Search

CN-122027819-A - Self-adaptive code rate 5G video stream edge intelligent regulation and control system

CN122027819ACN 122027819 ACN122027819 ACN 122027819ACN-122027819-A

Abstract

The invention discloses a 5G video stream edge intelligent regulation and control system with a self-adaptive code rate, which relates to the technical field of edge video regulation and control, and comprises a global snapshot modeling module, a hierarchical cache arrangement module, a version closed-loop correction module and an experience constraint distribution module, wherein the snapshot module collects link observation, edge resource allowance, terminal session state and interaction event count, calculates a rate matrix and a time delay matrix, decomposes a propagation item and a transcoding item of end-to-end live broadcast time delay, and encapsulates a state snapshot. The cache arrangement module analyzes the layering request, divides the shared layer and the non-shared layer, generates a cache plan, a scheduling plan and a code rate version table under the constraints of capacity, availability and interlayer dependence, and executes the level rollback synchronization correction code rate when the resources are insufficient. The closed loop correction module completes unique allocation of the terminal to the edge server and corrects the code rate under the constraint of bandwidth and calculation force. The experience constraint allocation module performs hard verification, conflict localization, least disturbance delta replacement, and versioning rollback output.

Inventors

  • LIU QIANG
  • SUN HUIHUI
  • Lin Gushan

Assignees

  • 北京微呼科技有限公司
  • 福克斯韦尔(深圳)科技有限公司

Dates

Publication Date
20260512
Application Date
20260414

Claims (10)

  1. 1. The intelligent regulation and control system for the 5G video stream edge of the self-adaptive code rate is characterized in that the modules are connected through signals; the global snapshot modeling module is used for collecting link observation, edge resource allowance, terminal session state and interaction event count, calculating link speed and propagation delay, constructing a speed matrix and a time matrix, decomposing the live delay into a plurality of sections of propagation items and transcoding items, establishing capacity constraint, availability constraint and interlayer dependency constraint of layered content, packaging into a state snapshot, and generating a candidate placement queue; The hierarchical cache arrangement module analyzes the hierarchical requests based on the state snapshot, counts concurrent request sets, divides a shared layer and a non-shared layer, calculates the scheduling relation between a cache placement scheme and the shared layer under the constraint of capacity, availability and dependence, performs feasibility verification on the bandwidth resources of the edge and the computing resources, performs hierarchical rollback and synchronously corrects the code rate version table when the bandwidth resources of the edge and the computing resources of the edge are not met, enables placement, scheduling and code rate decision to be consistent, and outputs a cache plan, a scheduling plan and a code rate version result; The version closed loop correction module forms an interactive experience item by using a live broadcast total time delay and a delay threshold value, forms a video quality item by using an expected code rate and an issued code rate, forms a code rate jitter loss item by using a historical average code rate, forms a switching loss item by using a connection state difference, selects unique allocation from a candidate edge server set for each terminal under the double constraint of bandwidth and computing power, carries out code rate self-adaptive correction and alternative server rollback, and outputs a terminal allocation table, a terminal code rate configuration table and a server resource surplus table; And the experience constraint distribution module is used for triggering the actual throughput, playing buffering, propagation delay, resource surplus and cache hit statistics of the complementary acquisition according to a scheduling period or an event, firstly executing hard verification of the resource constraint and the cache constraint, positioning a conflict set, carrying out incremental replacement on cache placement according to a minimum disturbance strategy on the premise of feasibility, then linking the re-decision distribution and the code rate, recording switching punishment, forming versioning output, and returning an effective version to a previous version and retaining traceable archiving when a rollback condition is met.
  2. 2. The adaptive bitrate 5G video stream edge intelligent regulation system of claim 1, wherein the global snapshot modeling module is configured to: establishing an edge server set and a terminal set, and determining an edge server candidate set which can be connected with each terminal for each terminal; abstracting the live video stream into a regulable code rate set, transcoding overhead, propagation delay and hierarchical content joint model; calculating the transmission rate and the transmission delay between nodes by using a preset algorithm based on the link observation parameters to form a rate matrix and a delay matrix; The end-to-end live delay when the terminal connects to the edge server is modeled as the sum of the multi-segment propagation delay and the transcoding delay.
  3. 3. The adaptive bitrate 5G video stream edge intelligent regulation system of claim 2 wherein the global snapshot modeling module is further configured to: Based on the interaction event count of the terminal in the sliding time window, obtaining an interaction strength index through normalization mapping, and taking the interaction strength index as interaction experience modeling input; Expressing video content as a schedulable hierarchical unit, and introducing a cache device set and capacity constraint thereof; establishing a cache capacity constraint, a hierarchical object availability constraint and an interlayer dependency constraint of the hierarchical content to form a supply side feasible region; and generating a candidate placement queue containing object identifiers, layer identifiers and priority key values according to the splitting rules of the shared and non-shared layers.
  4. 4. The adaptive bitrate 5G video stream edge intelligent regulation system of claim 1, wherein the hierarchical buffering arrangement module is configured to: analyzing layering requests of all terminals based on the state snapshot, determining the highest layer number required, and summarizing to obtain a request set; Dividing a video layer into a non-shared layer set and a shared layer set according to the number of request terminals, and constructing an ordered processing queue; The cache placement scheme is solved under the constraints of capacity, availability and layer dependence with the goal of minimizing the weighted sum of perceived latency and storage cost.
  5. 5. The adaptive bitrate 5G video stream edge intelligent regulation system of claim 4 wherein the layered buffering arrangement module is further configured to: For the non-sharing layer, selecting a node with the minimum comprehensive cost from the candidate equipment set meeting the residual cache capacity for placement; For the sharing layer, an alternating greedy method is adopted, a duplicate node set is determined firstly, and then a unique supply node is allocated to each request terminal; And carrying out the feasibility check of the bandwidth and the computing resources on each edge server, and if the feasibility check is not satisfied, executing the hierarchical rollback on the affected terminal and synchronously correcting the code rate version table.
  6. 6. The adaptive bitrate 5G video stream edge intelligent regulation system of claim 1, wherein the version closed loop correction module is further configured to: Constructing a terminal experience quality function, wherein the experience quality function comprises an interactive experience item based on a live broadcast total time delay and a delay threshold, a video quality item based on a desired code rate and a downlink code rate, a code rate jitter loss item based on a historical average code rate and a switching loss item based on a connection state difference; And selecting an edge server for each terminal and determining the code rate under the bandwidth constraint and the calculation constraint aiming at maximizing the sum of the experience quality of all terminals.
  7. 7. The adaptive bitrate 5G video stream edge intelligent regulation system of claim 6, wherein the version closed loop correction module is further configured to: Searching lower code rates in descending order in the code rate level set to meet the constraint when the selected server bandwidth is insufficient; when the computing resources of the selected server are insufficient, searching higher code rates in the code rate level set in an ascending order to reduce transcoding overhead; If the code rate correction still cannot meet the constraint, the current terminal is removed from the server candidate list, and a server with suboptimal experience quality is selected for distribution.
  8. 8. The adaptive bitrate 5G video stream edge intelligent regulation and control system according to claim 1, wherein the experience constraint allocation module is configured to: triggering according to a period or an event, and supplementing observed quantity of a terminal side, a link side, an edge side and a cache side to form a current round of observed vector; Based on the observation vector, performing hard verification on the edge bandwidth, the computing resource, the cache capacity and the layer dependency constraint, and positioning a conflict set; On the premise of meeting the hard constraint, the buffer storage placement scheme is subjected to incremental correction, and then terminal allocation and code rate decision are re-executed based on the corrected state.
  9. 9. The adaptive bitrate 5G video stream edge intelligent regulation system of claim 8, wherein the experience constraint allocation module is further configured to: identifying a hot spot layer and a low-yield layer according to request statistics and cache observed quantity, and carrying out local replacement on a cache placement scheme, wherein the hot spot layer adopts a multi-source placement and sharing layer strategy, and the non-hot spot layer adopts a single-source placement and non-sharing layer strategy; The re-decision strategy comprises the steps of introducing switching punishment when a server which enables the quality of experience to be maximum is selected for the terminal, and adjusting the code rate while keeping the connection state unchanged when the quality of experience difference is smaller.
  10. 10. The adaptive bitrate 5G video stream edge intelligent regulation system of claim 8, wherein the experience constraint allocation module is further configured to: distributing version numbers for each iteration, and packaging and outputting a cache placement version, a distribution and code rate version and a resource snapshot version; When the new version causes the overall target amount to drop beyond a threshold, or the switch and rate drop penalty increases beyond a threshold, or the cache target amount worsens beyond a threshold, the validated version is rolled back to the previous version and a traceable archive is persisted.

Description

Self-adaptive code rate 5G video stream edge intelligent regulation and control system Technical Field The invention relates to the technical field of edge video regulation and control, in particular to a 5G video stream edge intelligent regulation and control system with a self-adaptive code rate. Background With the development of fifth generation mobile communication networks, edge computing and content distribution technologies, the interactive video streaming service generally needs to cooperatively complete transmission, buffering, scheduling and code rate adjustment at the network side and the edge side. In order to adapt to link fluctuation and multi-terminal concurrent access, information such as link observation, edge resource allowance, terminal session state and the like is gradually introduced in engineering, and content layering supply relation and resource constraint are uniformly modeled by combining edge caching, transcoding processing and end-to-end time delay decomposition, so that computable state input is provided for subsequent edge side allocation and code rate regulation. In the existing scheme, code rate self-adaption is usually based on single link throughput or client strategy, constraint such as propagation delay, transcoding overhead, cache availability, interlayer dependence, edge bandwidth, computing power and the like are lack of uniform expression, so that inconsistency easily occurs among cache placement, scheduling relation and final code rate selection, and when network state, request concurrency or resource occupation changes, hard constraint verification oriented to conflict positioning, minimum disturbance correction of incremental replacement and traceable versioning output and rollback mechanism are lack, so that sustainable iterative edge regulation closed loop is difficult to form. The present invention proposes a solution to the above-mentioned problems. Disclosure of Invention In order to overcome the above-mentioned drawbacks of the prior art, an embodiment of the present invention provides a 5G video stream edge intelligent regulation and control system with adaptive code rate, so as to solve the problems set forth in the above-mentioned background art. In order to achieve the above purpose, the present invention provides the following technical solutions: The intelligent regulation and control system for the 5G video stream edge of the self-adaptive code rate comprises a global snapshot modeling module, a hierarchical cache arrangement module, an experience constraint distribution module and a version closed loop correction module, wherein the modules are connected through signals; the global snapshot modeling module is used for collecting link observation, edge resource allowance, terminal session state and interaction event count, calculating link speed and propagation delay, constructing a speed matrix and a time matrix, decomposing the live delay into a plurality of sections of propagation items and transcoding items, establishing capacity constraint, availability constraint and interlayer dependency constraint of layered content, packaging into a state snapshot, and generating a candidate placement queue; The hierarchical cache arrangement module analyzes the hierarchical requests based on the state snapshot, counts concurrent request sets, divides a shared layer and a non-shared layer, calculates the scheduling relation between a cache placement scheme and the shared layer under the constraint of capacity, availability and dependence, performs feasibility verification on the bandwidth resources of the edge and the computing resources, performs hierarchical rollback and synchronously corrects the code rate version table when the bandwidth resources of the edge and the computing resources of the edge are not met, enables placement, scheduling and code rate decision to be consistent, and outputs a cache plan, a scheduling plan and a code rate version result; The version closed loop correction module forms an interactive experience item by using a live broadcast total time delay and a delay threshold value, forms a video quality item by using an expected code rate and an issued code rate, forms a code rate jitter loss item by using a historical average code rate, forms a switching loss item by using a connection state difference, selects unique allocation from a candidate edge server set for each terminal under the double constraint of bandwidth and computing power, carries out code rate self-adaptive correction and alternative server rollback, and outputs a terminal allocation table, a terminal code rate configuration table and a server resource surplus table; And the experience constraint distribution module is used for triggering the actual throughput, playing buffering, propagation delay, resource surplus and cache hit statistics of the complementary acquisition according to a scheduling period or an event, firstly executing hard verification of the reso