A hierarchical multimedia transmission system

Abstract

The present invention relates to the field of multimedia transmission technology, and in particular, proposes a hierarchical multimedia transmission system. It aims to solve the challenges of delay, jitter and packet loss faced by multimedia data in network transmission. The system divides multimedia data into different priorities with a flexible time slice division mechanism, adopts differentiated exchange strategies, and realizes hierarchical service quality assurance of multimedia data. High-priority time slices adopt direct exchange based on connection matrix to minimize transmission delay; ordinary priority time slices adopt cache-based storage-forwarding mode for scheduling exchange, taking into account fairness and delay performance. In addition, the present invention also proposes a flexible heterogeneous network intercommunication mechanism and a high-precision time slice synchronization method to support end-to-end service quality assurance. Compared with traditional solutions, the hierarchical exchange system can better meet differentiated multimedia business needs, improve network real-time performance, reduce delay jitter, improve transmission efficiency, and has good application prospects.

Description

Technical Field

[0001] The present invention relates to the technical field of multimedia transmission, and in particular to a hierarchical multimedia transmission system. Background Art

[0002] With the rapid development of Internet technology, the demand for multimedia data such as audio, video, and images to be transmitted over the network is growing. Multimedia data has strict requirements on network performance indicators such as transmission bandwidth, latency, and packet loss rate. Traditional IP networks face many challenges in providing multimedia services. On the one hand, the huge volume of multimedia data places high demands on network bandwidth; on the other hand, multimedia services are usually real-time, requiring the network to provide stable low latency and low packet loss rate guarantees. In addition, multimedia data of different types and contents also have different service quality requirements.

[0003] In order to meet the challenges of network transmission in the multimedia era, academia and industry are actively exploring innovative network architectures and technical solutions. Typical practices include: building an overlay network for multimedia transmission on top of the IP network, such as CDN; researching more efficient streaming media transmission protocols, such as QUIC; introducing SDN, and flexibly adapting multimedia traffic diversion strategies through separation of control and forwarding, centralized flow table rule issuance, etc.; in new generation network standards such as 5G, native support for network slicing, edge computing and other functions, in order to optimize the multimedia experience from the network architecture level.

[0004] Although the industry has proposed a variety of solutions to address multimedia challenges, there are still many limitations in practice. Traditional IP networks are essentially designed for best-effort data transmission and lack a mechanism to ensure communication quality. When multimedia traffic is mixed with ordinary data traffic, it is easy for them to affect each other and compete for resources. In addition, the granularity of IP packet switching is large, making it difficult to fine-tune multimedia data management and control.

[0005] Network solutions dedicated to multimedia transmission, such as fiber-optic private networks, can provide stable low latency and high-quality transmission, but they are expensive and have poor scalability, making them difficult to deploy on a large scale. In addition, such private networks usually lack universality and are difficult to interconnect with the Internet.

[0006] Multimedia overlay networks and optimized transmission protocols based on IP have made up for the shortcomings of IP networks to a certain extent, but it is difficult to fundamentally solve the problem of mutual interference between multimedia and ordinary data. Moreover, most solutions only focus on specific application scenarios and lack versatility and flexibility. Summary of the invention

[0007] In view of the limitations of existing solutions, the present invention proposes an innovative hierarchical multimedia switching system. The system combines the advantages of circuit switching and packet switching, and uses a flexible time slice division mechanism to divide multimedia data into different priorities for transmission and exchange. Compared with traditional solutions, the main innovations are as follows:

[0008] (1) The integrated time slicing, slice grouping, and data packet transmission methods can meet the differentiated multimedia packet performance requirements;

[0009] (2) Delayed, quality-determined switching, compatible with efficient, multi-type packet switching;

[0010] (3) Determine scalability based on the delay of time recovery.

[0011] The hierarchical multimedia switching system proposed in the present invention aims to solve the technical challenges of delay, jitter and packet loss faced by multimedia data transmission from the perspective of network switching. Based on a flexible time slice division mechanism, the system loads multimedia data of different priorities into corresponding time slices for transmission, and adopts differentiated switching strategies for time slices of different levels, thereby achieving hierarchical service quality assurance for multimedia data.

[0012] like Figure 1 As shown, the system is mainly composed of three modules: multimedia data segment transmission module, switching module, and expansion module.

[0013] For the multimedia data slice transmission module, multimedia data is first divided into fixed-length time slices, and time slices of different priorities form slice groups. In each slice group, high-priority time slices enjoy the privilege of preempting transmission, while low-priority time slices compete for remaining bandwidth resources. This hierarchical transmission of time slices can not only meet the high-demand multimedia services, but also ensure the basic communication quality of ordinary data.

[0014] The switching module is designed with high performance and low latency as the goal. For high-priority time slices, the system directly performs switching based on the connection matrix, while other time slices are cached in queues and packet switched through scheduling algorithms. This ensures the real-time performance of high-quality traffic while taking into account the system's switching efficiency for various types of traffic.

[0015] In the expansion module, the system adopts a lossless expansion solution based on time recovery. When it is necessary to connect to an external network, the high-quality time slices are extracted and converted into IP packets, which are sent to the backbone network for transmission; after reaching the destination node, they are restored to the original time slice form according to the timestamp information. Ordinary data traffic is directly forwarded in the form of packets. In this way, the system can achieve seamless expansion between different network standards.

[0016] Specifically:

[0017] The hierarchical multimedia switching system mainly includes the following three modules:

[0018] (1) Multimedia data fragmentation transmission module:

[0019] This module is responsible for slicing multimedia data streams, including: division of fixed time slices, composition of multi-priority slice groups, intra-slice time slice preemption mechanism, data packetization and padding, etc.

[0020] Assume the length of the time slice is , then The time interval corresponding to the time slice is Multiple time slices form a slice group, and the slice group length is The system divides multimedia data into performance-oriented data with strong real-time performance and general capacity-oriented data, which correspond to high priority and normal priority respectively.

[0021] For performance data, the system timestamps it at the beginning of each slice group and loads it first to ensure high-quality transmission; while capacity data is filled into the remaining time slices and transmitted as much as possible. Under this preemption mechanism, performance data can obtain stable bandwidth and latency guarantees.

[0022] (2) Switching module:

[0023] The core of the switching system is to achieve efficient and low-latency switching of multimedia data. Two different switching modes are used for different levels of time slices:

[0024] For performance time slices, direct switching based on the connection matrix is ​​performed. The control plane calculates the connection matrix of each switching node in advance according to the network topology and business requirements, and sends it to the data plane. The high-priority time slices that arrive at the switching node will directly select the appropriate output port through the connection matrix without waiting in line, thereby minimizing the switching delay.

[0025] For capacity-based time slices, a packet-based store-and-forward method is used for exchange. The arriving packets are first extracted for valid data and cached, and then sent to the scheduling module for queuing. The scheduling module comprehensively considers factors such as fairness and latency, and selects appropriate packets for forwarding. Compared with direct exchange, this method can improve the system's ability to support multiple services.

[0026] (3) Expansion module:

[0027] The present invention utilizes the expansion module to realize the smooth expansion and interconnection of the network, and the key of the expansion module is the conversion between time slices and IP packets.

[0028] When the system needs to connect to an external IP network, it extracts data from high-priority time slices, adds timestamp information, and repackages it into IP packets for transmission in the backbone network. After several hops, the packets arrive at the destination node and are restored to time slice form, and inserted into the new slice group according to the timestamp information, completing end-to-end transmission.

[0029] Ordinary data packets can be directly encapsulated into IP packets and sent, and delivered to the receiving node in the form of packets after arrival. In this way, seamless connectivity can be achieved between heterogeneous networks without introducing additional delays.

[0030] The specific working process of each module is further explained below.

[0031] The working process of the multimedia data segment transmission module is as follows:

[0032] (1) Fixed time slices and time slice groups:

[0033] Multimedia data is transmitted at fixed time intervals Divide into time slices of equal length. Adjacent time slices are numbered satisfy:

[0034] ;

[0035] in is the absolute time of the system. It can be seen that the time slice sequence It has a discrete linear mapping relationship with the time axis.

[0036] Continuous A slice group :

[0037] ;

[0038] Length of the film group satisfy: ;

[0039] (2) Prioritization and preemption within a time slice:

[0040] Each time slice Divided into two parts: performance time slice and capacity time slice , used to transmit performance data and capacity data respectively.

[0041]

[0042] Performance time slice It has a higher priority and is allowed to interrupt the transmission of capacity time slices. Therefore, performance-oriented services can seize bandwidth resources at the beginning of the slice group and obtain a guaranteed transmission channel. However, capacity-oriented services can only use the remaining bandwidth to transmit data packets.

[0043] Set a special scheduling time slice at the beginning of each slice group The scheduling time slice is responsible for collecting resource applications from performance-oriented services and allocating time slices. The entire preemption process is shown in Algorithm 1 below.

[0044] Algorithm 1: Time slice preemption algorithm. The input of this algorithm is: performance business set , capacity service set and the length of the group ; Its output is: time slice allocation result.

[0045] The flowchart of Algorithm 1 is as follows Figure 2 The specific process is shown in the following pseudo-code:

[0046] 1. Initialize the time slice allocation vector ;

[0047] 2.foreach :

[0048] 3. / / Calculation business The number of time slices required;

[0049] 4.if : / / The remaining performance time slice is sufficient;

[0050] 5. Medium Mark performance time slices and updates the total number of allocated performance time slices ;

[0051] 6.else:

[0052] 7. Mark all remaining time slices as performance time slices and exit the loop;

[0053] 8.foreach :

[0054] 9. The packets are filled into the unmarked time slices in sequence;

[0055] Among them, the constant Represents the transmission capacity of a single time slice. Algorithm 1 first counts the time slice requirements of performance services and allocates resources as much as possible; if resource competition occurs, the service quality of existing services is prioritized and the scheduling of capacity service data is suspended.

[0056] (3) Data subpackaging and backlog processing

[0057] For data frames that cannot be accommodated in a single time slot, they need to be transmitted in fragments. The granularity of the fragments should not exceed the payload length of the time slot.

[0058] When filling data frames for performance services, the fragments should be as continuous as possible to avoid spreading them across multiple slice groups and introducing additional transmission delays. If the current slice group cannot accommodate a complete data frame, the system chooses to shelve the extra fragments and wait for the next slice group to start before transmitting, rather than blindly filling and causing the frame boundary to cross the slice group. This strategy can reduce inter-frame delay jitter, thereby ensuring the QoE of performance services.

[0059] For capacity services, the system should monitor the amount of data backlog to prevent continuous data accumulation. When the backlog exceeds the set threshold, the system automatically discards some packets to avoid further deterioration of latency.

[0060] The specific working process of the switching module is as follows:

[0061] (1) Direct exchange of performance time slices:

[0062] For performance time slices, this system adopts a direct exchange method based on a connection matrix to avoid queuing and minimize transmission delay.

[0063] Assume the system contains input ports and output ports, then the connection matrix It can be defined as:

[0064] ,in, Indicates input port To output port The connection status.

[0065] when When a dedicated channel is established between the two ports, the input port The high priority time slice will be sent directly to the output port , no additional exchange processing is required.

[0066] Connection Matrix The configuration is calculated by the control plane based on the network topology and business requirements. Specifically, the control plane calculates the optimal connection matrix of each switching node through a centralized flow table and sends it to the data plane through a southbound interface (such as the OpenFlow protocol).

[0067] (2) Cache exchange of capacity-based time slices:

[0068] For data packets carried by capacity-based time slices, the system uses a cache-based store-and-forward mechanism for switching. The core of the switching system is the packet cache and scheduling module.

[0069] When a packet arrives at the input port, the system extracts the payload data from the capacity time slice and stores the complete packet in the cache queue. The scheduling module is responsible for selecting the appropriate packet from the queue and sending it to the destination output port.

[0070] The design of the scheduling strategy needs to take into account both latency performance and fairness. On the one hand, the first packet should be transmitted first to reduce the overall latency; on the other hand, the problem of individual services occupying resources for a long time should be avoided. The system uses Algorithm 2 to solve this problem. Its input is the input port set , the cache queue of each port , the weight vector of each queue and scheduling slot length ; Its output is the selected group and its destination port.

[0071] The flowchart of Algorithm 2 is as follows Figure 3 The specific working process is shown in the following pseudo-code:

[0072] 1. Initialize the time slot counter ,make Indicates the last scheduling time;

[0073] 2.while : / / within a scheduling time slot;

[0074] 3.for to : / / Traverse all input ports;

[0075] 4.if Non-empty: / / queue There is a waiting group inside;

[0076] 5. Take the first team to group , remember its destination port is ;

[0077] 6.if port idle:

[0078] 7. Order ,send , and returns ;

[0079] 8.else:

[0080] 9. Continue the cycle;

[0081] 10. ;

[0082] 11. Return NULL;

[0083] Algorithm 2 checks the head of each port in turn in a scheduling time slot. Once a packet is found to be available for transmission, the loop stops immediately and the packet is returned. If no available packet is found after the time slot ends, an empty result is returned and the next round of scheduling is waited for. The weight vector is introduced , used to control the scheduling priority of each queue. Weight The larger the value, the higher the port The more packets are dispatched, the easier it is to be scheduled first. The system can dynamically adjust the weights according to factors such as business type and service level to achieve differentiated QoS guarantees. The status of the output port is also considered. The system only performs the switching operation when the destination port of the packet is idle, otherwise it continues to check other queues to avoid unnecessary packet backlogs.

[0084] The specific working process of the expansion module is as follows:

[0085] (1) Conversion between performance time slice and IP protocol:

[0086] When the system needs to communicate with an external IP network, the performance time slice will trigger the protocol conversion mechanism. This mechanism encapsulates the time slice data into IP packets at the ingress node (IngressNode) and transmits it through the backbone network; it decapsulates and restores it to the time slice format at the egress node (EgressNode). The conversion process needs to record and restore the priority and timing information of the original time slice to ensure that the end-to-end QoS semantics remain unchanged.

[0087] Specifically, for the high priority time slice to be sent , the entry node first extracts its payload data , add serial number and timestamp Generate data unit:

[0088] ;

[0089] Then, the system will As upper layer data, it generates standard IP packets through encapsulation :

[0090] ;

[0091] Grouping Enter the backbone network for transmission, and arrive at the exit node after multi-hop routing. The exit node decapsulates get , recover the original time slice payload . Using the serial number and the local time reference, the egress node can calculate the target time at which the time slice should be inserted :

[0092] ,in is the slice size, is the slice group period, which can be synchronized between the inlet and outlet nodes in advance. Using the above formula, the system can accurately locate the time slice Through this conversion mechanism, networks of different standards can be seamlessly connected, and the transmission timing of high-priority services can be maintained.

[0093] (2) Expansion of capacity-based time slices

[0094] For ordinary data packets carried by capacity-based time slices, the forwarding node can directly extract the payload and assemble it into IP packets, which are then sent to the backbone network for transmission. After the packets arrive at the aggregation node, they are also delivered in the form of packets without special processing. The expansion of capacity-based services does not involve timing synchronization and is relatively simple to implement.

[0095] (3) Time synchronization and recovery

[0096] To support high-precision time slice restoration, the source node and the sink node need to synchronize time and establish a unified time reference. Synchronization can be achieved using a variety of means such as GPS, atomic clocks, and PTP (Precision Time Protocol). In addition, factors such as synchronization error and network jitter need to be considered, and a certain protection margin should be reserved to avoid service quality degradation caused by time slice misalignment.

[0097] In addition, in abnormal situations such as network failures, the continuity of time slices may be destroyed. Therefore, the system also has a time slice recovery mechanism, which can repair and retransmit time slices through redundant transmission, FEC, etc. to ensure uninterrupted services. BRIEF DESCRIPTION OF THE DRAWINGS

[0098] Figure 1 A system block diagram of a hierarchical multimedia transmission system is provided for the present invention.

[0099] Figure 2 Flow chart of Algorithm 1 provided by the present invention.

[0100] Figure 3 Flow chart of Algorithm 2 provided by the present invention. DETAILED DESCRIPTION

[0101] For example, we need to transmit two video files at the same time: one is a real-time video conferencing video, which has strict requirements on transmission delay and jitter and is a performance-based service; the other is video on demand, which is not sensitive to delay and is a capacity-based service.

[0102] Time slice division of multimedia data: First, we divide the video data into fixed time intervals Divide into time slices of equal length. , then in Time, the time slice number to which the video frame belongs for:

[0103] , if we A slice group is composed of consecutive time slices. The length of the slice group is for:

[0104]

[0105] Within each time slice, we further divide it into performance time slices and capacity time slice The data packets of real-time video conference are marked as performance-oriented and take priority. ; Video on demand packets are marked as capacity-based and can only be used .

[0106] Direct exchange of performance-based packets: For real-time video conferencing, packets are filled into performance time slices. Assume that At that moment, there was a video frames need to be transmitted, and the payload of a single time slice is , the frame can be completely loaded into a performance time slice.

[0107] The control plane sends forwarding rules for performance-based services to each switching node based on the network topology and routing calculation results. The forwarding rules include a connection matrix, such as:

[0108] ;

[0109] Indicates input port Connect directly to the output port , input port Connect directly to the output port .

[0110] When the performance time slice arrives at the switching node, the node directly switches the time slice to the target port without queuing according to the connection matrix, thereby minimizing the forwarding delay. For example, for the above connection matrix, assuming that Performance time slice of the moment Arrival at input port , the node immediately switches it to the output port , the delay is close to .

[0111] After multi-hop direct exchange, the performance time slice reaches the destination node. The destination node extracts the payload data and restores the original video frame, achieving low-latency transmission for real-time video conferencing.

[0112] Buffer switching of capacity-based data packets: For video-on-demand services, data packets are filled into capacity time slices. Different from performance time slices, capacity time slices need to be cached and scheduled in the switching node.

[0113] exist When there are two video-on-demand packets and Arrives at the two input ports of the switching node. The node maintains a cache queue for each port. , Store them in the corresponding queues respectively.

[0114] Then, the node's scheduling module executes the packet scheduling algorithm and selects appropriate packets for exchange based on the queue weight and packet waiting time.

[0115] Input: Input port set , the cache queue of each port , the weight vector of each queue , scheduling time slot length ;

[0116] Output: selected packets and their destination ports.

[0117] Algorithm steps:

[0118] 1. Initialize the time slot counter ,make Indicates the last scheduling time

[0119] 2.while : / / In a scheduling time slot

[0120] 3.for to : / / Traverse all input ports

[0121] 4.if Non-empty: / / queue There are waiting groups

[0122] 5. Take the first team to group , remember its destination port is

[0123] 6.if port idle:

[0124] 7. Order ,send , and returns

[0125] 8.else:

[0126] 9. Continue the cycle

[0127] 10.

[0128] 11. Return NULL

[0129] For example, let the weights of the two queues be , , indicating that the two queues have the same scheduling priority. Let the scheduling time slot length , that is, each scheduling cycle is the length of two time slices.

[0130] Assume that after a scheduling time slot, the node selects the group To exchange. is sent to its target output port, while Continue to wait for scheduling.

[0131] After multiple scheduling time slots, all capacity-based packets are cleared and finally reach the destination node. Compared with performance-based packets, capacity-based packets allow for certain delays and jitters, but as long as the cache and scheduling algorithms are designed reasonably, a better transmission experience can still be obtained.

[0132] Interconnection between different networks: If the source node and the destination node are located in different networks, protocol conversion is required through a gateway or border node.

[0133] For performance time slices, the gateway restores them to IP packets and carries timestamp information in the packet header. After the packets are transmitted through the backbone network to the gateway on the other side, the gateway reassembles the IP packets into performance time slices according to the timestamps and adds them to the new slice group for exchange. In this way, end-to-end delay guarantee and QoS mapping can be achieved.

[0134] For capacity time slices, the gateway directly extracts the payload data, encapsulates it into ordinary IP packets and transmits it in the backbone network. After the packets reach the gateway on the other side, they are reassembled into capacity time slices and inserted into new slice groups to participate in switching scheduling.

[0135] Through the above-mentioned hierarchical exchange and protocol conversion mechanism, networks of different standards can be seamlessly connected, and both high-priority real-time video and ordinary video-on-demand services can obtain corresponding service quality guarantees.

[0136] Synchronization mechanism: In order to ensure lossless transmission of performance time slices between heterogeneous networks, the source node and the destination node need to establish a synchronization mechanism. The synchronization information can be transmitted through a separate signaling channel or attached to the time slice payload.

[0137] The synchronization information should at least include: Length, Size, The cycle and time slice number After receiving the performance IP packet with a timestamp, the destination gateway extracts the synchronization information and maintains a timeslot sequence number generator that is the same as that of the source node:

[0138]

[0139] in is the local time of the gateway, is the start time after synchronization, and factors such as transmission delay and clock drift need to be considered. The gateway fills the extracted video data into the serial number The performance time slice of the video is added to the current slice group, thereby achieving lossless restoration of the time slice.

[0140] Capacity time slices do not need strict synchronization, they only need to be filled into the slice group according to the order in which the packets arrive.

[0141] Through the above steps, the hierarchical multimedia switching system can simultaneously transmit real-time video conferencing and video on demand services efficiently and with high quality. The core mechanism of the system is the combination of performance-oriented direct switching and capacity-oriented cache scheduling, coupled with time slice synchronization and protocol conversion, which ultimately achieves end-to-end service quality assurance and seamless interconnection.

[0142] The above is a detailed implementation example of the hierarchical multimedia switching method. This method fully considers the QoS differences of different services, and not only meets the low latency requirements of real-time multimedia services, but also takes into account the transmission efficiency of general data. Among them, the division and preemption mechanism of time slices, the connection matrix of direct switching, the weighted polling algorithm of capacity scheduling, and key technologies such as time synchronization and protocol conversion together constitute a complete solution.

[0143] The hierarchical multimedia switching system proposed in the present invention realizes refined priority division and resource isolation at the network layer in response to the differentiated QoS requirements of multimedia data. Compared with the traditional solution, the present invention has the following beneficial effects:

[0144] Through the time slice preemption mechanism and direct exchange, high-priority services obtain deterministic bandwidth guarantee and extremely low transmission delay, avoiding competition for resources with ordinary services. The flexible time slice cutover mechanism allows QoS differentiation of different granularities to meet the diverse needs of multimedia services.

[0145] For performance-based time slices, the system directly exchanges based on the connection matrix to minimize queuing delays; for capacity-based packets, the system combines cache exchange scheduling to improve throughput while ensuring fairness. The combination of store-and-forward and direct exchange takes into account the support of both real-time traffic and ordinary traffic, and improves overall efficiency.

[0146] Supporting flexible scalability, the system realizes seamless interconnection of heterogeneous networks by introducing protocol-independent time slice abstraction. High-precision time synchronization mechanism ensures QoS of cross-network transmission. Flexible expansion mode supports backbone network connection and provides high-quality services nearby, which is convenient for realizing multi-level coverage.

[0147] The hierarchical switching solution of the present invention has a good balance of performance indicators in multiple aspects such as service quality, system efficiency, and expansion capability, and provides a new solution for the construction of the next generation multimedia network.

Claims

1. A hierarchical multimedia transmission system, characterized in that: The system includes a multimedia data slice transmission module, a switching module and a capacity expansion module; The multimedia data slice transmission module is responsible for slicing the multimedia data stream, including: the division of fixed time slices, the composition of multi-priority slice groups, the intra-slice time slice preemption mechanism, data subpackaging and filling; The switching module realizes efficient and low-latency switching of multimedia data. The system adopts two switching modes for different levels of time slices. The expansion module realizes smooth expansion and interconnection of the network; The specific working process of the multimedia data segment transmission module includes: (1) Fixed time slices and time slice groups; (2) Prioritization and preemption within time slices; (3) Data packetization and backlog processing; The fixed time slices and time slice groups specifically include: Multimedia data is transmitted at fixed time intervals Divided into time slices of equal length, adjacent time slices are numbered satisfy: ,in is the absolute time of the system, the time slice sequence It has a discrete linear mapping relationship with the time axis; Continuous A slice group : ； Length of the film group satisfy: ; The priority division and preemption within the time slice specifically include: Each time slice Divided into two parts: performance time slice and capacity time slice , used to transmit performance data and capacity data respectively, ;Performance time slice It has a higher priority and is allowed to interrupt the transmission of capacity time slices. Capacity services can only use the remaining bandwidth to transmit data packets. Set a special scheduling time slice at the beginning of each slice group ,The scheduling time slice is responsible for collecting the resource applications of ,performance-oriented services and allocating time slices. The entire preemption process is completed according to ,Algorithm 1; The input of Algorithm 1 is: performance business set , capacity service set and the length of the group ; Its output is: time slice allocation result, the specific process is as follows: Initialize the time slice allocation vector ; for each : ； if : exist Medium Mark performance time slices and updates the total number of allocated performance time slices ; else: Mark all remaining time slices as performance time slices and exit the loop; for each : Will The packets are filled into the unmarked time slices in sequence; Among them, the constant Indicates the transmission capacity of a single time slice; The data subpackaging and backlog processing specifically include: For data frames that cannot be accommodated in a single time slice, they are transmitted in fragments, and the granularity of the fragments should not exceed the payload length of the time slice; When filling data frames of performance services, the fragments are kept as continuous as possible to avoid introducing extra transmission delay by spreading them across multiple slice groups. If the current slice group cannot accommodate a complete data frame, the system chooses to shelve the extra fragments and wait for the next slice group to start before transmitting. For capacity services, the system should monitor the amount of data backlog to prevent continuous data accumulation. When the backlog exceeds the set threshold, the system automatically discards some packets. The specific working process of the switching module includes: (1) direct switching of performance-based time slices; (2) cache switching of capacity-based time slices.

2. The system according to claim 1, characterized in that The specific working process of the expansion module includes: (1) conversion of performance-type time slices and IP protocols; (2) expansion of capacity-type time slices; and (3) time synchronization and recovery.

3. The system according to claim 1, characterized in that The direct exchange of the performance time slice is specifically as follows: Adopt direct switching based on connection matrix to avoid queuing and minimize transmission delay; Assume the system contains input ports and output ports, then the connection matrix Defined as: ,in, Indicates input port To output port The connection status; when When a dedicated channel is established between the two ports, the input port The high priority time slice will be sent directly to the output port , no additional exchange processing is required; Connection Matrix The configuration is calculated by the control plane based on the network topology and business requirements. Specifically, the control plane calculates the optimal connection matrix of each switching node through the centralized flow table and sends it to the data plane through the southbound interface. The cache exchange of the capacity-based time slice is specifically as follows: The cache-based store-and-forward mechanism is used for switching. When a packet arrives at the input port, the payload data is extracted from the capacity time slice, and the complete packet is stored in the cache queue. The appropriate packet is selected from the queue and sent to the destination output port. This is done using Algorithm 2. The input of Algorithm 2 is: input port set , the cache queue of each port , the weight vector of each queue and scheduling slot length ; The output of Algorithm 2 is: the selected group and its destination port; The specific working process is as follows: Initialize the time slot counter ,make Indicates the last scheduling time; while : for to : if non-empty: Take the first team group , remember its destination port is ; If port idle: make ,send , and returns ; else: Continue the cycle; ； Returns NULL.

4. The system according to claim 2, characterized in that The conversion between the performance time slice and the IP protocol specifically includes: When the system needs to communicate with an external IP network, the time slice data is encapsulated into IP packets and transmitted through the backbone network. At the egress node, it is decapsulated and restored to the time slice format. The conversion process needs to record and restore the priority and timing information of the original time slice to ensure that the end-to-end QoS semantics remain unchanged. Specifically, for the high priority time slice to be sent , the entry node first extracts its payload data , add serial number and timestamp Generate data unit: ； Then, the system will As upper layer data, it generates standard IP packets through encapsulation : ； Grouping Enter the backbone network for transmission, and after multi-hop routing, reach the exit node, where it is decapsulated get , recover the original time slice payload , using the serial number and the local time reference, the egress node calculates the target time at which the time slice should be inserted : ,in is the slice size, It is the slice group period, which is synchronized between the ingress and egress nodes in advance; The expansion of the capacity-based time slice is specifically as follows: for ordinary data packets carried by the capacity-based time slice, the forwarding node directly extracts the payload and assembles it into an IP packet, which is sent to the backbone network for transmission. After the packet reaches the aggregation node, it is delivered in the form of a packet; The time synchronization and recovery are specifically as follows: the source node and the destination node perform time synchronization, establish a unified time reference, and use a variety of means including GPS, atomic clocks, PTP to achieve synchronization, reserve a protection margin, and avoid time slice misalignment causing service quality degradation; through redundant transmission and FEC, the time slice is repaired and retransmitted to ensure uninterrupted service.

Images