Dynamic Population Estimation Using Voronoi Cell Partitioning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for estimating population density using cellular data are static and unreliable for real-time applications, especially in small or non-recurring situations, as they fail to account for temporal variability and spatial distribution of people.
Innovation Solution
A method that learns coefficients by processing mobile telephony data through Voronoi cell partitioning and hybrid mapping, using a model that links mobile telephony parameters to population density, incorporating both surface and residential population weights, and adjusts for activity levels to provide accurate real-time estimates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If static coefficients are used for population estimation, then the method is simple to implement, but the accuracy deteriorates in real-time applications due to temporal variability
Solution Approach 1:
The patent transforms the static coefficient model into a dynamic one by introducing time-varying coefficients that adapt to temporal changes in population behavior. The model now uses coefficients that evolve over time based on observed patterns in cellular data, allowing it to capture diurnal variations and seasonal trends while maintaining computational efficiency.
Solution Approach 2:
The patent changes the parameters from fixed constants to time-dependent variables. By modeling coefficients as functions of time and incorporating temporal features into the estimation model, the system adapts to changing population dynamics without requiring complete remodelling, thus improving accuracy while controlling complexity.
2Adaptability or versatility
If traditional cellular data methods are used, then the coverage is broad, but the reliability deteriorates in small localities and occasional situations
Solution Approach 1:
The patent applies local quality by adapting the estimation model specifically for small localities and occasional situations. It introduces location-aware coefficients and adjusts the model parameters based on the characteristics of the specific area being estimated, whether urban, rural, or temporary gathering places, thereby improving reliability without sacrificing broad applicability.
Solution Approach 2:
The patent segments the population estimation problem into different contexts (routine vs. occasional situations, different locality sizes) and applies context-specific modeling approaches. By dividing the estimation task into specialized sub-models or context-aware parameters, it achieves high reliability across diverse scenarios while maintaining overall system versatility.
3Speed
If real-time estimation is implemented, then the timeliness is improved, but the complexity of the system increases
Solution Approach 1:
The patent performs preliminary actions by pre-computing and storing baseline coefficients, spatial weights, and temporal patterns during off-peak periods. This preprocessing allows the real-time estimation system to operate with reduced computational burden, using stored reference data combined with current observations to rapidly generate estimates without complex real-time calculations.
Solution Approach 2:
The patent implements partial action by focusing computational resources on the most critical estimation tasks and using simplified models for routine updates. It applies full complexity only when necessary (e.g., during significant events or initial model training), while using lighter-weight computations for regular real-time updates, thus achieving timely responses with controlled complexity.
Data Source
Figure 1
AI summary
The present invention relates to a method for learning population estimation coefficients in a reference area, comprising implementing, by means of data processing equipment (11) (10), the following steps: (a) For each of a plurality of relay antennas (2a, 2b) defining a first partition into Voronoi cells of said reference area, obtaining the value over time for said Voronoi cell of a first mobile telephony parameter representative of the presence of subscribers of a telephone operator; (b) For each block of a set of blocks defining a second partition of said reference area, calculating the value over time for the block of said first parameter, as a function, for each Voronoi cell having a non-zero intersection with said block: ∘ Of the value of said first parameter obtained for said cell;(c) The ratio of the surface area of said intersection to the surface area of all said cell; and (c) The ratio of the population of said block to the populations of blocks having a non-zero intersection with said cell; (c) Learning the coefficients of a model linking at least said first parameter to the population, from a training database. The present invention also relates to a method for estimating the population in a given area.