Deadline-Driven Packet Dispatch Rate Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current content delivery over shared networks faces challenges in efficiently managing bandwidth distribution among different services, leading to suboptimal customer experience due to the generic nature of HTTP and TCP protocols, which do not consider varying bandwidth requirements of applications.
Innovation Solution
A method and apparatus that determine a minimum packet dispatch rate based on delivery deadlines, adjust packet transmission rates based on predicted quality of experience compared to network quality of experience, and use a global quality of experience parameter (QoEmax) to optimize bandwidth sharing among multiple services.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If generic HTTP and TCP protocols are used for content delivery, then protocol simplicity and ease of operation are improved, but bandwidth sharing sensitivity to application requirements deteriorates
Solution Approach 1:
The patent implements a feedback mechanism where the server monitors network conditions and adjusts packet transmission rates dynamically. The server determines a minimum packet dispatch rate based on delivery deadlines, transmits packets at this rate, measures the resulting packet loss rate, and uses this feedback to determine a predicted quality of experience for the next segment. This closed-loop feedback system allows the protocol to adapt to network conditions while maintaining simplicity.
Solution Approach 2:
The patent introduces dynamic adjustment of packet transmission rates based on delivery deadlines and network conditions. Instead of using a static transmission rate, the server dynamically determines the minimum packet dispatch rate for each segment based on its delivery deadline, and adjusts this rate dynamically based on measured packet loss rates and predicted quality of experience. This dynamic approach enables the generic HTTP protocol to become sensitive to application requirements.
2Adaptability or versatility
If network QoS is implemented to prioritize packets, then bandwidth sharing sensitivity to service requirements is improved, but system complexity and ease of manufacture deteriorates
Solution Approach 1:
The patent enables the content delivery system to self-manage bandwidth allocation without requiring external QoS infrastructure. The server autonomously determines minimum packet dispatch rates based on delivery deadlines, transmits packets, measures packet loss rates, and adjusts transmission rates dynamically. This self-service approach eliminates the need for complex network QoS systems while achieving service-sensitive bandwidth sharing.
Solution Approach 2:
The patent uses feedback from packet loss rate measurements to dynamically adjust transmission rates. The server measures the packet loss rate resulting from transmissions at the minimum packet dispatch rate, determines a predicted quality of experience, and compares this with the network quality of experience parameter. This feedback loop enables the system to automatically prioritize traffic based on delivery deadlines without requiring external QoS marking or prioritization infrastructure.
3Ease of operation
If TCP friendliness is applied uniformly to all sessions, then ease of operation is improved, but customer experience and adaptability to different service requirements deteriorates
Solution Approach 1:
The patent applies local quality by treating each content segment with unique characteristics based on its delivery deadline. Instead of applying uniform TCP friendliness to all sessions, the server determines a specific minimum packet dispatch rate for each segment based on its individual delivery deadline. This allows each segment to receive customized transmission treatment tailored to its specific requirements, enabling service-specific optimization while maintaining ease of operation through automated deadline-based rate determination.
Data Source
AI summary
Examples of the present invention present a method of controlling content delivery in a network. A global quality of experience measure, QoEmax, is calculated based on the packet loss rate in the network. As packet loss rate varies as a result of congestion conditions in the network, so will QoEmax. A server delivering content over the network will attempt to reach QoEmax for the content in its respective session. Those sessions with a quality of experience, QoE, less than QoEmax will have its packet dispatch rate increased, and those with a QoE higher than QoEmax will reduce its packet dispatch rate, subject to any delivery deadlines associated with the session. If the delivery deadlines of the sessions can be met without exceeding QoEmax, then all sessions will end up achieving QoEmax. Since QoEmax is only a function of packet loss rate that all sessions miming over the same shared network agree upon, all sessions should converge on the same QoE.