HLS live broadcast method based on multi-cloud storage

Abstract

The invention discloses an HLS live broadcast method based on multi-cloud storage, and the key point of the technical scheme is that the method comprises the following steps: S1, segmenting a video stream to be live broadcast into a plurality of video clips; s2, dynamically adjusting the transmission size and request frequency of each video clip according to parameters such as real-time network bandwidth, delay and geographical location of audience; s3, an HLS Low-Latency protocol is adopted to carry out transmission of the video clips; s4, storing the video clips on a plurality of cloud storage platforms, and performing cross-platform distribution of the video clips through a load balancing algorithm; s5, monitoring the load and performance of each cloud storage platform in real time, and dynamically selecting the optimal storage position of the video clip through an intelligent scheduling system; according to the invention, in a high-concurrency and complex network environment, the stability, the reliability and the user experience of the HLS live broadcast system can be remarkably improved through the technologies of cooperation, intelligent scheduling, redundancy storage, edge calculation, load balancing, fault switching and the like of the multi-cloud storage platform.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of video live streaming, in particular to an HLS live streaming method based on multi-cloud storage. BACKGROUND

[0002] With the rapid development of Internet technology, especially the widespread application of high-definition video and live streaming content, the live streaming industry has increasingly demanding requirements for low-latency and high-quality video transmission. In the traditional HTTP Live Streaming (HLS) protocol, due to its segmented video stream-based approach, it often faces high transmission latency and unstable playback quality, especially in environments with high concurrency, bandwidth fluctuations, and wide geographical distribution. These problems result in the viewing experience of live streaming users being affected, with phenomena such as large latency and frequent freezing.

[0003] Traditional HLS protocols usually rely on a single cloud storage platform for video content storage and distribution. This single-platform architecture is difficult to meet the needs of efficient distribution and disaster recovery when facing high concurrency and large-scale user access. The single platform has obvious limitations in terms of bandwidth bottlenecks, computing resource loads, and network latency, and cannot flexibly cope with cross-regional distribution and real-time load scheduling, thereby affecting the stability and availability of the overall service.

[0004] In order to optimize the live streaming experience, the combination of low-latency transmission and multi-cloud storage has become a new solution. Through the multi-cloud storage architecture, different cloud platforms can intelligently schedule according to real-time network status and user location, thereby better coping with bandwidth fluctuations and high concurrency requests. However, the existing multi-cloud storage systems still have certain deficiencies in dynamic load balancing, cache consistency, data redundancy, and automatic fault switching, making it difficult to achieve optimal video live streaming results in complex environments. To solve the above problems, we propose an HLS live streaming method based on multi-cloud storage. SUMMARY

[0005] In view of the deficiencies of the prior art, the present application provides an HLS live streaming method based on multi-cloud storage, which solves the problems raised in the background art.

[0006] The above technical purpose of the present application is achieved by the following technical scheme:

[0007] An HLS live streaming method based on multi-cloud storage, the method comprising the following steps:

[0008] S1, segmenting a video stream to be live streamed into multiple video segments, each video segment having a duration of between 2 seconds and 10 seconds;

[0009] S2, dynamically adjust the transmission size and request frequency of each video segment according to real-time network bandwidth, delay, and audience geographic location parameters;

[0010] S3, use HLS Low-Latency protocol for video segment transmission to ensure that the playback delay of each video segment is as low as possible;

[0011] S4, store the video segments on multiple cloud storage platforms and distribute the video segments across platforms through a load balancing algorithm;

[0012] S5, monitor the load and performance of each cloud storage platform in real time, and dynamically select the optimal storage location for the video segments through an intelligent scheduling system to ensure fast data access.

[0013] Preferably, in step S2, dynamically adjusting the size and transmission frequency of each video segment includes:

[0014] According to the monitoring results of real-time bandwidth B real―time and delay D real―time , calculate the request frequency R request of the video segment, the formula is:

[0015]

[0016] Where f(D real―time ) is a delay compensation function designed to adjust the request frequency to ensure low delay playback;

[0017] According to the real-time bandwidth and delay, select the duration T segment of each video segment, which is determined by the following formula:

[0018]

[0019] Where the video segment duration is limited to between 2 seconds and 10 seconds to ensure smooth video playback.

[0020] Preferably, in step S3, the low-latency transmission protocol is HLS Low-Latency protocol or other low-latency optimized HTTP protocol, where the transmission delay L latency of the protocol can be optimized by the following formula:

[0021]

[0022] Where R request is the request frequency, and f(B real―time , D real―time ) is a function optimized according to real-time bandwidth and delay, designed to reduce transmission time.

[0023] Preferably, in step S4, the load balancing algorithm selects the optimal storage node based on the following parameters:

[0024] Bandwidth usage B of the cloud storage platform platform , computing resource load C platform , storage capacity S platform , and network latency D with the audience platform and the following load balancing formula F load to select the optimal node:

[0025]

[0026] where α and β are adjustment coefficients reflecting the weights of storage capacity and latency, and the platform with the smallest F load value is selected as the preferred node.

[0027] According to the geographical distribution, the cloud platform with the shortest distance to the user is selected to store the video segment.

[0028] Preferably, in step S5, the intelligent scheduling system includes:

[0029] Real-time collection and analysis of performance data of the cloud platform, dynamic adjustment of storage location of the video segment based on the current load situation, and intelligent scheduling algorithm as follows:

[0030] A schedule (t) = argmin node (P cloud (t))

[0031] where P cloud (t) represents the comprehensive performance index of the cloud platform at time t, and the intelligent scheduling system selects the node with the best performance for video segment storage and distribution;

[0032] Automatic selection of the cloud platform with the best performance for video segment storage and distribution.

[0033] Preferably, the method further includes caching part of the video segment on the edge computing node to reduce transmission delay and improve viewing experience.

[0034] Preferably, the method further includes synchronizing the cached video segments among multiple cloud platforms to ensure consistency of video content among all cloud platforms, and synchronizing transmission when the cache is updated to avoid cache invalidation.

[0035] Preferably, the method further includes using real-time fault detection and automatic fault switching mechanism to automatically switch the video stream to a cloud platform with lower load or normal performance when a cloud platform fails or performance decreases.

[0036] Preferably, the method further comprises: during the transmission of the video segments, using a redundant data technology to store each video segment in a redundant piece, ensuring that when a certain cloud platform fails, the video stream can be recovered from the redundant copy of other cloud platforms, reducing the risk of live interruption or data loss.

[0037] In summary, the present application mainly has the following beneficial effects:

[0038] 1、The present application adopts a HLS live streaming method based on multi-cloud storage, especially by dynamically adjusting the transmission size, request frequency and video segment length, which can optimize the transmission process according to the real-time network bandwidth and delay, and this adaptive adjustment method ensures low latency in live streaming, avoids playback stuttering caused by bandwidth fluctuations or network congestion, and significantly improves the viewing experience.

[0039] 2、The present application realizes load balancing and intelligent scheduling among multiple cloud storage platforms, ensuring that video segment distribution and storage are more efficient and reliable, and when a cloud platform fails, the system can automatically detect the fault and switch to other normal cloud platforms, avoiding the risk of live interruption or data loss, in addition, the application of redundant data technology further enhances the disaster recovery capability of live streaming, ensuring that the video content remains consistent among multiple cloud platforms and can be quickly recovered.

[0040] 3、The present application uses multi-cloud storage platforms and edge computing node caching technology, which can dynamically select the optimal storage and distribution node according to the user's geographic location and demand, thus realizing low-latency cross-platform distribution, and the real-time synchronization and cache update mechanism among multiple cloud platforms effectively avoids the problem of data inconsistency or cache invalidation, improving the reliability and performance of the entire live streaming system. BRIEF DESCRIPTION OF DRAWINGS

[0041] Figure 1 is a flowchart of the HLS live streaming method based on multi-cloud storage. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme of the embodiments of the present application will be described clearly and completely below in conjunction with the drawings of the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the described embodiments of the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of protection of the present application.

[0043] The following examples are used to illustrate the present application, but cannot be used to limit the scope of protection of the present application. The conditions in the examples can be further adjusted according to specific conditions, and simple improvements to the method of the present application under the concept of the present application also belong to the scope of protection of the present application.

[0044] Example 1: HLS live streaming process based on multi-cloud storage, reference Figure 1

[0045] 1.1 Video stream segmentation: The video stream to be live broadcast is segmented into multiple video segments, and the duration of each video segment is set to 2-10 seconds. This duration range can reduce the delay in video playback while maintaining video smoothness.

[0046] 1.2 Real-time monitoring of network bandwidth and delay, according to the monitoring results of real-time bandwidth B real―time and delay D real―time , dynamically adjust the request frequency R requeest and segment size of the video segment, the specific calculation formula is:

[0047] Request frequency:

[0048]

[0049] Where f(D real―time ) is a delay compensation function designed to adjust the request frequency to ensure low-delay playback.

[0050] Video segment duration selection:

[0051]

[0052] This formula ensures that the duration of the video segment is between 2-10 seconds, suitable for different bandwidth and delay conditions, and ensures the smoothness of video playback.

[0053] 1.3 Low-delay transmission:

[0054] Use HLS Low-Latency protocol for video segment transmission, optimize transmission delay L latency :

[0055]

[0056] This optimization function reduces transmission time and improves real-time video playback effect.

[0057] 1.4 Load balancing and multi-cloud storage distribution:

[0058] Video segments are stored on multiple cloud storage platforms, and the system selects the optimal storage node through a load balancing algorithm. The load balancing formula is: ​

[0059]

[0060] where B platform is the bandwidth usage, C platform is the computational load, S platform is the storage capacity, D platform is the network latency, and the adjustment coefficients α and β are optimized according to the weights of storage capacity and latency for storage node selection.

[0061] 1.5, Intelligent Scheduling System:

[0062] The intelligent scheduling system collects and analyzes the performance data of the cloud platform in real time, dynamically adjusts the storage location of video segments based on the current load situation, and the intelligent scheduling algorithm is as follows:

[0063] A schedule (t) = argmin node (P cloud (t))

[0064] where P cloud (t) represents the comprehensive performance index of the cloud platform at time t, and the system selects the node with the best performance for video segment storage and distribution.

[0065] 1.6, Redundant Data and Fault Switching:

[0066] Redundant data technology is used to store video segments redundantly among multiple cloud platforms, ensuring that video streams can be recovered from copies on other platforms in the event of a failure of a certain cloud platform. Real-time fault detection and automatic fault switching mechanism will automatically switch the video stream to a cloud platform with normal performance, reducing the risk of live interruption.

[0067] Example 2: Edge Computing Node Cache and Synchronization Mechanism

[0068] In the multi-cloud storage HLS live streaming system, edge computing nodes serve as an important optimization method by caching video content to network nodes close to users, significantly reducing transmission latency and improving viewing experience. The following details the edge computing node caching mechanism and synchronization scheme.

[0069] 2.1, Working Principle of Edge Computing Node

[0070] The edge computing node is responsible for caching video segments from multiple cloud platforms. To ensure that these video segments can be efficiently transmitted to end users, the edge node will be dynamically adjusted based on the following factors:

[0071] 2.11, User Request Analysis:

[0072] The system predicts the video segments that the user is most likely to watch based on their viewing behavior, request frequency, and historical viewing records, and pre-caches these segments at the edge nodes. The selection of cached segments is based on an intelligent prediction model that can dynamically adjust to audience demand.

[0073] 2.12 Real-time bandwidth and latency monitoring:

[0074] The edge nodes monitor the bandwidth and latency between themselves and the cloud platform in real-time. A bandwidth and latency estimation model is used to select the most suitable video segments for caching. Through real-time monitoring of bandwidth B_(edge) and latency D_(edge), the caching strategy is dynamically adjusted to maximize the use of network resources.

[0075] 2.13 Dynamic cache management:

[0076] The cache management strategy of the edge computing node is optimized based on the following formula:

[0077]

[0078] where f(B edge ,D edge ,T user ) is a weighted function that considers bandwidth, latency, and user viewing time to dynamically select the most needed segments for caching.

[0079] 2.2 Synchronization mechanism

[0080] To ensure that the video content remains consistent across multiple cloud platforms and edge computing nodes, the following synchronization strategies are implemented:

[0081] 2.21 Real-time synchronization:

[0082] Whenever a video segment is updated or a new video segment is generated, the cloud platform will push the content to each edge node through an efficient synchronization protocol. The synchronization frequency is dynamically adjusted based on network bandwidth and storage capacity to ensure that the latest data is synchronized in a timely manner.

[0083] 2.22 Conflict resolution in synchronization:

[0084] In the synchronization of cross-platform caching, if a cache conflict occurs (e.g., different versions of the same segment), the system will use the following strategies to resolve the conflict:

[0085] 2.221 According to the update timestamp of the cache, the latest segment is selected for replacement;

[0086] 2.222 Based on user viewing history, the segments with higher user viewing frequency are preferentially retained, and other low-frequency segments are cleaned up or updated;

[0087] 2.23, Synchronization Algorithm:

[0088] To ensure the efficiency and consistency of the synchronization process, a distributed synchronization algorithm such as Paxos or Raft is used to ensure the consistency of data between multiple platforms. In practical applications, the appropriate synchronization frequency is selected based on network conditions, and incremental synchronization is used to reduce bandwidth consumption.

[0089] Embodiment 3: Multi-cloud storage optimization in high concurrency scenarios

[0090] When facing large-scale user concurrency access, the live streaming system must be able to intelligently schedule cloud platform resources and dynamically adjust the load to ensure stability and smoothness in high concurrency scenarios. The following is a detailed description of this embodiment:

[0091] 3.1, Dynamic load balancing under high concurrency access

[0092] In high concurrency scenarios, traditional load balancing strategies often face problems such as bandwidth bottlenecks, uneven distribution of computing resources, and excessive delays. Therefore, the system needs to use real-time load balancing algorithms to optimize platform selection and ensure that video clips can be quickly and accurately distributed to viewers.

[0093] 3.11, Load balancing formula:

[0094] In order to select the optimal storage platform, the system uses the following load balancing function F load Dynamically select the most suitable cloud platform for storage and distribution:

[0095]

[0096] Where B platform is the bandwidth utilization of the platform, C platform is the computing resource load of the platform, S platform is the storage capacity, D platform is the network delay to the audience, U platform is the current user load of the platform, and α and β and γ are adjustment coefficients reflecting the importance of bandwidth, storage, and user load.

[0097] By considering these factors, the system selects the platform with the smallest F load to store and distribute video clips. Whenever the platform load increases, the system will dynamically adjust resource allocation to avoid overloading a single platform.

[0098] 3.2, Intelligent scheduling and priority adjustment

[0099] In high concurrency scenarios, the system dynamically adjusts the storage and distribution strategy of video clips through intelligent scheduling algorithms. The scheduling algorithm is based on the following comprehensive performance index P cloudTo select the optimal cloud platform:

[0100] P cloud (t) = α1·L latency (t) + α2·B load (t) + α3·F fail (t)

[0101] Where L latency is the transmission delay of the platform, B load is the bandwidth usage rate of the platform, F fail is the failure rate of the platform, and α1, α2, α3 are adjustment coefficients.

[0102] Based on these real-time monitoring data, the intelligent scheduling system automatically selects the platform with the best performance, ensuring that video clips can be transmitted to users under the conditions of the lowest delay and the highest bandwidth.

[0103] 3.3, Redundant storage and failure switching

[0104] In order to deal with system failure or network failure, the system adopts redundant data storage and automatic failure switching mechanism, each video clip is redundantly stored between multiple cloud platforms, and the following redundant shard storage strategy is used for disaster recovery processing:

[0105] 3.31, Redundant storage strategy:

[0106] Each video clip is divided into multiple small blocks (redundant shards), which are stored on different cloud platforms. Whenever a platform fails, the system will automatically recover the video stream from the redundant copy on other platforms, ensuring seamless switching;

[0107] 3.32, Automatic failure detection and switching:

[0108] The system automatically identifies faulty nodes by monitoring the health status of the platform in real time, and determines whether to switch the video stream to other platforms through the following failure switching formula:

[0109]

[0110] If the performance of the current platform decreases or fails, the system will automatically switch the video stream to a cloud platform with normal performance, thereby reducing the risk of live interruption;

[0111] 3.4, Cross-platform data synchronization in high concurrency optimization

[0112] In high concurrency scenarios, it is crucial to ensure data synchronization between all cloud platforms. The system uses an efficient data synchronization protocol and a distributed consistency algorithm (such as Raft algorithm) to ensure consistency of video content on all platforms, avoiding data conflicts and cache invalidation;

[0113] 3.41, Cross-platform synchronization algorithm:

[0114] Whenever a new video segment is generated, the system will automatically synchronize it to all redundant storage cloud platforms. During the synchronization process, the system will prefer the network path with the lowest latency and the widest bandwidth for data transmission to ensure fast and stable synchronization.

[0115] 3.42, Incremental synchronization and cache invalidation processing:

[0116] The system synchronizes only the recently updated segments through incremental synchronization, avoiding the full transmission of the entire video file, thereby reducing the bandwidth burden. At the same time, the system uses a cache invalidation detection mechanism to promptly clear invalid caches, ensuring that the caches on all platforms are always up-to-date.

[0117] Through the above embodiments, the present application can improve the stability, reliability and user experience of the HLS live streaming system significantly in a high-concurrency and complex network environment through the cooperation of multi-cloud storage platforms, intelligent scheduling, redundant storage, edge computing, load balancing and fault switching, etc.

[0118] Although embodiments of the present application have been shown and described, it will be understood by those skilled in the art that, unless otherwise defined, technical terms or scientific terms used in the present application should be understood as their usual meanings understood by those skilled in the art in the field to which the present application belongs, and the similar words such as "include" or "contain" in the present application mean that the elements or objects before the words cover the elements or objects listed after the words and their equivalents.

[0119] Although embodiments of the present application have been shown and described, it will be understood by those skilled in the art that, unless otherwise defined, technical terms or scientific terms used in the present application should be understood as their usual meanings understood by those skilled in the art in the field to which the present application belongs, and the similar words such as "include" or "contain" in the present application mean that the elements or objects before the words cover the elements or objects listed after the words and their equivalents.

Claims

1. A multi-cloud storage based HLS live streaming method, characterized in that, The method comprises the following steps: S1, segmenting the live video stream into multiple video segments, each video segment having a duration of between 2 seconds and 10 seconds; S2, dynamically adjusting the transmission size and request frequency of each video segment according to real-time network bandwidth, delay, and audience geographic location parameters; S3, using the HLS Low-Latency protocol for video segment transmission to ensure that the playback delay of each video segment is as low as possible; S4, storing the video segments on multiple cloud storage platforms and distributing the video segments across platforms using a load balancing algorithm; S5, monitoring the load and performance of each cloud storage platform in real time and dynamically selecting the optimal storage location for the video segments through an intelligent scheduling system to ensure fast data access.

2. The HLS live streaming method based on multi-cloud storage according to claim 1, characterized in that, In step S2, dynamically adjusting the size and transmission frequency of each video segment includes: According to the monitoring results of real-time bandwidth B real―time and delay D real―time , the request frequency R request of the video clip is calculated, and the formula is: where f(D real―time ) is a delay compensation function, which aims to adjust the request frequency to ensure low delay playback; According to the real-time bandwidth and delay, the length T of each video segment is selected segment is determined by the following formula: The duration of the video segment is limited to between 2 seconds and 10 seconds to ensure smooth video playback. 3.The HLS live streaming method based on multi-cloud storage of claim 1, wherein, In step S3, the low-latency transmission protocol is the HLS Low-Latency protocol or other low-latency optimized HTTP protocols.

4. The HLS live streaming method based on multi-cloud storage according to claim 1, characterized in that, In step S4, the load balancing algorithm selects the optimal storage node based on the following parameters: Bandwidth usage B of the cloud storage platform platform Compute resource load C platform Storage capacity S platform and network latency D to the audience platform etc. information, combined with the following load balancing formula F load Select the optimal node: where a and β are adjustment factors, reflecting the weight of storage capacity and delay, and F is finally selected as the platform with the minimum value of F load the platform with the minimum value of F as the preferred node; Select the cloud platform that is closest to the user in terms of geographic distribution to store the video segments.

5. The HLS live streaming method based on multi-cloud storage according to claim 1, characterized in that, In step S5, the intelligent scheduling system includes: Collect and analyze cloud platform performance data in real time, and dynamically adjust the storage location of the video segments based on the current load situation; Automatically select the cloud platform with the best performance for video segment storage and distribution.

6. The HLS live streaming method based on multi-cloud storage according to claim 1, characterized in that, The method further comprises: Caching some video segments on edge computing nodes to reduce transmission delay and improve viewing experience.

7. The multi-cloud storage based HLS live streaming method of claim 6, wherein, The edge computing node cache content is dynamically adjusted based on user requests and viewing history. 8.The multi-cloud storage based HLS live streaming method of claim 1, wherein, The method further comprises: Synchronizing cached video segments across multiple cloud platforms to ensure that video content remains consistent across all cloud platforms and is transmitted synchronously when the cache is updated to avoid cache invalidation. 9.The HLS live streaming method based on multi-cloud storage of claim 1, wherein, The method further comprises: Using real-time fault detection and automatic fault switching mechanisms to automatically switch video streams to a cloud platform with lower load or normal performance when a cloud platform fails or performance decreases.

10. The multi-cloud storage based HLS live streaming method of claim 1, wherein, The method further comprises: During the transmission of video segments, use redundancy data technology to store each video segment in redundant slices to ensure that when a cloud platform fails, the video stream can be recovered from the redundant copy on other cloud platforms, reducing the risk of live streaming interruption or data loss.

Images