A media real-time transport system and method

Abstract

A media real-time transmission system and method, wherein the media real-time transmission system comprises a central signaling cluster and a plurality of edge media nodes; the central signaling cluster is deployed in a central cloud platform, and the central signaling cluster is integrated with a scheduling module with a built-in scheduling logic set; the scheduling module is used for intelligently selecting the optimal media transmission path and the edge media node according to the network parameters reported by the terminal and the scheduling logic set; the plurality of edge media nodes are connected with the central signaling cluster respectively; the plurality of edge media nodes are distributedly deployed in local edge machine rooms, and are wirelessly connected with terminals around the local machine rooms, and are used for completing media processing tasks; at least one MediaSoup SFU server is deployed in each edge media node; and a local STUN server is additionally deployed in each edge media node at least.

Description

Technical Field

[0001] This invention relates to the field of WebRTC real-time audio and video transmission technology, and in particular to a media real-time transmission system and method. Background Technology

[0002] With the rapid development of cloud phones, live streaming, and remote collaboration, higher demands are being placed on the low latency, high stability, low cost, and controllability of WebRTC (Web Real-Time Communication) media transmission. Currently, existing WebRTC media transmission solutions suffer from the following technical shortcomings, making them unsuitable for large-scale, multi-datacenter deployment scenarios: Option 1: Transmission scheme based on a single-center SFU (Selective Forwarding Unit) cluster.

[0003] By centrally deploying SFU servers such as MediaSoup (an open-source WebRTC media server designed for real-time audio and video communication) in a single central data center, all terminal media streams need to be transmitted back to the central cluster across regions for forwarding. This results in high cross-regional transmission latency, huge backbone network bandwidth consumption, and a single cluster failure can cause a global service avalanche, leading to poor stability and making it difficult to control bandwidth costs.

[0004] Option 2: Pure P2P (Peer-to-Peer) transmission scheme.

[0005] Relying on STUN (Session Traversal Utilities for NAT) servers or TURN (Traversal Using Relays around NAT) servers for hole punching, when hole punching is successful, the media stream connects directly end-to-end, saving server bandwidth. However, there are significant drawbacks: First, the success rate of P2P hole punching is limited by NAT type and carrier network, resulting in insufficient stability. Second, when hole punching fails, it automatically degrades to a TURN server relay, causing the media stream to fully occupy the cloud server bandwidth, and sudden traffic can easily cause a surge in billing. Third, it cannot implement media stream recording, transcoding, watermarking, rate limiting, and other control functions, making it difficult to meet the security auditing and quality control needs of scenarios such as cloud phones.

[0006] Option 3: Transmission scheme based on hybrid deployment of signaling and media.

[0007] The mixed deployment of signaling services and SFU media services results in poor architectural scalability, an inability to achieve unified global scheduling, and difficulty in optimizing media stream transmission paths, further exacerbating latency and bandwidth waste.

[0008] In summary, existing technologies cannot simultaneously solve the four core problems of low latency, low cost, high stability, and strong management and control in multi-data center scenarios. Summary of the Invention

[0009] This invention provides a real-time media transmission system and method to solve the technical problems mentioned in the background.

[0010] To achieve the above objectives, the present invention adopts the following technical solution: This invention provides a real-time media transmission system, comprising: The central signaling cluster is deployed on the central cloud platform. The central signaling cluster only transmits control signaling. The central signaling cluster integrates a scheduling module with a built-in set of scheduling logic. The scheduling module is used to intelligently select the optimal media transmission path and edge media nodes based on the network parameters reported by the terminal and the set of scheduling logic. Multiple edge media nodes, each connected to the central signaling cluster, are distributed across edge data centers in various locations and wirelessly connected to terminals around these data centers to complete media processing tasks. Each edge media node is equipped with at least one MediaSoup SFU server, and each edge media node is also equipped with a local STUN server, a local TURN server, and an intranet penetration and routing module.

[0011] Furthermore, the central signaling cluster serves as a unified global entry point for terminal authentication, terminal discovery, session description protocol exchange, coordination of interactive connection establishment, room management, access control, and global state synchronization.

[0012] Furthermore, the network parameters include the location of the terminal in the computer room, IP address, network address translation type, bandwidth status, and network latency; The media processing tasks include media stream forwarding, encoding / transcoding, audio mixing, screen recording, and watermarking for the local data center and surrounding terminals.

[0013] Furthermore, the scheduling logic set includes hierarchical scheduling logic, bandwidth priority logic, and node load balancing logic; The hierarchical scheduling logic is used to determine the set of candidate edge media nodes based on the region where the terminal is located, the network topology, and the preset regional priority rules, and to select edge media nodes according to the distance priority principle. The bandwidth priority logic is used to sort candidate edge media nodes and make access decisions based on terminal service type, media bitrate requirements and remaining bandwidth resources of edge media nodes. The node load balancing logic is used to dynamically load balance and schedule candidate edge media nodes based on their processing capacity, resource utilization, number of media sessions, and real-time operating status.

[0014] Furthermore, each edge media node is equipped with two parallel MediaSoup SFU servers, which are the MediaSoup SFU master server and the MediaSoup SFU backup server, respectively. If network fluctuations or node failures occur during transmission, the scheduling module automatically switches from the MediaSoup SFU master server to the MediaSoup SFU backup server.

[0015] Furthermore, the local STUN server is used to adapt the encoding parameters of the screen streams of terminals around the local data center; the local TURN server is used for media relay of terminals around the local data center, without cross-regional forwarding.

[0016] Furthermore, the intranet penetration and routing module is used to realize direct intranet connection between terminals in the same data center, media stream cascading forwarding between terminals in different data centers, and auxiliary adaptation for P2P hole punching; it is also responsible for adapting to differences in intranet segments, firewalls, and network address translation rules in data centers in different regions.

[0017] In another aspect, the present invention provides a method for real-time media transmission, comprising the following steps: S1. Select the hierarchical scheduling logic from the scheduling logic set, start each terminal, connect each terminal to the central signaling cluster, complete identity authentication, and report its own network parameters to the central signaling cluster. S2. The scheduling module of the central signaling cluster receives the network parameters reported by the terminal. Based on the hierarchical scheduling logic and combined with the media transmission requests initiated by the terminals at both ends, it performs hierarchical scheduling on the transmission paths of the terminals at both ends to obtain the scheduling results. The scheduling results include the optimal media transmission path and edge media nodes. S3. Based on the scheduling results, the two terminals establish corresponding media channels, and the edge media nodes perform transcoding, screen recording, and watermarking on the media streams that need to be controlled. S4. The terminal initiates a session termination request, and the central signaling cluster notifies the relevant edge media nodes to release resources and close the media channel.

[0018] Furthermore, the hierarchical scheduling in S2 includes the following steps: S21. Perform hierarchical judgment on the transmission paths of the two terminals. If the two terminals are in the same data center or the same intranet, proceed to S22; if the two terminals are in different data centers and in different regions, proceed to S23; if the terminal is in a set NAT environment, proceed to S24. S22: The scheduling module allocates MediaSoup SFU servers to local edge media nodes. Both ends of the terminal establish DTLS media channels or SRTP media channels with the allocated MediaSoup SFU servers. The media stream is transmitted in a closed loop within the data center intranet without passing through the public network or cross-regional backbone networks. Then it enters S3. Here, local edge media nodes refer to edge media nodes in the same intranet. S23. The scheduling module prioritizes guiding both terminals to attempt P2P direct connection, exchanging ICE candidate addresses through the local STUN server to verify the feasibility of P2P hole punching; if P2P hole punching is successful and no media control is required, the media stream is transmitted in P2P direct connection mode; if P2P hole punching fails, or media control is required, the scheduling module allocates the nearest edge media node to both terminals; then proceeds to S3. S24. The scheduling module directly allocates local edge media nodes and relays media streams through the local TURN server; then it proceeds to S3.

[0019] Furthermore, the following steps are included after step S3: If network fluctuations or node failures occur during transmission, the scheduling module will automatically switch from the MediaSoup SFU master server to the MediaSoup SFU backup server.

[0020] Compared with traditional media transmission systems and methods, the present invention has the following advantages: 1. Low-latency transmission; In this invention, media streams are preferentially transmitted via closed-loop network within the data center. In cross-regional scenarios, P2P direct connection or forwarding to the nearest edge media node is preferred, avoiding cross-regional backbone network relay, minimizing media transmission latency, and adapting to scenarios with extremely high real-time requirements such as cloud phones.

[0021] 2. Costs are controllable; This invention adopts local media stream processing, which significantly reduces cross-regional backbone network bandwidth consumption; the P2P direct connection mode saves server bandwidth and avoids sudden surges in TURN traffic caused by pure P2P hole punching failure, allowing for precise control of bandwidth costs.

[0022] 3. High stability and high availability; In this invention, multiple edge media nodes are deployed in a distributed manner. A single data center or node failure only affects the local terminal, preventing a global service collapse. The central signaling cluster employs a multi-site active-active architecture, ensuring uninterrupted signaling services. Simultaneously, in this architecture, control commands are synchronized between data centers (including the central signaling cluster and edge media nodes) via a "signaling channel," and data content is distributed via a "media channel." This ensures that even if a failure occurs in one area, other nodes can seamlessly take over the service, guaranteeing business continuity. The automatic switching mechanism of the scheduling module further enhances transmission stability.

[0023] 4. Strong control and management capabilities; This invention realizes functions such as screen recording, transcoding, watermarking, rate limiting, and auditing of media streams through edge media nodes, solving the defects of pure P2P transmission that cannot be controlled, and meeting the security and quality requirements of scenarios such as cloud phones and enterprise meetings.

[0024] 5. High scalability; The central signaling cluster in this invention is completely decoupled from multiple edge media nodes, and the number of central signaling clusters and edge media nodes can be flexibly increased according to business needs, adapting to large-scale terminal access scenarios and having strong scalability. Attached Figure Description

[0025] Figure 1 This is a structural diagram of the real-time media transmission system in this invention. Detailed Implementation

[0026] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Preferred embodiments of the invention are shown in the drawings. However, the invention can be implemented in many other different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to provide a thorough and complete understanding of the disclosure of the invention.

[0027] To address the shortcomings mentioned in the background technology, this invention provides a real-time media transmission system and method to solve the technical problems of high latency, high cost, and poor stability in existing single-center SFU systems, as well as the instability, weak control capabilities, and uncontrollable turn traffic of pure P2P transmission. This achieves low-latency, low-cost, highly available, and strongly controlled transmission of media streams in multi-datacenter scenarios; specifically as follows: Reference Figure 1 This application provides a real-time media transmission system, including: The central signaling cluster is deployed on the central cloud platform. The central signaling cluster only transmits control signaling, which is low-frequency, small-packet data and occupies negligible bandwidth. The central signaling cluster integrates a scheduling module with a built-in set of scheduling logic. The scheduling module is used to intelligently select the optimal media transmission path and edge media nodes based on the network parameters reported by the terminal and the set of scheduling logic. Multiple edge media nodes, each connected to the central signaling cluster, are distributed across edge data centers in various locations and wirelessly connected to terminals surrounding these data centers. They perform media processing tasks such as media stream forwarding, encoding / transcoding, audio mixing, screen recording, and watermarking for both the local data center and surrounding terminals, achieving local closed-loop transmission of media streams and avoiding cross-regional media stream relay. Each edge media node deploys at least one MediaSoup SFU (MediaSoup Selective Forwarding Unit, a high-performance, scalable audio / video selective forwarding server). The MediaSoup SFU is used by the edge media node to process and forward media streams from both the local data center and surrounding terminals, avoiding cross-regional media stream relay. Each edge media node is also equipped with a local STUN server, a local TURN server, and an intranet penetration and routing module. Cascading communication can be achieved between the edge media nodes for cross-data center media stream forwarding. In this embodiment, multiple edge media nodes are connected to the central signaling cluster via the network and simultaneously establish signaling channels with the central signaling cluster through the network to achieve identity authentication, parameter reporting, and signaling interaction.

[0028] In some embodiments, the number of central signaling clusters and edge media nodes can be flexibly increased according to business needs to adapt to large-scale terminal access scenarios.

[0029] In some embodiments, the central signaling cluster serves as a unified global entry point for terminal authentication, terminal discovery, Session Description Protocol (SDP) exchange, Interactive Connectivity Establishment (ICE) coordination, room management, access control, and global state synchronization.

[0030] In some embodiments, the network parameters include the location of the terminal in the computer room, IP (Internet Protocol Address), network address translation type, bandwidth status, and network latency.

[0031] In some embodiments, the scheduling logic set includes hierarchical scheduling logic, bandwidth priority logic, and node load balancing logic.

[0032] The hierarchical scheduling logic is used to determine the set of candidate edge media nodes based on the region where the terminal is located, the network topology, and the preset regional priority rules, and to select edge media nodes according to the distance priority principle. The bandwidth priority logic is used to sort candidate edge media nodes and make access decisions based on terminal service type, media bitrate requirements and remaining bandwidth resources of edge media nodes. The node load balancing logic is used to dynamically load balance and schedule candidate edge media nodes based on their processing capacity, resource utilization, number of media sessions, and real-time operating status.

[0033] In some embodiments, each edge media node is equipped with two parallel MediaSoup SFU servers, which are a MediaSoup SFU master server and a MediaSoup SFU backup server, respectively. If network fluctuations or node failures occur during transmission, the scheduling module automatically switches from the MediaSoup SFU master server to the MediaSoup SFU backup server.

[0034] In some embodiments, the local STUN server is used to adapt the encoding parameters of the screen streams of terminals around the local data center; the local TURN server is used for media relay of terminals around the local data center and does not forward across regions.

[0035] In some embodiments, the intranet penetration and routing module is used to realize direct intranet connection between terminals in the same data center, media stream cascading forwarding between terminals in different data centers, and auxiliary adaptation for P2P hole punching; at the same time, it is responsible for adapting to the differences in intranet segments, firewalls, and network address translation rules of data centers in different regions, so as to improve the success rate of media stream transmission.

[0036] First embodiment: The following example illustrates the deployment of a real-time media transmission system: Assuming we deploy three edge data centers located in regions A, B, and C respectively, and one central cloud platform, the specific deployment is as follows: Central Signaling Cluster: Deployed on the central cloud platform, it consists of 3 servers forming a geographically dispersed active-active cluster, running the WebSocket protocol, and is responsible for global signaling forwarding, terminal authentication, and room management; the signaling cluster uses a load balancer to distribute requests and ensure high availability.

[0037] Edge Media Nodes: One edge media node is deployed in each of regions A, B, and C. Each node consists of two MediaSoup SFU servers, one local STUN server, and one local TURN server. The local STUN server is used to adapt the encoding parameters of high-bitrate screen streams from cloud phones, supporting H.264 / HEVC (High Efficiency Video Coding) transcoding. The TURN server is used only for media relay within the local data center terminal and does not forward media across regions. The two MediaSoup SFU servers operate in a master-slave mode, serving as the MediaSoup SFU master server and the MediaSoup SFU backup server. Scheduling module: Integrated into the central signaling cluster, with preset scheduling rules: terminals in the same data center are given priority to be assigned to the local MediaSoup SFU server, terminals in different data centers are given priority to try P2P, if P2P fails, the nearest MediaSoup SFU server is assigned, and in complex NAT environments, the local MediaSoup SFU server is directly assigned.

[0038] Intranet penetration and routing module: Deployed at each edge media node, it adapts to the intranet segments of each data center, enabling direct intranet connection between terminals on the same network segment and forwarding of terminals across network segments through SFU cascading.

[0039] In another aspect, the present invention provides a method for real-time media transmission, comprising the following steps: S1. Select the hierarchical scheduling logic from the scheduling logic set, start each terminal, connect each terminal to the central signaling cluster, complete identity authentication, and report its own network parameters to the central signaling cluster. S2. The scheduling module of the central signaling cluster receives the network parameters reported by the terminal. Based on the hierarchical scheduling logic and combined with the media transmission requests (such as calls and streaming) initiated by the terminals at both ends, it performs hierarchical scheduling on the transmission paths of the terminals at both ends to obtain the scheduling results. The scheduling results include the optimal media transmission path and edge media nodes. S3. Based on the scheduling results, the two terminals establish corresponding media channels, and the edge media nodes perform transcoding, screen recording, watermarking and other processing on the media streams that need to be controlled. S4. The terminal initiates a session termination request, and the central signaling cluster notifies the relevant edge media nodes to release resources and close the media channel.

[0040] In some embodiments, the hierarchical scheduling in S2 includes the following steps: S21. Perform hierarchical judgment on the transmission paths of the two terminals. If the two terminals are in the same data center or the same intranet, proceed to S22. If the two terminals are in different data centers and in different regions, proceed to S23. If the terminal is in a set NAT environment (such as a complex NAT environment, which includes multi-layer NAT and symmetric NAT), proceed to S24. S22: The scheduling module allocates a MediaSoup SFU server to the local edge media node. Both ends of the terminal establish a DTLS (Datagram Transport Layer Security) media channel or an SRTP (Secure Real-time Transport Protocol) media channel with the allocated MediaSoup SFU server. The media stream is transmitted in a closed loop within the data center intranet, without passing through the public network or cross-regional backbone network; then it enters S3. Here, the local edge media node refers to the edge media node in the same intranet. S23. The scheduling module prioritizes guiding both terminals to attempt P2P direct connection, exchanging ICE candidate addresses through the local STUN server to verify the feasibility of P2P hole punching. If P2P hole punching is successful and media control (such as screen recording or watermarking) is not required, the media stream is transmitted using the P2P direct connection mode. If P2P hole punching fails, or media control is required, the scheduling module allocates the nearest edge media nodes to both terminals, and forwards the media stream through cascading edge media nodes to avoid backhauling to the central signaling cluster. Then, proceed to S3. S24. The scheduling module directly allocates local edge media nodes and relays media streams through the local TURN server to ensure transmission stability and avoid bandwidth waste caused by cross-regional TURN relay; then it proceeds to S3.

[0041] In some embodiments, the following steps are included after step S3: If network fluctuations or node failures occur during transmission, the scheduling module will automatically switch from the MediaSoup SFU master server to the MediaSoup SFU backup server.

[0042] Second embodiment: The following description, based on the deployment of the first embodiment and using a cloud mobile terminal as an example, illustrates the real-time media transmission method: Terminal access: Cloud mobile terminal a in data center A and cloud mobile terminal b in data center B are started and connected to the central signaling cluster. After completing identity authentication, they report their own network parameters: terminal a (data center in region A, IP: 192.168.1.100, NAT type: cone NAT, bandwidth: 100Mbps), terminal b (data center in region B, IP: 192.168.2.200, NAT type: symmetric NAT, bandwidth: 100Mbps).

[0043] Scenario 1 (Two terminals across data centers): Scheduling Judgment: Terminal a initiates a screen stream pull request with terminal b. The scheduling module of the central signaling cluster determines that the two terminals are in different data centers and that terminal b is a symmetric NAT with a low success rate of P2P hole punching. Therefore, the scheduling module assigns terminal a to the MediaSoup SFU master server of the edge media node in data center A, and terminal b to the MediaSoup SFU master server of the edge media node in data center B, using the edge media node cascading forwarding mode.

[0044] Media stream establishment: Terminal a establishes a DTLS or SRTP channel with the MediaSoup SFU master server in data center A, and pushes the cloud phone screen stream to the MediaSoup SFU master server in data center A; the MediaSoup SFU master server in data center A forwards the media stream in a concatenation manner to the MediaSoup SFU master server in data center B; terminal b establishes a media channel with the MediaSoup SFU master server in data center B, and pulls the forwarded screen stream; the central signaling cluster monitors the media stream status at both ends, and the MediaSoup SFU master server in data center A adds watermarks to the screen stream.

[0045] Scenario 2 (Both terminals are in the same data center): Scheduling decision: If terminal a in data center A and terminal c in data center C (IP of terminal c: 192.168.1.101) initiate a call request, the scheduling module of the central signaling cluster determines that the two terminals are in the same data center and allocates a local SFU in data center A. Media stream establishment: Both terminal a and terminal c establish a media channel with the local MediaSoup SFU server. The media stream is transmitted within the intranet of the data center in region A, with latency controlled within 10ms and without occupying public network bandwidth, which greatly saves costs.

[0046] Anomaly Handling: If the MediaSoup SFU master server in the data center of region A fails, the scheduling module will automatically switch the MediaSoup SFU master server to the MediaSoup SFU backup server. The central signaling cluster will synchronously update the scheduling information. Terminal a will not need to reconnect, and the media stream transmission will not be interrupted.

[0047] Session End: Terminal a closes the streaming request, and the central signaling cluster notifies the MediaSoupSFU servers (including the MediaSoup SFU master server and the MediaSoup SFU backup server) in the data centers of regions A and B to release resources, close the media channel, and the session ends.

[0048] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Furthermore, the technical solutions of the various embodiments of the present invention can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A real-time media transmission system, characterized in that, include: The central signaling cluster is deployed on the central cloud platform. The central signaling cluster only transmits control signaling. The central signaling cluster integrates a scheduling module with a built-in set of scheduling logic. The scheduling module is used to intelligently select the optimal media transmission path and edge media nodes based on the network parameters reported by the terminal and the set of scheduling logic. Multiple edge media nodes, each connected to the central signaling cluster, are distributed and deployed in edge data centers in various locations. They are also wirelessly connected to terminals around the data centers in various locations to complete media processing tasks. Each edge media node is equipped with at least one MediaSoup SFU server, and each edge media node is also equipped with a local STUN server, a local TURN server, and an intranet penetration and routing module. The scheduling logic set includes hierarchical scheduling logic, bandwidth priority logic, and node load balancing logic; The hierarchical scheduling logic is used to determine the set of candidate edge media nodes based on the region where the terminal is located, the network topology, and the preset regional priority rules, and to select edge media nodes according to the distance priority principle. The bandwidth priority logic is used to sort candidate edge media nodes and make access decisions based on terminal service type, media bitrate requirements and remaining bandwidth resources of edge media nodes. The node load balancing logic is used to dynamically load balance and schedule candidate edge media nodes based on their processing capacity, resource utilization, number of media sessions, and real-time operating status. The MediaSoup SFU server is used for edge media nodes to process and forward media streams from the local data center and surrounding terminals; the local TURN server is used for media relay from surrounding terminals in the local data center, without forwarding across regions. The hierarchical scheduling includes the following steps: S21. Perform hierarchical judgment on the transmission paths of the two terminals. If the two terminals are in the same data center or the same intranet, proceed to S22; if the two terminals are in different data centers and in different regions, proceed to S23; if the terminal is in a set NAT environment, proceed to S24. S22. The scheduling module allocates MediaSoup SFU servers to local edge media nodes. Both ends of the terminal establish DTLS media channels or SRTP media channels with the allocated MediaSoup SFU servers. The media stream is transmitted in a closed loop within the data center intranet without passing through the public network or cross-regional backbone networks. Here, local edge media nodes refer to edge media nodes in the same intranet. S23. The scheduling module prioritizes guiding both terminals to attempt P2P direct connection, exchanging ICE candidate addresses through the local STUN server to verify the feasibility of P2P hole punching; if P2P hole punching is successful and no media control is required, the media stream is transmitted in P2P direct connection mode; if P2P hole punching fails, or media control is required, the scheduling module allocates the nearest edge media node to both terminals. S24. The scheduling module directly allocates local edge media nodes and relays media streams through the local TURN server.

2. The media real-time transmission system according to claim 1, characterized in that, The central signaling cluster serves as a unified global entry point, used for terminal authentication, terminal discovery, session description protocol exchange, coordination of interactive connection establishment, room management, access control, and global state synchronization.

3. The media real-time transmission system according to claim 1, characterized in that, The network parameters include the location of the terminal in the computer room, IP address, network address translation type, bandwidth status, and network latency; The media processing tasks include media stream forwarding, encoding / transcoding, audio mixing, screen recording, and watermarking for the local data center and surrounding terminals.

4. A real-time media transmission system according to claim 1, characterized in that, Each edge media node is equipped with two parallel MediaSoup SFU servers, which are the MediaSoup SFU master server and the MediaSoup SFU backup server, respectively. If network fluctuations or node failures occur during transmission, the scheduling module automatically switches from the MediaSoup SFU master server to the MediaSoup SFU backup server.

5. A real-time media transmission system according to claim 1, characterized in that, The intranet penetration and routing module is used to enable direct intranet connection between terminals in the same data center, media stream cascading forwarding between terminals in different data centers, and auxiliary adaptation for P2P hole punching; it is also responsible for adapting to differences in intranet segments, firewalls, and network address translation rules in data centers in different regions.

6. A real-time media transmission method based on a real-time media transmission system according to any one of claims 1 to 5, characterized in that, Includes the following steps: S1. Select the hierarchical scheduling logic from the scheduling logic set, start each terminal, connect each terminal to the central signaling cluster, complete identity authentication, and report its own network parameters to the central signaling cluster. S2. The scheduling module of the central signaling cluster receives the network parameters reported by the terminal. Based on the hierarchical scheduling logic and combined with the media transmission requests initiated by the terminals at both ends, it performs hierarchical scheduling on the transmission paths of the terminals at both ends to obtain the scheduling results. The scheduling results include the optimal media transmission path and edge media nodes. S3. Based on the scheduling results, the two terminals establish corresponding media channels, and the edge media nodes perform transcoding, screen recording, and watermarking on the media streams that need to be controlled. S4. The terminal initiates a session termination request, and the central signaling cluster notifies the relevant edge media nodes to release resources and close the media channel.

7. A real-time media transmission method according to claim 6, characterized in that, Following step S3, the following steps are also included: If network fluctuations or node failures occur during transmission, the scheduling module will automatically switch from the MediaSoup SFU master server to the MediaSoup SFU backup server.

Images