Traffic Prediction Using Cluster Dissimilarity and Tag Elimination
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current traffic prediction methods are inadequate for long-term predictions and require significant computational resources, making them unsuitable for use on low-resource devices like mobile phones.
Innovation Solution
A method that clusters traffic profiles with associated tags, determines a reference cluster, calculates dissimilarity measures, and iteratively eliminates request tags to find the closest matching cluster for predicting traffic behavior, optimizing computational effort while enhancing accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If standard clustering algorithms are used for traffic profile analysis, then traffic patterns can be identified, but the computational resources required are too high for mobile devices
Solution Approach 1:
The patent segments the traffic prediction problem into two distinct phases: an offline training phase that runs on powerful servers to create pre-computed lookup tables, and an online prediction phase that runs on mobile devices using simple table lookups. This segmentation allows complex computations to be performed only once during training, while the mobile device performs only lightweight operations during actual use, resolving the contradiction between prediction accuracy and computational resource consumption.
Solution Approach 2:
The patent performs preliminary action by pre-computing traffic profile clusters and their characteristic vectors during an offline training phase before the mobile device needs to make predictions. The lookup tables containing cluster assignments and characteristic vectors are created in advance and stored on the mobile device, eliminating the need for complex real-time computations during prediction, thus enabling accurate predictions on resource-constrained devices.
2Duration of action of moving object
If existing traffic prediction methods are implemented, then short-term forecasts can be generated, but they cannot provide reliable long-term predictions
Solution Approach 1:
The patent uses copying by creating characteristic vectors that represent typical traffic patterns for different days of the week and times of day. During prediction, the system copies the appropriate characteristic vector from the lookup table based on the requested time and day, providing consistent and reliable predictions for long-term forecasts. This approach allows the system to maintain prediction reliability across different time horizons by referencing historical pattern copies rather than performing complex real-time simulations.
3Measurement precision
If complex clustering algorithms are used to improve prediction accuracy, then better traffic behavior prediction is achieved, but the device complexity increases
Solution Approach 1:
The patent segments the computational complexity into the offline training phase where complex clustering algorithms are executed on powerful servers, and the online prediction phase where only simple lookup operations are performed on mobile devices. This segmentation allows the system to achieve high prediction accuracy through sophisticated algorithms while keeping the mobile device complexity low, as the complex processing is performed only once during training and the results are stored in compact lookup tables.
Data Source
AI summary
The present subject matter relates to a method of predicting a traffic behaviour in a road system, comprising the following steps carried out by at least one processor connected to a database, which contains a set of traffic profiles: clustering the traffic profiles; for each cluster, determining a dissimilarity measure; receiving a request having one or more request tags; checking whether all request tags match one of the characteristic vectors of the clusters, and if so, outputting a predicted traffic behaviour, else, eliminating one of the request tags from the request and returning to the beginning of said step of checking.