CDN Server Selection Using Model Predictive Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current path control by ISPs and server selection by CDN operators are independently performed, leading to challenges in optimizing delivery paths and server selection, especially with dynamic changes in content demand and prediction errors, resulting in frequent user request rejections in CDNs.
Innovation Solution
A design apparatus using model predictive control to determine the probability of selecting a delivery server or route, considering network parameters and estimated demand, to reduce excessive delivery requests by optimizing cache server and delivery path selection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If path control is performed by ISP and server selection is performed by CDN operator independently, then each party can perform control according to their own policy, but the overall delivery optimization is compromised and request rejections increase
Solution Approach 1:
The patent combines path control and server selection into a unified joint optimization framework. The system simultaneously determines both the delivery path and selected server by considering network state information exchanged between ISP and CDN operator, rather than performing these controls independently as separate processes.
Solution Approach 2:
The system implements feedback mechanisms where the CDN operator provides network state information to the ISP, and the ISP returns path control information. This iterative information exchange enables both parties to coordinate their decisions, adjusting server selection and path routing based on real-time network conditions to minimize request rejections.
2Device complexity
If static server selection and path control are used, then the system is simple to implement, but it cannot adapt to dynamic changes in content demand and prediction errors
Solution Approach 1:
The patent implements dynamic server selection and path control that adapts to changing network conditions and content demand patterns. The system uses prediction algorithms to forecast future demand and adjusts server selection probabilities and path routing decisions in real-time based on actual network state information, rather than relying on static configurations.
Solution Approach 2:
The system performs preliminary actions by predicting future content demand patterns and pre-adjusting server selection probabilities and path configurations before actual requests occur. This predictive approach allows the system to proactively optimize for upcoming demand changes rather than reacting passively to actual requests.
Data Source
AI summary
A design apparatus is provided for designing a probability that a delivery server or a delivery route is selected for content delivery in response to a delivery request from a user. The design apparatus receives, as input parameter values, a parameter related to a network and an estimated amount of demand for delivery requests generated by users for content, and designs a selection probability that a cache server is selected as a delivery server for content delivery or a selection probability that a delivery route is selected, using model predictive control. The model predictive control uses an objective function for reducing the number of excessive delivery requests, each of which is a delivery request generated by a user for a content item and then rejected, and said delivery route is a combination of the delivery server and a delivery path on which the content is delivered by the delivery server.