Predicting Traffic Congestion Propagation Patterns

Abstract

A traffic congestion in a road network may propagate to upstream road segments. Such a congestion propagation may make a series of connected road segments congested in the near future. Given a spatial-temporal network and congested road segments in current time, the aim of predicting traffic congestion propagation pattern is to predict where those congestion will propagate to. This can provide users (e.g. city officials) with valuable information on how congestion will propagate in the near future to help mitigating emerging congestions. However, it is challenging to predict in realtime due to complex propagation process between roads and high computational intensity caused by large dataset. Recent studies have been focusing on finding frequent or most likely congestion propagation patterns in historical data. In contrast, this research will address the problem of predicting congestion propagation patterns in the near future. We predict the footprint of congestion propagation as Propagation Graphs (Pro-Graphs) where the root of each Pro-Graph is a set of congested roads propagating congestion to nearby roads. We propose an efficient algorithm called PPI_Fast to achieve this prediction. Our experiments on real-word dataset from Shenzhen, China shows that the PPI_Fast is able to predict near future propagations with AUC of 0.75 and improves the running time of the baseline algorithm. Two case studies have been done to show our work can find meaningful patterns.

Mining Topological Dependencies of Recurrent Congestion in Road Networks

“…Another line of research investigates the evolution of RC patterns. Current approaches typically analyse the propagation of RC within a spatial grid [3,4,22] or a road network graph [23][24][25].…”

“…Algorithm 1 presents an incremental greedy approach to merge spatially overlapping affected subgraphs. The algorithm consist of a main loop (line 6-24) where the individual steps include candidate generation (line 9-11), similarity computation (line 12-14) and merging (line [15][16][17][18][19][20][21][22][23][24]. For the candidate generation, we consider all subgraph pairs that share at least one unit as candidates (line 13).…”

Dynamic Spatiotemporal Causality Analysis for Network Traffic Flow Based on Transfer Entropy and Sliding Window Approach

“…Chawla et al [18] proposed an optimized mining algorithm framework for inferring the root cause of anomalies from large taxis GPS data. Xiong et al [19] developed a propagation graph approach to predict traffic congestion patterns in the near future based on the large real-world vehicle trajectory data.…”

“…Based on the dynamic spatiotemporal causality graphs, four indicators are proposed to evaluate the impacts of any road segment in the road network from the perspective of causal dependence. e input degree D t in (i) is defined as (19), denoting the impacts of the traffic states of the other n − 1 road segments on that of r i at time t. e output degree D t out (i) is defined as (20), denoting the influence of traffic state for road segment r i on the other n − 1 road segments. e sum of input degrees SumD in (i) and the sum of output degrees SumD out (i) are defined to quantify the cause-effect relationship between the road segment r i and the other road segments during the time period T, as shown in ( 21) and (22).…”

Traffic Congestion Forecasting in Shanghai Based on Multi-Period Hotspot Clustering

“…Do et al used an attention mechanism to mine the temporal and spatial characteristics of data with their proposed model, thereby improving the prediction effect of the model [28]. Xiong, Haoyi et al noted that traffic congestion could propagate across roads [29]. To solve the prediction problem, they predicted the footprint of congestion propagation as propagation graphs and proposed the PPI_Fast algorithm.…”