A data channel service system and implementation method for IMS network

By using the data channel server based on the SCTP protocol and the priority management device FPM in the IMS network, the priority and congestion window size of the SCTP data flow are dynamically adjusted, and the problems of data channel transmission efficiency, utilization and delay in the IMS network are solved. It realizes support and performance improvement of multimedia call applications.

CN116032893BActive Publication Date: 2024-11-26EB INFORMATION TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202211614952.3
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-12-14
Publication Date
2024-11-26
Estimated Expiration
2042-12-14

AI Technical Summary

Technical Problem

In IMS networks, SCTP over DTLS has performance problems in implementing data channels, including low transmission efficiency, low utilization rate and large transmission delay, especially when many-to-many traffic shares the same network bottleneck, the existing Research focuses on the coupled flow between peers and fails to effectively solve this problem.

Method used

By establishing a data channel server based on SCTP protocol in the IMS network, the transmission rate is calculated using a congestion control algorithm based on delay and loss, and dynamically adjusting the priority and congestion window size for multiple SCTP data flows through the priority management device FPM, ensuring that Fair distribution and efficient transmission of traffic sets.

Benefits of technology

It realizes efficient transmission of data channels in the IMS network, reduces overall delay and loss, supports multimedia call applications, improves the performance and utilization of data channels, and can introduce rich interactive information before, during and after calls.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116032893B_ABST
    Figure CN116032893B_ABST
Patent Text Reader

Abstract

A data channel service system and implementation method for an IMS network, the method comprising: step 1, each mobile terminal UE and a data channel server establish a data channel based on the SCTP protocol; step 2, the data channel server receives a network information feedback packet sent by each UE, and then simultaneously uses a delay-based congestion control algorithm and a loss-based congestion control algorithm to calculate the corresponding sending rate, and selects a small sending rate to send an SCTP data stream to the UE. The present invention relates to the field of communications, can implement a data channel based on SCTP based on an IMS network, and effectively optimize the performance of the data channel in terms of transmission efficiency, utilization, transmission delay, etc.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to a data channel service system for an IMS network and an implementation method thereof, and relates to the communication field. Background Art

[0002] In the past, operators' native video calling services based on IMS were not popularized, mainly due to the low popularity of terminals supporting video calling, high tariffs, insufficient network coverage, and the inability to communicate across networks. In the 5G era, these obstacles have been or are being eliminated, and IMS-based video calling has the foundation for large-scale commercial use.

[0003] WebRTC is a new standard defined by the Internet Engineering Task Force (IETF) and the World Wide Web Consortium (W3C) for implementing real-time communications in web browsers. It includes the ability to transmit audio and video over so-called media channels using the Secure Real-time Transport Protocol (SRTP), and the ability to transmit arbitrary data over data channels based on the Stream Control Transmission Protocol (SCTP).

[0004] However, there are a series of performance issues in implementing the data channel server using SCTP over DTLS in the IMS network. The role of DTLS is to encrypt data channel data to ensure data security and add a link certificate verification mechanism to prevent network attacks. Unlike TLS over TCP, the UDP layer does not handle the disorder and packet loss of data messages, which will result in the inability to guarantee the link certificate verification negotiation. Therefore, DTLS needs to add a reliable transmission mechanism to the handshake message when creating a connection. Since UDP has high requirements for the real-time performance of data, after the link is established and stabilized, when transmitting application data, no data reliability transmission mechanism is provided, which requires the SCTP protocol or application layer to guarantee. However, since SCTP uses a congestion control algorithm similar to TCP and uses packet loss as a congestion signal, the performance of the algorithm will deteriorate when the packet loss rate of the link is large.

[0005] Existing research has proposed a solution for RTP streams called "flow state exchange" (FSE), combined with congestion control that shares the same bottleneck. FSE uses the congestion controller of the flow to let them share information with each other. FSE-NG combines active FSE and ROSIEEE algorithms to support traffic priorities while still being able to couple and manage loss- and delay-based flows. However, existing research at home and abroad only focuses on coupled flows running between two peers, while DCS may exist where all flows sent by a given peer to one or more destinations may share the same network bottleneck.

[0006] Therefore, how to implement SCTP-based data channels based on the IMS network and effectively optimize the performance of the data channels in terms of transmission efficiency, utilization, transmission delay, etc. has become a technical issue that technical personnel have focused on. Summary of the invention

[0007] In view of this, the purpose of the present invention is to provide a data channel service system and implementation method for an IMS network, which can implement an SCTP-based data channel based on the IMS network and effectively optimize the performance of the data channel in terms of transmission efficiency, utilization, transmission delay, etc.

[0008] In order to achieve the above object, the present invention provides a data channel service system for an IMS network, including a data channel server and a plurality of mobile terminals UE, wherein:

[0009] The data channel server establishes a data channel based on the SCTP protocol with each UE, then receives the network information feedback packet sent by each UE, and uses the delay-based congestion control algorithm and the loss-based congestion control algorithm to calculate the corresponding sending rate, and selects a small sending rate to send the SCTP data stream to the corresponding UE.

[0010] The data channel server also includes a priority control device and a priority management device FPM, wherein:

[0011] The priority control device registers multiple SCTP data streams through the same network path to FPM and inputs the congestion window size and network round-trip delay of each SCTP data stream: CC_CWND(f i ) and last_rtt(f i ), where f i Indicates the i-th SCTP data stream, CC_CWND(f i ) is f i The congestion window size, last_rtt(f i ) is f i The network round trip delay is

[0012] FPM further includes:

[0013] The congestion window allocation unit forms a traffic set for all SCTP data flows passing through the same network path and saves the network round-trip delay of each SCTP data flow: last_rtt(f i ), and then calculate the output rate of each SCTP data flow in the traffic set: CC_R(f i ) is f i The output rate of the flow set is calculated to obtain the current total output rate S_CR of the flow set: N is the total number of SCTP data streams in the traffic set. Then set the priority of each SCTP data stream in the traffic set and calculate the benefit of successful transmission of each data stream:

[0014] Q(f i ) is f i The profit of successful transmission, time now is the current time, time create (f i ) is f i The creation time of the corresponding data block, blcok remainsize (f i ) is f i The remaining size of the corresponding data block, P(f i ) is f i The priority of all data flows in the traffic set is calculated, and the sum of the benefits of successful transmission S_Q is calculated as follows: Finally, according to the benefits of successful transmission of each data stream, the congestion window size of each SCTP data stream in the traffic set is calculated: FPM_CWND(f i )=FPM_R(f i )×last_rtt(f i ), FPM_R(f i ) is f i The priority algorithm output rate, FPM_CWND(f i ) is f i The congestion window size output by the priority algorithm is L_R, which is the current remaining allocation rate of the traffic set. Its value is set to S_CR, and the congestion window size output by the priority algorithm is calculated and allocated to the corresponding SCTP data stream in the traffic set.

[0015] In order to achieve the above object, the present invention also provides a method for implementing a data channel service for an IMS network, comprising:

[0016] Step 1: Each mobile terminal UE and the data channel server establish a data channel based on the SCTP protocol;

[0017] Step 2: The data channel server receives the network information feedback packet sent by each UE, and then uses both the delay-based congestion control algorithm and the loss-based congestion control algorithm to calculate the corresponding sending rate, and selects a smaller sending rate to send the SCTP data stream to the UE.

[0018] The data channel server includes a priority control device and a priority management device FPM. When the priority control device finds that multiple SCTP data streams pass through the same network path, it also includes:

[0019] Step B1: The priority control device registers multiple SCTP data streams passing through the same network path to FPM, and inputs the congestion window size and network round-trip delay of each SCTP data stream: CC_CWND(f i ) and last_rtt(f i ), where f i Indicates the i-th SCTP data stream, CC_CWND(f i ) is f i The congestion window size, last_rtt(f i ) is f i The network round trip delay

[0020] Step B2: FPM forms a traffic set with all SCTP data flows passing through the same network path, and saves the network round trip delay of each SCTP data flow: last_rtt(f i ), and then calculate the output rate of each SCTP data flow in the traffic set: CC_R(f i ) is f i The output rate of the flow set is calculated based on the current total output rate S_CR: Where N is the total number of SCTP data flows in the traffic set;

[0021] Step B3, FPM sets the priority of each SCTP data stream in the traffic set, and then calculates the benefit of successful transmission of each data stream: Where Q(f i ) is f i The profit of successful transmission, time now is the current time, time create (f i ) is f i The creation time of the corresponding data block, blcok remainsize (f i ) is f i The remaining size of the corresponding data block, P(f i ) is f i The priority of the data flow is used to calculate the sum of the benefits of successful transmission of all data flows in the traffic set S_Q:

[0022] Step B4: FPM calculates the congestion window size of each SCTP data flow in the traffic set according to the revenue of successful transmission of each data flow: FPM_CWND(f i )=FPM_R(f i )×last_rtt(f i ), where FPM_R(fi ) is f i The priority algorithm output rate, FPM_CWND(f i ) is f i The congestion window size output by the priority algorithm is L_R, which is the current remaining allocation rate of the traffic set. Its value is set to S_CR, and then the congestion window size output by the calculated priority algorithm is allocated to the corresponding SCTP data flow in the traffic set.

[0023] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention builds an available data channel server based on the IMS proprietary bearer, creates a new call data channel on top of the existing audio and video capabilities, and introduces a congestion control algorithm into the data channel service, thereby introducing richer interactive information such as text, pictures, expressions, positions, actions, and even sharing of mobile phone desktops before, during, and after the call on the basis of the original audio and video channels, upgrading the call from a single media to multimedia, achieving multi-dimensional interaction, and supporting a series of emerging multimedia call applications; the present invention extends FPM to the data channel service, sharing multiple SCTP streams on the same path, assigning different priorities according to different service types, and dynamically adjusting the rate of streams of different priorities. For high-priority service types, more bandwidth is allocated, thereby fairly allocating available bandwidth, reducing overall delays and losses, and solving related performance problems after the introduction of data channel services. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 The present invention is a schematic diagram of the structure of a data channel service system for an IMS network.

[0025] Figure 2 The present invention is a flow chart of a method for implementing a data channel service in an IMS network.

[0026] Figure 3 It is a flowchart of the specific steps for FPM to dynamically allocate congestion window sizes to flows of different priorities when the priority control device finds that multiple SCTP data flows pass through the same network path.

[0027] Figure 4 It is a flowchart of the specific steps for FPM to dynamically adjust the congestion window size of all data flows in the traffic set when the congestion window size CWND of any SCTP data flow in the traffic set changes. DETAILED DESCRIPTION

[0028] In order to make the objectives, technical solutions and advantages of the present invention more clear, the present invention is further described in detail below with reference to the accompanying drawings.

[0029] like Figure 1As shown, the present invention provides a data channel service system for an IMS network, including a data channel server and a plurality of mobile terminals UE, wherein:

[0030] The data channel server establishes a data channel based on the SCTP protocol with each UE, then receives the network information feedback packet sent by each UE, and uses the delay-based congestion control algorithm and the loss-based congestion control algorithm to calculate the corresponding sending rate, and selects a small sending rate to send the SCTP data stream to the corresponding UE.

[0031] The data channel server further includes:

[0032] The data channel construction device first completes the media description information SDP negotiation exchange with the UE, then exchanges the candidate address Candidate with the UE to establish a PeerConnection connection, and finally establishes a DataChannel connection with the UE.

[0033] The data channel construction device further includes:

[0034] When the SDP negotiation unit receives the Offer SDP from the UE, it sets the remote SDP description information to the local by using the SetRemoteDescription method, then creates the Answer SDP by using the CreateAnswer method, and sets the answer to the local by using the setLocalDescription method, and sends the Answer SDP to the UE;

[0035] The Candidate switching unit listens to OnICECandidate and triggers this method after setting the local SDP description information. When the ice candidate is available, the candidate is sent to the UE. When the candidate is received from the UE, the AddICECandidate method of adding the other party's network information is executed to add the candidate to the local.

[0036] The UE further comprises:

[0037] The UE-SDP negotiation device creates OfferSDP to the data channel server through the CreateOffer method, sets LocalDescription and sends it to the data channel server, then receives the Answer SDP sent by the data channel server, and sets it to the local through SetRemoteDescription;

[0038] The UE-Candidate switching device monitors OnICECandidate, and when an ice candidate is available, sends the candidate to the data channel server, and when receiving the candidate sent by the data channel server, executes AddICECandidate to add the candidate to the local.

[0039] When creating a DataChannel connection, the data channel construction device of the data channel server can configure reliability through the ordered (whether to ensure sequential transmission), maxRetransmitTimeMs (maximum time allowed for retransmission), and maxRetransmits (maximum number of retransmissions allowed) parameters of DataChannel.Init.

[0040] The present invention can also use the reinforcement learning optimization algorithm GCC to adjust the parameters in the delay-based congestion control algorithm or the loss-based congestion control algorithm in the data channel server, and the data channel server also includes:

[0041] Reinforcement learning optimization device, build reinforcement learning model GCC, the input is the state space Among them, T t represents the throughput vector at time t, D t represents the inter-packet delay jitter vector at time t, R t represents the RTT vector at time t, L t Represents the packet loss rate vector at time t, and the output is the predicted sending rate. The reward function reward takes throughput as positive feedback, and delay and packet loss rate as negative feedback: reward = α*throughput + β*delay + γ*packet loss rate, where α, β, and γ are the weight coefficients of throughput, delay, and packet loss rate, respectively, representing the impact of throughput, delay, and packet loss rate on reward, and α>0, β<0, γ<0. A fully connected network is used for feature extraction, the number of fully connected layers is 2, and the number of neurons in each layer is set to 64.

[0042] When multiple data streams pass through the same network path, they will compete for shared bandwidth. The resulting rate is usually the result of competition at the bottleneck. The present invention can also dynamically adjust the congestion window size for flows of different priorities through the priority management device FPM, thereby fairly allocating available bandwidth and reducing overall delay and loss. The data channel server also includes a priority control device and FPM, wherein:

[0043] The priority control device registers multiple SCTP data streams through the same network path to FPM and inputs the congestion window size and network round-trip delay of each SCTP data stream: CC_CWND(f i) and last_rtt(f i ), where f i Indicates the i-th SCTP data stream, CC_CWND(f i ) is f i The congestion window size, last_rtt(f i ) is f i The network round trip delay is

[0044] FPM further includes:

[0045] The congestion window allocation unit forms a traffic set for all SCTP data flows passing through the same network path and saves the network round-trip delay of each SCTP data flow: last_rtt(f i ), and then calculate the output rate of each SCTP data flow in the traffic set: CC_R(f i ) is f i The output rate of the flow set is calculated to obtain the current total output rate S_CR of the flow set: N is the total number of SCTP data streams in the traffic set. Then set the priority of each SCTP data stream in the traffic set and calculate the benefit of successful transmission of each data stream: Q(f i ) is f i The profit of successful transmission, time now is the current time, time create (f i ) is f i The creation time of the corresponding data block, blcok remainsize (f i ) is f i The remaining size of the corresponding data block, P(f i ) is f i The priority of all data flows in the traffic set is calculated, and the sum of the benefits of successful transmission S_Q is calculated as follows: Finally, according to the benefits of successful transmission of each data stream, the congestion window size of each SCTP data stream in the traffic set is calculated: FPM_CWND(f i )=FPM_R(f i )×last_rtt(f i ), FPM_R(f i ) is f i The priority algorithm output rate, FPM_CWND(f i ) is f iThe congestion window size output by the priority algorithm is L_R, which is the current remaining allocation rate of the traffic set. Its value is set to S_CR, and the congestion window size output by the priority algorithm is calculated and allocated to the corresponding SCTP data stream in the traffic set.

[0046] When the congestion window size CWND of a SCTP data stream in the traffic set changes, FPM can also dynamically adjust the congestion window size of all data streams in the traffic set. a When the congestion window size CWND changes, f a The current CC_CWND(f a ) and last_rtt(f a ) is sent to FPM, which also includes:

[0047] Congestion window adjustment unit, updates the locally saved f a last_rtt(f a ), calculate f a The output rate CC_R(f a ): Then query f a Belong to the traffic set, adjust f a The current total output rate S_CR′ of the traffic set to which it belongs: S_CR′=S_CR+CC_R(f a )-FPM_R(f a ), FPM_R(f a ) is FPM for f a The output rate of the priority algorithm last allocated, and then f a The value of the current remaining allocation total rate L_R of the traffic set to which it belongs is updated to S_CR′, and f is recalculated a The congestion window size output by the priority algorithm of each SCTP data flow in the traffic set: FPM_CWND(f j )=FPM_R(f j )×last_rtt(f j ), f j Yes a The jth SCTP data flow in the traffic set, then

[0048] Assign the calculated congestion window size output by the priority algorithm to f a The corresponding SCTP data stream in the traffic set.

[0049] like Figure 2 As shown, the present invention provides a data channel server implementation method for an IMS network, comprising:

[0050] Step 1: Each mobile terminal UE and the data channel server establish a data channel based on the SCTP protocol;

[0051] Step 2: The data channel server receives the network information feedback packet sent by each UE, and then uses both the delay-based congestion control algorithm and the loss-based congestion control algorithm to calculate the corresponding sending rate, and selects a smaller sending rate to send the SCTP data stream to the UE. Step 2 can specifically use the Google congestion control algorithm to calculate the sending rate.

[0052] Figure 2 Step one may further include:

[0053] Step A1: UE and data channel server complete the negotiation and exchange of media description information SDP;

[0054] Step A2: The UE and the data channel server exchange candidate addresses Candidate, and then establish a PeerConnection connection;

[0055] Step A3: The UE and the data channel server establish a DataChannel connection.

[0056] Step A1 may further include:

[0057] Step A11: The UE creates an Offer SDP to the data channel server through the CreateOffer method, sets LocalDescription, and sends it to the data channel server;

[0058] Step A12: The data channel server receives the Offer SDP from the UE and sets it locally through the method SetRemoteDescription to set the remote SDP description information;

[0059] Step A13: The data channel server creates an Answer SDP by using the remote create answer method CreateAnswer, sets the answer to the local by using the local SDP description information method setLocalDescription, and sends the Answer SDP to the UE.

[0060] Step A14: The UE receives the Answer SDP sent by the data channel server, and sets it locally through SetRemoteDescription.

[0061] Step A2 may further include: the data channel server (or UE) monitors OnICECandidate, and when icecandidate is available, sends the candidate to the UE (or data channel server), and the UE (or data channel server) executes the method AddICECandidate to add the other party's network information to add the candidate locally.

[0062] When creating a DataChannel connection in step A3, reliability can be configured through the parameters ordered (whether to ensure sequential transmission), maxRetransmitTimeMs (maximum time allowed for retransmission), and maxRetransmits (maximum number of retransmissions allowed) of DataChannel.Init.

[0063] Figure 2 In step 2, the present invention may also use the reinforcement learning optimization algorithm GCC to adjust the parameters in the delay-based congestion control algorithm or the loss-based congestion control algorithm, and also includes:

[0064] Construct a reinforcement learning model GCC, the input is the state space Among them, T t represents the throughput vector at time t, D t represents the inter-packet delay jitter vector at time t, R t represents the RTT vector at time t, L t Represents the packet loss rate vector at time t, and the output is the predicted sending rate. The reward function reward takes throughput as positive feedback, and delay and packet loss rate as negative feedback: reward = α*throughput + β*delay + γ*packet loss rate, where α, β, and γ are the weight coefficients of throughput, delay, and packet loss rate, respectively, representing the impact of throughput, delay, and packet loss rate on reward, and α>0, β<0, γ<0. A fully connected network is used for feature extraction, the number of fully connected layers is 2, and the number of neurons in each layer is set to 64.

[0065] When multiple data flows pass through the same network path, they will compete for shared bandwidth. The resulting rate is usually the result of competition at the bottleneck. The present invention can also dynamically allocate congestion window sizes to flows of different priorities through the priority management device FPM, thereby fairly allocating available bandwidth and reducing overall delay and loss. Figure 3 As shown, the data channel server includes a priority control device and an FPM. When the priority control device finds that multiple SCTP data streams pass through the same network path, it also includes:

[0066] Step B1: The priority control device registers multiple SCTP data streams passing through the same network path to FPM, and inputs the congestion window size and network round-trip delay of each SCTP data stream: CC_CWND(f i ) and last_rtt(f i ), where f i Indicates the i-th SCTP data stream, CC_CWND(f i ) is f i The congestion window size, last_rtt(f i ) is f i The network round trip delay

[0067] Step B2: FPM forms a traffic set with all SCTP data flows passing through the same network path, and saves the network round trip delay of each SCTP data flow: last_rtt(f i ), and then calculate the output rate of each SCTP data flow in the traffic set: CC_R(f i ) is f i The output rate of the flow set is calculated based on the current total output rate S_CR: Where N is the total number of SCTP data flows in the traffic set;

[0068] Step B3, FPM sets the priority of each SCTP data stream in the traffic set, and then calculates the benefit of successful transmission of each data stream: Where Q(f i ) is f i The profit of successful transmission, time now is the current time, time create (f i ) is f i The creation time of the corresponding data block, blcok remainsize (f i ) is f i The remaining size of the corresponding data block, P(f i ) is f i The priority of the data flow is used to calculate the sum of the benefits of successful transmission of all data flows in the traffic set S_Q:

[0069] Step B4: FPM calculates the congestion window size of each SCTP data flow in the traffic set according to the revenue of successful transmission of each data flow: FPM_CWND(f i )=FPM_R(f i )×last_rtt(f i ), where FPM_R(f i) is f i The priority algorithm output rate, FPM_CWND(f i ) is f i The congestion window size output by the priority algorithm is L_R, which is the current remaining allocation rate of the traffic set. Its value is set to S_CR, and then the congestion window size output by the calculated priority algorithm is allocated to the corresponding SCTP data flow in the traffic set.

[0070] like Figure 4 As shown in Figure 2, when the congestion window size CWND of any SCTP data flow in the traffic set changes, FPM can also dynamically adjust the congestion window size of all data flows in the traffic set to a For example, it also includes:

[0071] Step C1: The priority control device sets f a The current CC_CWND(f a ) and last_rtt(f a )Send to FPM;

[0072] Step C2: FPM updates the locally saved f a last_rtt(f a ), and calculate f a The output rate CC_R(f a ): Then query f a Belong to the traffic set, adjust f a The current total output rate S_CR′ of the traffic set to which it belongs: S_CR′=S_CR+CC_R(f a )-FPM_R(f a ), where FPM_R(f a ) is FPM for f a The last assigned priority algorithm output rate;

[0073] Step C3, FPM converts f a The value of the current remaining allocation total rate L_R of the traffic set to which it belongs is updated to S_CR′, and f is recalculated a The congestion window size output by the priority algorithm of each SCTP data flow in the traffic set: FPM_CWND(f j )=FPM_R(f j )×last_rtt(f j ), where f j Yes a The jth SCTP data flow in the traffic set belongs to, and then the congestion window size calculated by the priority algorithm output is assigned to fa The corresponding SCTP data stream in the traffic set.

[0074] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A data channel service system for an IMS network, characterized in that: It includes a data channel server and multiple mobile terminals UE, wherein: The data channel server establishes a data channel based on the SCTP protocol with each UE, then receives the network information feedback packet sent by each UE, and uses the delay-based congestion control algorithm and the loss-based congestion control algorithm to calculate the corresponding sending rate, and selects a small sending rate to send the SCTP data stream to the UE. The data channel server also includes a priority control device and a priority management device FPM, wherein: The priority control device registers multiple SCTP data streams through the same network path to FPM and inputs the congestion window size and network round-trip delay of each SCTP data stream: CC_CWND(f i ) and last_rtt(f i ), where f i Indicates the i-th SCTP data stream, CC_CWND(f i ) is f i The congestion window size, last_rtt(f i ) is f i The network round trip delay is FPM further includes: The congestion window allocation unit forms a traffic set for all SCTP data flows passing through the same network path and saves the network round-trip delay of each SCTP data flow: last_rtt(f i ), and then calculate the output rate of each SCTP data flow in the traffic set: CC_R(f i ) is f i The output rate of the flow set is calculated to obtain the current total output rate S_CR: N is the total number of SCTP data streams in the traffic set. Then set the priority of each SCTP data stream in the traffic set and calculate the benefit of successful transmission of each data stream: Q(f i ) is f i The profit of successful transmission, time now is the current time, time create (f i ) is f i The creation time of the corresponding data block, blcok remainsize (f i ) is f i The remaining size of the corresponding data block, P(f i ) is f i The priority of all data flows in the traffic set is calculated, and the sum of the benefits of successful transmission S_Q is calculated as follows: Finally, according to the benefits of successful transmission of each data stream, the congestion window size of each SCTP data stream in the traffic set is calculated: FPM_CWND(f i )=FPM_R(f i )×last_rtt(f i ), PPM_R(f i ) is f i The priority algorithm output rate, FPM_CWND(f i ) is f i The congestion window size output by the priority algorithm is L_R, which is the current remaining allocation rate of the traffic set. Its value is set to S_CR, and the congestion window size output by the priority algorithm is calculated and allocated to the corresponding SCTP data stream in the traffic set.

2. The system according to claim 1, characterized in that The data channel server further includes: The data channel construction device first completes the media description information SDP negotiation exchange with the UE, then exchanges the candidate address Candidate with the UE to establish a PeerConnection connection, and finally establishes a DataChannel connection with the UE.

3. The system according to claim 2, characterized in that The data channel construction device further includes: When the SDP negotiation unit receives the Offer SDP from the UE, it sets the remote SDP description information to the local by using the SetRemoteDescription method, then creates the Answer SDP by using the CreateAnswer method, and sets the answer to the local by using the setLocalDescription method, and sends the Answer SDP to the UE; The candidate switching unit monitors OnICECandidate, and when the ice candidate is available, it sends the candidate to the UE, and when receiving the candidate sent by the UE, it executes the AddICECandidate method to add the other party's network information to add the candidate to the local. The UE further comprises: The UE-SDP negotiation device creates an Offer SDP to the data channel server through the CreateOffer method, sets LocalDescription and sends it to the data channel server, then receives the AnswerSDP sent by the data channel server, and sets it to the local through SetRemoteDescription; The UE-Candidate switching device monitors OnICECandidate, and when an ice candidate is available, sends the candidate to the data channel server, and when receiving the candidate sent by the data channel server, executes AddICECandidate to add the candidate to the local.

4. The system according to claim 1, characterized in that The data channel server also includes: Reinforcement learning optimization device, build reinforcement learning model GCC, the input is the state space Among them, T t represents the throughput vector at time t, D t represents the inter-packet delay jitter vector at time t, R t represents the RTT vector at time t, L t Represents the packet loss rate vector at time t, and the output is the predicted sending rate. The reward function reward takes throughput as positive feedback, and delay and packet loss rate as negative feedback: reward = α*throughput + β*delay + γ*packet loss rate, where α, β, and γ are the weight coefficients of throughput, delay, and packet loss rate, respectively, representing the impact of throughput, delay, and packet loss rate on reward, and α>0, β<0, γ<0. A fully connected network is used for feature extraction, the number of fully connected layers is 2, and the number of neurons in each layer is set to 64.

5. The system according to claim 1, characterized in that The priority control device detects a SCTP data flow f in the traffic concentration when monitoring a When the congestion window size CWND changes, f a The current CC_CWND(f a ) and last_rtt(f a ) is sent to FPM, FPM also includes: Congestion window adjustment unit, updates the locally saved f a last_rtt(f a ), calculate f a The output rate CC_R(f a ): Then query f a Belong to the traffic set, adjust f a The current total output rate S_CR′ of the traffic set to which it belongs: S_CR′=S_CR+CC_R(f a )_FPM_R(f a ), FPM_R(f a ) is FPM for f a The output rate of the priority algorithm last allocated, and then f a The value of the current remaining allocation total rate L_R of the traffic set to which it belongs is updated to S_CR′, and f is recalculated a The congestion window size output by the priority algorithm of each SCTP data flow in the traffic set: FPM_CWND(f j )=FPM_R(f j )×last_rtt(f j ), f j Yes a The jth SCTP data flow in the traffic set belongs to, and then the congestion window size calculated by the priority algorithm output is assigned to f a The corresponding SCTP data stream in the traffic set.

6. A method for implementing a data channel server for an IMS network, characterized in that: Included are: Step 1: Each mobile terminal UE and the data channel server establish a data channel based on the SCTP protocol; Step 2: The data channel server receives the network information feedback packet sent by each UE, and then uses both the delay-based congestion control algorithm and the loss-based congestion control algorithm to calculate the corresponding sending rate, and selects a smaller sending rate to send the SCTP data stream to the UE. The data channel server includes a priority control device and a priority management device FPM. When the priority control device finds that multiple SCTP data streams pass through the same network path, it also includes: Step B1: The priority control device registers multiple SCTP data streams passing through the same network path to FPM, and inputs the congestion window size and network round-trip delay of each SCTP data stream: CC_CWND(f i ) and last_rtt(f i ), where f i Indicates the i-th SCTP data stream, CC_CWND(f i ) is f i The congestion window size, last_rtt(f i ) is f i The network round trip delay Step B2: FPM forms a traffic set with all SCTP data flows passing through the same network path, and saves the network round trip delay of each SCTP data flow: last_rtt(f i ), and then calculate the output rate of each SCTP data flow in the traffic set: CC_R(f i ) is f i The output rate of the flow set is calculated based on the current total output rate S_CR: Where N is the total number of SCTP data flows in the traffic set; Step B3, FPM sets the priority of each SCTP data stream in the traffic set, and then calculates the benefit of successful transmission of each data stream: Where Q(f i ) is f i The profit of successful transmission, time now is the current time, time create (f i ) is f i The creation time of the corresponding data block, blcok remainsize (f i ) is f i The remaining size of the corresponding data block, P(f i ) is f i The priority of the data flow is used to calculate the sum of the benefits of successful transmission of all data flows in the traffic set S_Q: Step B4: FPM calculates the congestion window size of each SCTP data flow in the traffic set according to the revenue of successful transmission of each data flow: FPM_CWND(f i )=FPM_R(f i )×last_rtt(f i ), where FPM_R(f i ) is f i The priority algorithm output rate, FPM_CWND(f i ) is f i The congestion window size output by the priority algorithm is L_R, which is the current remaining allocation rate of the traffic set. Its value is set to S_CR, and then the congestion window size output by the calculated priority algorithm is allocated to the corresponding SCTP data flow in the traffic set.

7. The method according to claim 6, characterized in that Step 1 includes: Step A1: UE and data channel server complete the negotiation and exchange of media description information SDP; Step A2: The UE and the data channel server exchange candidate addresses Candidate, and then establish a PeerConnection connection; Step A3: The UE and the data channel server establish a DataChannel connection.

8. The method according to claim 7, characterized in that Step A1 includes: Step A11: The UE creates an Offer SDP to the data channel server through the CreateOffer method, sets LocalDescription, and sends it to the data channel server; Step A12: The data channel server receives the Offer SDP from the UE and sets it locally through the method SetRemoteDescription to set the remote SDP description information; Step A13: The data channel server creates Answer SDP by remotely creating the answer using the CreateAnswer method, sets the answer to the local side using the setLocalDescription method, and sends the Answer SDP to the UE. Step A14: UE receives the Answer SDP sent by the data channel server and sets it locally via SetRemoteDescription. Step A2 includes: The data channel server or UE listens to OnICECandidate, and when the ice candidate is available, the candidate is sent to the UE or the data channel server. The UE or the data channel server executes the AddICECandidate method to add the other party's network information to add the candidate to the local.

9. The method according to claim 6, characterized in that The reinforcement learning optimization algorithm GCC is used to adjust the parameters of the delay-based congestion control algorithm or the loss-based congestion control algorithm, including: Construct a reinforcement learning model GCC, the input is the state space Among them, T t represents the throughput vector at time t, D t represents the inter-packet delay jitter vector at time t, R t represents the RTT vector at time t, L t Represents the packet loss rate vector at time t, and the output is the predicted sending rate. The reward function reward takes throughput as positive feedback, and delay and packet loss rate as negative feedback: reward = α*throughput + β*delay + γ*packet loss rate, where α, β, and γ are the weight coefficients of throughput, delay, and packet loss rate, respectively, representing the impact of throughput, delay, and packet loss rate on reward, and α>0, β<0, γ<0. A fully connected network is used for feature extraction, the number of fully connected layers is 2, and the number of neurons in each layer is set to 64.

10. The method according to claim 6, characterized in that When the congestion window size CWND of an SCTP data flow in the traffic set changes, the congestion window size CWND of the data flow f a For example, it also includes: Step C1: The priority control device sets f a The current CC_CWND(f a ) and last_rtt(f a )Send to FPM; Step C2, FPM updates the locally saved f a last_rtt(f a ), and calculate f a The output rate CC_R(f a ): Then query f a Belong to the traffic set, adjust f a The current total output rate S_CR′ of the traffic set to which it belongs: S_CR′=S_CR+CC_R(f a )-FPM_R(f a ), where FPM_R(f a ) is FPM for f a The last assigned priority algorithm output rate; Step C3, FPM converts f a The value of the current remaining allocation total rate L_R of the traffic set to which it belongs is updated to S_CR′, and f is recalculated a The congestion window size output by the priority algorithm of each SCTP data flow in the traffic set: FPM_CWND(f j )=FPM_R(f j )×last_rtt(f j ), where f j Yes a The jth SCTP data flow in the traffic set belongs to, and then the congestion window size calculated by the priority algorithm output is assigned to f a The corresponding SCTP data stream in the traffic set.