Granular Media Stream Distribution Across Hosts With Asynchronous Memory Pools
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing protocols like SMPTE 2022-6 and SMPTE 2110 are not suitable for asynchronous media processing in cloud computing environments, imposing tight time constraints and limiting the efficient use of computing resources when multiple hosts are involved.
Innovation Solution
The Asynchronous Media Fabric (AMF) framework allows media processing functions to run asynchronously, using shared memory pools and a supervisory software system to manage real-time constraints, enabling efficient resource allocation and synchronization across multiple hosts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If SMPTE 2022-6 and SMPTE 2110 protocols are used to transport uncompressed media over ethernet, then real-time transmission capability is improved, but device complexity and difficulty of operation increase due to tight time constraints and specialized requirements
Solution Approach 1:
The patent introduces an intermediary layer (the media processing system with synchronized clocks and buffer management) that translates between the synchronous requirements of SMPTE protocols and the asynchronous nature of standard ethernet, thereby maintaining real-time capability without requiring all devices to directly implement complex SMPTE protocol stacks
Solution Approach 2:
The system enables standard ethernet infrastructure to serve multiple functions - both traditional data transmission and real-time media transport - by implementing synchronization and buffer management software that allows generic ethernet hardware to handle synchronous media streams without requiring specialized hardware
2Speed
If SMPTE 2022-6 and SMPTE 2110 protocols are used for media transport, then real-time transmission is enabled, but adaptability to cloud computing environments and asynchronous processing deteriorates
Solution Approach 1:
The patent implements dynamic buffer management and flexible synchronization mechanisms that allow the system to adapt between synchronous and asynchronous operating modes, enabling cloud-based hosts to process media streams at different rates while maintaining overall real-time performance through adjustable buffer sizes and clock synchronization tolerances
Solution Approach 2:
The system segments the media stream into discrete packets with timestamp information, allowing individual packets to be processed asynchronously across multiple cloud hosts while the overall stream maintains synchronization through buffer management, enabling both parallel processing and real-time delivery
3Manufacturing precision
If tight time constraints are imposed on ethernet packet transmission to simulate SDI timing characteristics, then synchronous stream accuracy is improved, but productivity and resource utilization worsen due to limited flexibility in processing
Solution Approach 1:
The patent implements preliminary buffer allocation and pre-synchronization of clocks across hosts before media transmission begins, allowing the system to establish timing references and allocate resources in advance, thereby reducing the need for tight real-time constraints during actual transmission and improving overall resource utilization
Solution Approach 2:
The system creates copies of media data in buffer memory with associated timestamp information, allowing multiple consumers to access identical data at different rates without affecting the original stream timing, thereby enabling asynchronous processing while maintaining synchronous accuracy for the primary stream
Data Source
AI summary
Including an asynchronous media fabric service in each of a plurality of hosts. The hosts operating together to asynchronously process real-time media streams under control of a media-processing application. A media-processing function process operates to capture from a real-time synchronous input stream and/or output to a real-time synchronous output stream. A grain transfer link is established for a transfer of grains between a first instance of a first shared memory pool located on a first host and a second instance of the first shared memory pool located on a second host. Following a first local write operation of a first grain to the first instance of the first shared memory pool, the asynchronous media fabric service is notified by a producer function process that the first grain is available for transfer to the second instance of the first shared memory pool on the second host via the grain transfer link.