Predictive Content Delivery with Randomized Ad Request Scheduling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Content providers face significant processing delays and network congestion due to simultaneous requests for advertising content from numerous playback devices during high-viewership events, leading to a poor viewing experience.
Innovation Solution
Implementing a system that predicts events within a content stream and generates predicted markers, allowing playback devices to wait a randomly generated time before requesting advertising content, thereby reducing the instantaneous load on the content delivery system.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If simultaneous requests for advertising content are processed during high-viewership events, then targeted advertisements can be delivered to viewers, but processing delays and network congestion occur
Solution Approach 1:
The system performs preliminary actions by predicting future ad opportunities and pre-fetching advertisement content before the actual ad break occurs. The content delivery network proactively requests and caches advertisement content in advance based on predicted events, so that when the actual ad break arrives, the content is already available for immediate playback without processing delays.
2Quantity of substance
If millions of simultaneous viewers request advertising content, then comprehensive ad delivery can be achieved, but the content provider's decision system becomes burdened with extreme instantaneous load
Solution Approach 1:
The system distributes prediction logic to individual playback devices through manifest files containing predicted event markers. Each device independently predicts ad opportunities locally without burdening the central decision system. This shifts the computational load from the server to the client devices, allowing the decision system to handle only the actual ad delivery when content is already cached.
Solution Approach 2:
Playback devices perform self-service by autonomously detecting predicted event markers in the content stream and automatically requesting and caching associated advertisement content without requiring real-time decisions from the content provider's system. This self-service mechanism reduces the instantaneous load on the decision system during high-viewership events.
3Loss of time
If advertisement content is requested in advance, then processing delays are reduced, but network traffic increases
Solution Approach 1:
The system implements local quality by allowing each playback device to independently determine its own prediction and caching strategy based on local detection of predicted event markers. Different devices may request different advertisement content at different times based on their individual playback schedules and local predictions, distributing network traffic more evenly rather than all devices requesting the same content simultaneously.
Data Source
AI summary
Methods and systems are described herein to predict events within a content stream. Based on the predicted events, information may be embedded within the stream. A computing device that receives the content stream may detect the predicted event and may wait a randomly generated time before requesting information associated with the predicted event, such as requesting an advertisement to output. After detecting the predicted event, the computing device may wait a randomly generated time before requesting an advertisement. The system may respond to the computing device with an instruction or with supplemental content, such as with an advertisement for later output. The computing device may cache the advertisement. The computing device may detect the actual marker indicating a timestamp for when the advertisement should be output. The computing device may output the advertisement at the time indicated by the timestamp.