 5, threat parameter evaluation is performed on the received sampling frame, and hierarchical response is executed based on the evaluation result.

2. The video transmission and hierarchical early warning method based on 5G and Wi-Fi hybrid networking according to claim 1, characterized in that, The three types of core hardware modules include 5G NR baseband module, Wi-Fi 6E / 7 radio frequency module and protocol conversion core chip. 3.The video transmission and hierarchical early warning method based on 5G and Wi-Fi hybrid networking of claim 1, wherein, The specific steps for determining the number of anchor points according to the anchor point spacing are as follows: Collect the key vertex coordinates of the XR scene boundary through a laser range finder or CAD scene drawing, and generate the minimum circumscribed polygon of the scene; Divide the minimum circumscribed polygon into uniform square grids, the grid side length is Rmin / 2, and each grid is marked as an effective grid or an invalid grid, and the total number of effective grids Ngrid is counted, wherein the effective grid represents that the space occupied by the grid is completely within the scene, and the invalid grid represents that the space occupied by the grid is partially or completely outside the scene; The measured fusion anchor point in the current scene is obtained, the 5G coverage radius is R1, the Wi-Fi coverage radius is R2, and the single-anchor dual-link coverage radius is Rmin, and the single-anchor physical coverage area is calculated as wherein Rmin = min(R1, R2). The coverage of adjacent anchor points needs to overlap, and the overlap rate is set as , the effective coverage area considering the overlap rate is calculated ; The effective area of the scene is calculated as The preliminary number of anchor points is obtained as wherein represents rounding up. The triangular grid layout is adopted, that is, the adjacent anchor point connecting lines form a triangle, Ninitial anchor points are uniformly placed in the minimum circumscribed polygon, and the distance between adjacent anchor points is The position and size of the fixed obstacle in the scene are collected and marked as an occlusion region. Detect the blind area by using the ray method: for the center of each effective grid, emit four rays, including up, down, left and right, if at least one ray can reach a certain anchor point without obstruction, that is, the ray does not pass through the shielding area, then the grid has no blind area; if all rays are blocked, mark it as a blind area grid; If there is a blind area grid, supplement anchor points near the blind area, and re-detect the blind area for each additional anchor point until all effective grids have no blind area, and finally determine the number of anchor points Nfinal.

4. The video transmission and hierarchical early warning method based on 5G and Wi-Fi hybrid networking according to claim 1, characterized in that, The specific operation of dividing T and B into two grades is as follows: If , then it is classified as high real-time; if , then it is classified as normal real-time. If , then it is classified into the high bandwidth category; if , then it is classified into the normal bandwidth category. Wherein, Tth and Bth are the delay threshold and bandwidth threshold respectively.

5. The video transmission and hierarchical early warning method based on 5G and Wi-Fi hybrid networking according to claim 1, characterized in that, The specific mapping rules of 5G and Wi-Fi QoS parameters are as follows: If it is high real-time grade + high bandwidth grade, it is mapped to 5QI=5+AC_VO; if it is other demand combination, it is mapped to 5QI=9+AC_VI; The other demand combinations include high real-time grade + ordinary bandwidth grade, ordinary real-time grade + high bandwidth grade, and ordinary real-time grade + ordinary bandwidth grade.

6. The video transmission and hierarchical early warning method based on 5G and Wi-Fi hybrid networking according to claim 1, characterized in that, Each element in M is the average of the parameters in the window, while Each element in M is normalized, where, is the maximum value of the jth column parameter in the three windows.

7. The video transmission and hierarchical early warning method based on 5G and Wi-Fi hybrid networking according to claim 1, characterized in that, Calculate the integrated decision value for each time window and take the average of the three windows as the final decision value Davg: if Davg≥Dth1, then K takes 3; if Dth2≤Davg<Dth1, then K takes 4; if Davg<Dth2, then K takes 5; where Dth1 and Dth2 are the upper and lower limits of the final decision value, respectively.

8. The video transmission and hierarchical early warning method based on 5G and Wi-Fi hybrid networking according to claim 7, characterized in that, For the sampling frame collected per K frames, its value index is calculated according to the formula If , the corresponding sampling frame is marked as a high-value frame and is returned through a 5G link, otherwise, it is marked as a normal frame and is returned through a Wi-Fi link, wherein, is a service priority weight, is a normalized service priority of a time window t to which the current sampling frame belongs, is a real-time complexity of the current sampling frame, is a value index threshold. 9.The method of claim 1, wherein, The specific steps of threat parameter evaluation on the sampling frame are as follows: Select Kiu frames of normal sampling frames in the past Miu days of the same XR scene, and focus on the transmission dimension and the content dimension, wherein Miu≥30, Kiu≥ ; In the transmission dimension, the content complexity Chist of each frame is calculated, specifically: Chist=number of changed pixels / total number of pixels; the transmission time consumption Thist of each frame is calculated; and a linear regression algorithm is used to fit the correlation curve, specifically: Tfit=k×Chist+b, wherein Tfit is the predicted transmission time consumption, and k and b are fitting coefficients; and a confidence interval C is marked; In the content dimension, two adjacent frames are taken in chronological order, the SHA-256 check value Hamming distance of each pair of adjacent frames is calculated, and then divided by the total number of bits to obtain a series of check value difference rates Vhist; a window size of 100 frames is used to slide the Vhist sequence, and the 2.5% quantile and the 97.5% quantile of Vhist in each window are calculated; the minimum value of the 2.5% quantile of all windows is taken as Vlow, and the maximum value of the 97.5% quantile of all windows is taken as Vhigh, to form a baseline band [Vlow, Vhigh]; The transmission time consumption Treal, the content complexity Creal and the check value variability Vreal are collected in real time; Creal is substituted into the above correlation curve to obtain Tfit, the transmission deviation rate Rdev=(Treal-Tfit) / Tfit is calculated, and if |Rdev|>C%, it is determined that there is a transmission abnormality suspicion; When Vreal [Vlow, Vhigh], it is determined that the content is suspicious. If neither of the two dimensions is marked, it is classified as non-threatening; if one of the two dimensions is marked, it is classified as moderate threat; and if both of the two dimensions are marked, it is classified as severe threat.

Images