A Distributed Video Proxy Routing Method and System

By employing a distributed video proxy routing method, combined with end-to-end quality assessment and multi-dimensional node health, the problem of unstable routing in wide area network video surveillance systems was solved, achieving stability and load balancing in video stream forwarding and improving user experience.

CN123123244APending Publication Date: 2026-11-17CHINA TOWER CO LTD XIANGTAN BRANCH +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611471201.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-09-21
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing video surveillance systems in wide area network environments suffer from problems such as inflexible static routing, video stuttering caused by hard switching of single indicators, limited resource evaluation dimensions, and susceptibility to cascading failures. Existing solutions cannot comprehensively evaluate the quality of the entire link and the computing power of nodes, resulting in unstable video stream forwarding.

Method used

A distributed video proxy routing method is adopted, which realizes intelligent routing decision through end-to-end quality assessment and multi-dimensional node health. Combined with a weighted random mechanism to avoid single-point overheating, a distributed architecture is constructed to realize automatic optimization of video stream forwarding paths and traffic balancing.

Benefits of technology

It achieves stable video stream forwarding in a wide area network environment, avoids cascading failures caused by single points of failure, has traffic load balancing capabilities, and improves user experience.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123123244A_ABST
    Figure CN123123244A_ABST
Patent Text Reader

Abstract

This invention discloses a distributed video proxy routing method and system, belonging to the fields of video surveillance and edge computing. The system includes access nodes, egress nodes, and relay nodes, with at most one relay between the access and egress points. When a user requests a video stream, the quality score of the direct connection is first calculated as a benchmark. If the direct connection deteriorates (i.e., the benchmark value is lower than a preset deterioration threshold), a comprehensive score is generated for each candidate relay based on its resource utilization, health, remaining capacity, and the quality of the two links. After filtering the candidate set, if the gain of the optimal score relative to the direct connection benchmark reaches a threshold, the target relay is selected and the proxy link is switched to transmit, with higher gain carrying greater weight; otherwise, the direct connection is maintained. This invention solves the problems of blind switching and ignoring node computing power and capacity, and possesses anti-avalanche and load balancing capabilities.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of video surveillance and edge computing technology, and particularly relates to a dynamic proxy routing and traffic scheduling method for camera video streams in a wide area network environment. Background Technology

[0002] Currently, in large-scale video surveillance systems, cameras are distributed across local city and county data centers (access nodes), while user access points are typically concentrated in provincial or central data centers (exit nodes). Due to the complexity of the wide area network environment, high latency and high packet loss often occur between access nodes and exit nodes.

[0003] Existing routing and scheduling schemes mainly suffer from the following problems: Static routing lacks flexibility: it typically uses DNS or configuration files to specify fixed forwarding paths. Once the physical link deteriorates, it cannot dynamically switch, causing video stuttering.

[0004] Single-metric hard handover: Existing technologies often force a specific relay node to forward packets when the direct connection latency or packet loss exceeds a threshold. This method ignores the relay node's own load and the superimposed losses of the "access → relay" and "relay → exit" links, often resulting in a "the more you hop, the more laggy it gets" phenomenon.

[0005] Resource assessment is limited to a single dimension: existing load balancing algorithms mostly focus on CPU or memory, while video stream forwarding is a typical network and I / O intensive service. Disk I / O, network card throughput, and concurrent stream capacity are also bottlenecks that determine forwarding capabilities. Existing solutions lack a unified computing power and capacity assessment model.

[0006] It can easily lead to cascading failures: Traditional solutions often select the "optimal" node for centralized forwarding. Once this node fails due to excessive load or network jitter, all traffic will be interrupted or transferred to the suboptimal node instantly, which can easily cause a service avalanche.

[0007] Therefore, there is an urgent need for an intelligent routing method that can comprehensively evaluate the quality of the entire link and the computing power of nodes, and has the ability to distribute traffic. Summary of the Invention

[0008] To address the aforementioned technical issues, this invention provides a distributed video proxy routing method and system. By comparing the overall benefits of direct connection and relay, and combining the multidimensional load of nodes with the quality of dual-segment links, the optimal path selection is achieved, and a weighted random mechanism is used to avoid single-point overheating.

[0009] This invention achieves automated optimization and traffic balancing of video stream forwarding paths by constructing a dual-core intelligent routing mechanism of "full-link quality assessment + node multi-dimensional health".

[0010] The system adopts a distributed architecture, consisting of access nodes, exit nodes, and relay nodes distributed in different cities. The nodes work together to form a complete routing decision closed loop, and a single node can be responsible for access, relay, and exit simultaneously.

[0011] Specifically, this invention collects network link quality parameters and node load parameters in real time, calculates the network quality score of the direct link between the access node and the egress node as the direct link benchmark value; when the quality of the direct link is lower than a preset degradation threshold, for each candidate relay node, its CPU load, memory utilization, disk I / O, network I / O, flow capacity load, and the latency, packet loss, and jitter of the two links are comprehensively evaluated to obtain a one-hop end-to-end comprehensive score; the top three nodes with the highest scores are selected as the candidate set. If the gain ratio of the optimal one-hop score to the direct link score exceeds a preset threshold, a target relay node is selected in the candidate set in a weighted random manner to construct a proxy link of "access node → relay node → egress node"; otherwise, the direct link is maintained.

[0012] Therefore, this invention realizes a complete closed loop from link detection to node evaluation to routing decision, solving the problem that the traditional solution blindly jumps based on the quality of a single network segment and ignores the computing power and capacity loss of nodes, resulting in "the experience after jumping is not as good as the direct connection". At the same time, it has the ability to prevent avalanche and load balancing.

[0013] The anti-avalanche mechanism of the present invention is achieved through the following three-fold design: (1) weighted random election distributes traffic to multiple candidate nodes to avoid a single node from bearing the load; (2) the capacity factor is linearly reduced when the node's flow carrying capacity exceeds 60% of the rated carrying capacity, so that the traffic is diverted to other nodes in advance; (3) the candidate set retains the top 30% (at least 3) of nodes, so that when the optimal node fails, other candidate nodes can quickly take over and avoid complete service interruption.

[0014] According to one aspect of the present invention, a distributed video proxy routing method is provided, comprising the following steps: The system collects network link quality parameters, node load parameters, and node attribute parameters in real time. The network link quality parameters include the quality parameters of the direct link between the access node and the egress node, as well as the quality parameters of the segmented links between the access node, egress node, and each candidate relay node. The node load parameters include the resource usage parameters and current service capacity of the candidate relay nodes. The node attribute parameters include the rated capacity of the candidate relay nodes. The connection between the access node and the egress node passes through at most one relay node. Based on the quality parameters of the direct link, the network quality score of the direct link is calculated as the direct link benchmark value. When the direct link benchmark value is lower than the preset degradation threshold, for each candidate relay node: its available processing capacity is evaluated based on the resource occupancy parameters of the candidate relay node as a health factor; its remaining capacity is evaluated based on the remaining ratio of the current service carrying capacity of the candidate relay node to the rated carrying capacity as a capacity factor; and the overall transmission quality after the two links are connected is evaluated based on the segmented link quality parameters between the candidate relay node and the access node, and the segmented link quality parameters between the candidate relay node and the egress node as a link quality factor. A comprehensive score for the candidate relay nodes is generated based on the health factor, the capacity factor, and the link quality factor; the comprehensive scores of each candidate relay node are then filtered to form a candidate set. If the gain of the best comprehensive score in the candidate set relative to the direct connection benchmark value is less than a preset gain threshold, then the relay jump is rejected and transmission is maintained via the direct connection link; If the gain level is greater than or equal to a preset gain threshold, then selection weights are assigned to each candidate relay node in the candidate set in such a way that the higher the gain level, the greater the weight is assigned, and the target relay node is selected from the candidate set based on the selection weights. Establish a proxy link connecting the access node and the exit node via the target relay node, and switch the video stream from the direct link to the proxy link for transmission.

[0015] In another aspect, the present invention provides a distributed video proxy routing system, comprising: Access nodes, distributed across different cities, are used to connect to cameras of various brands and provide raw video streams; The exit nodes are distributed in different cities and are used to provide a unified gateway domain name to receive user requests. Relay nodes, distributed across different cities, are used to forward video streams between access nodes and egress nodes; One node is responsible for at least one of the functions of access, relay, and exit. The access node, the exit node, and the relay node are used together to execute the above-described distributed video proxy routing method.

[0016] Compared with the prior art, the present invention has the following beneficial effects: End-to-end quality assessment: comprehensively considers the quality of the two links from access to exit, as well as the computing power and capacity of relay nodes, to avoid blind spots in single-point assessment.

[0017] Gain decision mechanism: Relay is only activated when the overall quality improvement after the jump exceeds a preset ratio, thus avoiding invalid forwarding.

[0018] Traffic load balancing: By using Top-K candidates and weighted random allocation, traffic is distributed to reduce the risk of single point of failure.

[0019] Soft rate limiting protection: Automatically reduces the weight when the node's flow capacity approaches the upper limit, realizing soft rate limiting and early flow diversion. Attached Figure Description

[0020] 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.

[0021] Figure 1 This is a schematic diagram of the distributed video gateway network architecture provided in an embodiment of the present invention, illustrating the network topology relationship and user access path between access nodes, egress nodes, and relay nodes; Figure 2 This is a flowchart of the full-link decision-making process of the distributed video proxy routing method provided in this embodiment of the invention, which shows the complete decision-making process from indicator collection, direct connection benchmark calculation, degradation triggering, relay scoring, Top-K pruning, gain decision to weighted random election. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] The embodiments of this invention are written in a progressive manner.

[0024] To facilitate understanding of the technical solution of this invention, some technical terms appearing in this specification are explained as follows: Access node: refers to the edge node deployed in the local data center, used to connect to local cameras and provide raw video stream interfaces (such as RTSP / RTMP).

[0025] Egress node: refers to the central node deployed in the provincial or central computer room, which is used to provide a unified gateway domain name to receive user requests and provide streaming media exit.

[0026] Relay node: refers to a relay node that provides video stream forwarding services between access nodes and egress nodes. It can be deployed reused with access nodes or deployed independently.

[0027] Direct link: refers to a direct transmission link between the access node and the exit node without any intermediate proxy node.

[0028] Proxy link: refers to the video stream transmission path from the access node to the exit node via a relay node, i.e., "access node → relay node → exit node".

[0029] Candidate relay nodes: refer to all available relay nodes that the system can select as intermediate nodes for proxy links within the current routing decision cycle.

[0030] Rated capacity: refers to the maximum number of video stream channels that a relay node can simultaneously support under a preset resource threshold, which is obtained through dynamic calibration via benchmark stress testing.

[0031] Concurrent stream count: refers to the number of video stream channels that a relay node forwards simultaneously.

[0032] Health factor: A quantitative indicator used to characterize the current available processing capacity of a relay node, calculated based on CPU load rate, memory utilization rate, disk I / O utilization rate, and network I / O utilization rate.

[0033] Capacity factor: A quantitative indicator used to characterize the current remaining flow carrying capacity of a relay node, calculated based on the remaining proportion of the current service carrying capacity relative to the rated carrying capacity.

[0034] Link quality factor: A quantitative indicator used to characterize the overall transmission quality after the two links from the access node to the exit node are connected via the relay node. It is calculated based on the combined delay, packet loss rate and jitter of the two links.

[0035] Overall score: A quantitative indicator used to characterize the overall suitability of candidate relay nodes as intermediate nodes of proxy links. It is obtained by weighted multiplication of health factor, capacity factor and link quality factor.

[0036] Gain level: refers to the improvement ratio of the candidate relay node's one-hop end-to-end comprehensive score relative to the direct connection benchmark value, used to determine whether relay switching has a net benefit.

[0037] The terms used above are only for explaining specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Those skilled in the art will understand that other existing technologies with the same or similar functions can be employed without departing from the concept of the present invention.

[0038] Example 1: System Architecture like Figure 1As shown, this invention provides a distributed video proxy routing system. Its core lies in establishing an intelligent, multi-dimensional evaluation relay selection mechanism between access nodes and egress nodes to achieve automated optimization of video stream transmission paths. The system adopts a distributed architecture, consisting of access nodes, egress nodes, and relay nodes distributed in different cities. A single node can simultaneously handle access, relay, and egress.

[0039] Specifically, the system includes the following functional nodes: Access nodes are distributed across city-level data centers. Each access node connects to a local camera and provides an RTSP / RTMP raw stream interface. Access nodes are responsible for collecting local network link quality parameters and node load parameters, and reporting them to the system's decision center. Access nodes and egress nodes continuously send ICMP Ping packets and UDP probe packets (1400 bytes per packet, analog video stream MTU) via a lightweight agent to probe network quality. The sampling period is 5 seconds, the reporting period is 1 minute, and the data retention window is 15 minutes.

[0040] Egress nodes: Distributed in provincial or central data centers across different cities, these nodes provide a unified gateway domain name to receive user requests and offer streaming media egress. When a user requests a video stream from a camera via the egress node's domain name, the egress node locates its corresponding access node based on the camera's encoding, triggering a routing decision process.

[0041] Relay nodes: Distributed across different cities, these nodes forward video streams between access and egress nodes. Relay nodes can be deployed reused or independently as access nodes, and they possess stream relay capabilities. The entire system is limited to a maximum of one intermediate proxy node, meaning only a single hop ("access → relay → egress") is allowed, and multiple hops ("access → relay A → relay B → egress") are not permitted to avoid latency aggregation and amplified link jitter.

[0042] Node load collection mechanism: The system obtains the following load metrics through the proc file system (Linux) or performance counters (Windows) of the node operating system: CPU load rate C (calculated from 1-idle_time / total_time in / proc / stat); memory utilization rate M (calculated from (total memory - available memory) / total memory); disk IO utilization rate D (read from / proc / diskstats, calculate the percentage of sector read / write operations per unit time relative to the disk's rated IOPS); network IO utilization rate N (read from the network interface traffic, calculate the ratio of current throughput to the physical bandwidth limit).

[0043] Rated capacity calibration mechanism: When a new node is launched or its configuration is changed, the system starts an automated load testing tool to simulate mixed video stream access at different resolutions (720P, 1080P, 4K). The number of concurrent streams is gradually increased until the node's CPU load rate reaches 80%, memory utilization reaches 80%, or network throughput reaches 80% of the physical bandwidth. The number of concurrent streams at this point is the maximum rated capacity of the node. During the load test, mixed video streams of 720P, 1080P, and 4K resolutions are used, with the mixing ratio set to 3:5:2 (720P:1080P:4K) according to the current network service distribution to simulate the node load characteristics under real business scenarios. Implementers can adjust this ratio according to the actual service distribution.

[0044] The nodes communicate with each other through well-defined APIs to ensure the scalability and stability of the system.

[0045] Example 2: General Process like Figure 2 As shown, this invention provides a distributed video proxy routing method, comprising the following steps: Indicator data collection mechanism The system collects network link quality parameters and node load parameters in real time, with a sampling period of 5 seconds, a reporting period of 1 minute, and a data retention window of 15 minutes.

[0046] Network quality detection (latency L, packet loss rate P, jitter J): Between access nodes and egress nodes, and between access nodes and candidate relay nodes, lightweight agents continuously send ICMP Ping packets and UDP probe packets (packet size 1400 bytes, simulating video stream MTU). Latency L is the average value within a sliding window; packet loss rate P is the ratio of lost packets to total packets sent within a period; jitter J is calculated using the RTP jitter algorithm defined in RFC 3550.

[0047] Node load data (CPU load rate C, memory utilization M, disk I / O utilization D, network I / O utilization N): obtained through the node operating system's proc file system (Linux) or performance counters (Windows).

[0048] Direct connection benchmark score calculation A user requests a video stream from a camera via the domain name of the exit node, and the exit node locates the access node to which the camera belongs based on the camera's code.

[0049] Define the normalized calculation formula for each network metric of the link, with a minimum score of 0: ; ; ; Define the link quality calculation formula, and use the above score weighting to calculate the baseline network quality score B for direct connection (access node → exit node): ; Degradation trigger judgment If the direct connection baseline score is below 50, the relay election process will begin; otherwise, the direct connection address will be returned directly to maintain direct transmission.

[0050] The preset degradation threshold is set at 50 points (out of 100). Testing showed that when the quality score of the direct link falls below 50, the retransmission rate of 1080P video streams encoded in H.264 exceeds 15%, and the frame rate drops to below 60% of the original frame rate, resulting in a significant decline in user experience. Setting this threshold ensures that relay election is only initiated when the direct link truly cannot meet basic transmission requirements, avoiding unnecessary system overhead.

[0051] Multidimensional scoring of relay nodes For each candidate relay node, calculate: (a) Node overall load health factor (Four-dimensional weighted): ; Where C is the CPU load rate, M is the memory utilization rate, D is the disk I / O utilization rate, and N is the network I / O utilization rate.

[0052] (b) Capacity Load Factor : when hour, ; when hour, ; when hour, ; Where Cur represents the current number of flows carried, and Cap represents the node's rated capacity; when the load is below 60%, the node's load factor is not considered; when the load is above 60%, the score decreases linearly. The node will be eliminated at that time.

[0053] (c) Quality factor Q of the two-segment link: Since the network link includes both incoming and outgoing links after the introduction of relay nodes, the relevant network latency metrics need to be calculated by merging parameters before being substituted into the link quality calculation formula: The delay is the sum of the link delay and: ; The packet loss rate is calculated by multiplying the probability: ; Jitter is represented by RSS (Root Sum Square). ; The total link delay obtained Packet loss rate and shaking Substituting the values ​​into the link quality calculation formula, you can obtain the total link network quality score. .

[0054] (d) Overall score for one-jump end-to-end jump : ; in Preset basic weights manually (the default is 1, and node scheduling weights can be manually configured).

[0055] Top-K pruning Sort all candidate nodes in descending order of S, and retain only the top 30% (at least 3) to enter the final candidate set.

[0056] If the total number of candidate relay nodes is less than 10, the first 3 are retained; if the total number is more than 10, the first 30% are retained (rounded up) to ensure that the candidate set contains at least 3 nodes, so as to guarantee the diversity of subsequent weighted random election.

[0057] Gain decision Calculate the ratio of the optimal single-hop composite score to the score of the directly connected network: ; Set gain threshold : The gain threshold G is set at 1.2, requiring the overall score after relay hops to be at least 20% higher than the direct connection baseline. This threshold is based on the following considerations: introducing relay nodes adds an extra hop of forwarding latency and node processing overhead; if the improvement is less than 20%, the improvement in user experience is insufficient to offset the added system complexity and link failure risk. Actual testing has verified that when the gain ratio is below 1.2, the user's subjective perception of improved video playback smoothness is not significant.

[0058] like If the redirection is deemed highly profitable, the system enters the traffic allocation phase, selecting nodes from the Top-K candidate nodes according to... The percentage is used as a weight for weighted random selection. The larger the Δ, the higher the probability of being selected, thus avoiding traffic concentration on a single relay node while ensuring link quality.

[0059] like If the gain is less than 20%, it is determined that the benefit of the redirect cannot offset the cost of the additional forwarding, so the redirect is abandoned and the direct connection is maintained.

[0060] Weighted random election The system obtains the candidate set and their weights Then, a random election is conducted: (1) Calculate the total weights ; (2) Generate a random number rand in the interval [0,1); (3) Traverse the candidate set and accumulate the weights. ; (4) When When selecting a node, choose the current node.

[0061] Taking two candidate nodes as an example, if node A has a weight of 0.48 and node B has a weight of 0.36, then the total weight is 0.84. The probability of node A being selected is 0.48 / 0.84≈57%, and the probability of node B being selected is 0.36 / 0.84≈43%. The system generates a random number in the interval [0,1). If the random number is less than 0.57, then node A is selected; otherwise, node B is selected.

[0062] Establish proxy links Establish a proxy link from the access node through the target relay node to the exit node, i.e., "access node → relay node → exit node", and switch the video stream from the direct link to the proxy link for transmission.

[0063] Example 3: Specific Application Scenarios Taking a distributed video gateway cluster as an example, access nodes are located in city-level data centers such as A, B, C, and X. Each access node connects to a local camera and provides an RTSP / RTMP raw stream interface. The egress node is located in a provincial data center, receiving requests from users throughout the province and providing a unified service interface and streaming media egress. Relay nodes and access nodes can be reused or deployed independently and have stream relay capabilities.

[0064] A user requests access to the video stream from a camera on node X. The specific decision-making process is as follows: 1. Direct connection to reference detection Detection of access node X → exit node: delay L=2000ms, packet loss rate P=5%, jitter J=50ms.

[0065] Substitute into the formula to calculate: ; ; ; ; Direct connection benchmark score If the score falls below the 50-point threshold, a relay election is triggered.

[0066] 2. Data Acquisition from Relay Nodes The data for each candidate relay node is as follows:

[0067] 3. Relay node score calculation Node A: CPU / memory idle, current flow count low. Both link segments tested are of good quality. Calculated overall score. .

[0068] Node B: Load is normal, link quality is slightly poor. Calculations show... .

[0069] Node C: High CPU load and poor link quality. Calculations show... .

[0070] Calculate the gain ratio: , , In the gain decision stage, due to Eliminated.

[0071] 4. Decision Results The gain ratios of both Node A and Node B are greater than 1.2, meeting the jump condition. Weight calculation: Node A's weight percentage is (1.48-1) / [(1.48-1)+(1.36-1)]≈57%; Node B's percentage is approximately 43%. The system will subsequently distribute requests for this camera to Node A and Node B in a 57:43 ratio.

[0072] Parameter setting instructions Normalized denominator (0.1 packet loss, 2000ms latency, 300ms jitter): based on current network statistics. When the packet loss rate > 10% or the latency > 2s, TCP transmission has experienced severe congestion, and the video player cannot maintain smooth playback; therefore, this is used as the cutoff point for the full score.

[0073] Weighting coefficients (P: 0.35, L: 0.35, J: 0.3): The video stream is extremely sensitive to packet loss (mosaic), so it is given the highest weight; jitter can be eliminated by buffering, so it is given a slightly lower weight.

[0074] Health weights (CPU: 0.3, Mem: 0.3, Disk: 0.2, Net: 0.2): Video transcoding / forwarding mainly consumes CPU and memory for encoding, decoding and caching, so it has a high weight; the disk is mainly used for log writing, and the network is limited by bandwidth, so each accounts for 20%.

[0075] Degradation trigger threshold (50 points): According to internal stress testing, when the link score is below 50 points, the retransmission rate of 1080P video stream will exceed 15%, and the user experience will be significantly reduced.

[0076] The 60% inflection point of the flow capacity factor C: Setting a 40% buffer is to prevent nodes from oscillating frequently near the full load critical point, thus achieving smooth soft current limiting.

[0077] Experimental data verification To verify the effectiveness of this solution, comparative tests were conducted in a simulated weak WAN environment. The test environment included 7 access nodes, 1 egress node, and 7 candidate relay nodes. The comparison was with a traditional solution (based solely on the lowest latency proxy).

[0078]

[0079] Experimental data show that the solution of this invention is significantly better than the traditional solution in terms of first screen loading time, stuttering rate, load balancing and system stability.

[0080] One or more embodiments in this application are intended to cover all such substitutions, modifications, and variations that fall within the broad scope of this application. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of one or more embodiments in this application should be included within the protection scope of this application.

[0081] If flowcharts are used in this application, they are used to illustrate the operations performed by the system according to embodiments of this application. It should be understood that the preceding or following operations are not necessarily performed in exact order. Instead, the steps can be processed in reverse order or simultaneously. Furthermore, other operations can be added to these flows, or one or more steps can be removed from these flows.

[0082] The above provides a detailed description of a distributed video proxy routing method and system provided in this application. The above description of the disclosed embodiments enables those skilled in the art to implement or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A distributed video proxy routing method, characterized in that, Includes the following steps: The system collects network link quality parameters, node load parameters, and node attribute parameters in real time. The network link quality parameters include the quality parameters of the direct link between the access node and the egress node, as well as the quality parameters of the segmented links between the access node, egress node, and each candidate relay node. The node load parameters include the resource usage parameters and current service capacity of the candidate relay nodes. The node attribute parameters include the rated capacity of the candidate relay nodes. The connection between the access node and the egress node passes through at most one relay node. Based on the quality parameters of the direct link, the network quality score of the direct link is calculated as the direct link benchmark value. When the direct link benchmark value is lower than the preset degradation threshold, for each candidate relay node: its available processing capacity is evaluated based on the resource occupancy parameters of the candidate relay node as a health factor; its remaining capacity is evaluated based on the remaining ratio of the current service carrying capacity of the candidate relay node to the rated carrying capacity as a capacity factor; and the overall transmission quality after the two links are connected is evaluated based on the segmented link quality parameters between the candidate relay node and the access node, and the segmented link quality parameters between the candidate relay node and the egress node as a link quality factor. A comprehensive score for the candidate relay nodes is generated based on the health factor, the capacity factor, and the link quality factor; the comprehensive scores of each candidate relay node are then filtered to form a candidate set. If the gain of the best comprehensive score in the candidate set relative to the direct connection benchmark value is less than a preset gain threshold, then the relay jump is rejected and transmission is maintained via the direct connection link; If the gain level is greater than or equal to a preset gain threshold, then selection weights are assigned to each candidate relay node in the candidate set in such a way that the higher the gain level, the greater the weight is assigned, and the target relay node is selected from the candidate set based on the selection weights. Establish a proxy link connecting the access node and the exit node via the target relay node, and switch the video stream from the direct link to the proxy link for transmission.

2. The distributed video proxy routing method according to claim 1, characterized in that, The resource usage parameters include CPU load rate, memory usage rate, disk I / O usage rate, and network I / O usage rate; The health factor is determined based on the following formula: ; in, For the aforementioned health factor, The CPU load rate, The memory usage rate, For the disk input / output utilization, The network input / output utilization rate, These are the corresponding preset weights.

3. The distributed video proxy routing method according to claim 2, characterized in that, The rated load capacity is obtained in the following manner: When a new relay node is launched or its configuration is changed, the number of concurrent streams is gradually increased by simulating mixed video streams of different resolutions until any one of the CPU load rate, memory usage rate and network input / output usage rate of the relay node reaches a preset threshold. At this point, the corresponding number of concurrent streams is taken as the rated capacity of the relay node, where the number of concurrent streams is the number of video stream channels that the relay node forwards simultaneously.

4. The distributed video proxy routing method according to claim 2, characterized in that, The capacity factor is determined based on the following formula: when hour, ; when hour, ; when hour, ; in, For the capacity factor, The current service capacity, The rated load capacity, The preset load buffer threshold, .

5. The distributed video proxy routing method according to claim 1, characterized in that, The network link quality parameters include latency, packet loss rate, and jitter; The network quality score is determined based on the following formula: ; ; ; ; in, The network quality score is given. For packet loss rate, For delay, For shaking, To normalize the packet loss rate score, To delay normalization of scores, To normalize the jitter score, These are the preset weights corresponding to packet loss rate, latency, and jitter, respectively. These are preset index factors corresponding to packet loss rate, latency, and jitter, respectively. These are the preset normalized baseline values ​​corresponding to packet loss rate, latency, and jitter, respectively.

6. The distributed video proxy routing method according to claim 5, characterized in that, The overall transmission quality, including delay, packet loss rate, and jitter, after the two links are connected is determined based on the following formula: ; ; ; in, For the overall transmission delay, For the overall packet loss rate, For overall transmission jitter, and These represent the delays of the two link segments, and These represent the packet loss rates of the two links, respectively. and These represent the jitter of the two links respectively.

7. The distributed video proxy routing method according to claim 4, characterized in that, The overall score is determined based on the following formula: ; in, The overall score is as follows. To preset the basic weights, For the aforementioned health factor, For the capacity factor, This refers to the link quality factor.

8. The distributed video proxy routing method according to claim 1, characterized in that, The process of filtering the candidate relay nodes based on their comprehensive scores to form a candidate set involves sorting all candidate relay nodes from highest to lowest according to their comprehensive scores, and retaining the nodes that rank in the top 30% and have at least 3 nodes as the candidate set.

9. The distributed video proxy routing method according to claim 8, characterized in that, The step of filtering the comprehensive scores of each candidate relay node to form a candidate set also includes: Candidate relay nodes whose comprehensive score is lower than the direct connection benchmark value are removed from the candidate set.

10. A distributed video proxy routing system, characterized in that, It includes access nodes, egress nodes, and relay nodes, wherein there are one or more relay nodes; The access node is used to connect to the camera and acquire the video stream from the camera, as well as to collect network link quality parameters and node load parameters; The exit node is used to receive user access requests for the video stream from the camera and to provide a unified streaming media access interface. The access node, the egress node, and the relay node are used together to execute the distributed video proxy routing method as described in any one of claims 1 to 9.