Multi-Loop Control for Real-Time Bidding Pace
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In the real-time bidding environment for online advertisements, existing systems face challenges in efficiently managing the pace of purchasing advertisements due to the need for rapid decision-making within a short time frame, often resulting in difficulties in evaluating and responding to numerous bid requests within the limited time allocated.
Innovation Solution
A multi-loop control algorithm is employed by a control system to dynamically adjust a pacing parameter, allowing for real-time evaluation and decision-making on bid requests, incorporating an outer control loop and an inner control loop to optimize the pace of bidding through a PID controller and Kalman filtering.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If a bidding server evaluates numerous bid requests in real-time with very short response time (under 50 ms), then the speed of bidding response is improved, but the ability to run large scale models to predict advertisement purchases deteriorates
Solution Approach 1:
The patent segments the bidding control process into multiple independent loops: an outer control loop that manages overall pacing parameters and an inner control loop that handles real-time bid decisions. This segmentation allows complex pacing logic to operate at higher levels while simpler rules execute at the millisecond-level bid response stage, resolving the contradiction between response speed and model complexity.
Solution Approach 2:
The system performs preliminary actions by pre-calculating pacing parameters and bid strategies in advance of actual bid requests. The outer control loop establishes pacing targets and constraints before the bidding campaign begins, allowing the inner loop to make rapid decisions based on pre-established criteria rather than computing complex models in real-time.
2Measurement precision
If the bidding server runs complex models to predict advertisement purchases, then the precision of bid decisions is improved, but the response time required increases beyond the 50 ms limit
Solution Approach 1:
The patent implements dynamic pacing parameters that adjust in real-time based on campaign performance and market conditions. The outer control loop continuously monitors bid outcomes and dynamically modifies pacing targets, while the inner loop adapts to these changes instantaneously. This dynamic approach maintains high decision accuracy without requiring static complex models to be re-evaluated for each bid.
Solution Approach 2:
The outer control loop acts as an intermediary between complex pacing strategies and simple real-time bid execution. It translates high-level campaign objectives into actionable pacing parameters that the inner loop can apply immediately, serving as a mediator that bridges the gap between strategic complexity and operational simplicity.
3Quantity of substance
If the bidding system processes billions of random bid requests throughout the day, then the quantity of ad placements available is improved, but the difficulty of evaluating each request within milliseconds increases
Solution Approach 1:
The control system serves itself by automatically adjusting pacing parameters based on performance feedback without requiring manual intervention or complex external optimization. The outer loop monitors campaign progress and self-adjusts pacing targets, while the inner loop autonomously makes bid decisions based on these parameters, enabling the system to handle billions of requests with simple, automated rules rather than complex evaluation logic.
Data Source
AI summary
A method for controlling a purchasing pace in a real-time bidding environment is provided. A pacing parameter is dynamically adjusted using a multi-loop control algorithm including an outer control loop and an inner control loop, wherein the outer control loop outputs a control parameter, and the inner control loop uses this control parameter as an input and outputs a current value of the pacing parameter, which is adjusted over time by the multi-loop control algorithm. A series of bid requests generated by a real-time bidding exchange are received, each identifying a respective available ad placement in a respective webpage being loaded. For each received bid request, a decision is made in real-time, during the loading of the respective webpage, whether to submit a bid to purchase the respective ad placement based at least on the current value of the dynamically adjusted pacing parameter.