Abstract:Sparse mobile crowdsensing (MCS) is a promising paradigm for the large-scale urban sensing, which allows us to collect data from only a few areas (cell selection) and infer the data of other areas (data inference). It can significantly reduce the sensing cost while ensuring high data quality. Recently, large urban sensing systems often require multiple types of sensing data (e.g., publish two tasks on temperature and humidity respectively) to form a multi-dimensional urban sensing map. These multiple types of sensing data hold some inherent correlations, which can be leveraged to further reduce the sensing cost and improve the accuracy of the inferred results. In this paper, we study the multi-dimensional urban sensing in sparse MCS to jointly address the data inference and cell selection for multi-task scenarios. We exploit the intra- and inter-task correlations in data inference to deduce the data of the unsensed cells through the multi-task compressive sensing and then learn and select the most effective $\langle $ cell, task $\rangle $ pairs by using reinforcement learning. To effectively capture the intra- and inter-task correlations in cell selection, we design a network structure with multiple branches, where branches extract the intra-task correlations for each task, respectively, and then catenates the results from all branches to capture the inter-task correlations among the multiple tasks. In addition, we present a two-stage online framework for reinforcement learning in practical use, including training and running phases. The extensive experiments have been conducted on two real-world urban sensing datasets, each with two types of sensing data, which verify the effectiveness of our proposed algorithms on multi-dimensional urban sensing and achieve better performances than the state-of-the-art mechanisms.

Multi-Task Multi-Attention Graph Neural Network for Mobile Crowd Sensing Data Reconstruction and Prediction

Spatial-Temporal Data Inference With Graph Attention Neural Networks in Sparse Mobile Crowdsensing

Spatiotemporal Urban Inference and Prediction in Sparse Mobile CrowdSensing: A Graph Neural Network Approach

Spatiotemporal Fracture Data Inference in Sparse Mobile Crowdsensing: A Graph-and Attention-Based Approach

Spatiotemporal Transformer for Data Inference and Long Prediction in Sparse Mobile CrowdSensing.

Vehicular Crowdsensing Inference and Prediction with Multi Training Graph Transformer Networks

Sparse Mobile Crowdsensing for Cost-Effective Traffic State Estimation with Spatio-Temporal Transformer Graph Neural Network

Multi-stage attention spatial-temporal graph networks for traffic prediction

Multi-Dimensional Urban Sensing in Sparse Mobile Crowdsensing

Multi-Task Synchronous Graph Neural Networks for Traffic Spatial-Temporal Prediction.

Learning Multiaspect Traffic Couplings by Multirelational Graph Attention Networks for Traffic Prediction

Fine-grained Urban Prediction Via Sparse Mobile CrowdSensing.

Multi-View Matrix Factorization for Sparse Mobile Crowdsensing

MSASGCN : Multi-Head Self-Attention Spatiotemporal Graph Convolutional Network for Traffic Flow Forecasting

Multi-Relational Graph Attention Network for Social Relationship Inference from Human Mobility Data

MRA-DGCN: Multi-Range Attention-Based Dynamic Graph Convolutional Network for Traffic Prediction

Multi-Scale Convolution Multi-Graph Attention Neural Networks for Traffic Flow Forecasting

Multicomponent Spatial-Temporal Graph Attention Convolution Networks for Traffic Prediction with Spatially Sparse Data

Multivariate and Propagation Graph Attention Network for Spatial-Temporal Prediction with Outdoor Cellular Traffic

Active Sparse Mobile Crowd Sensing Based on Matrix Completion

Toward Time-Continuous Data Inference in Sparse Urban CrowdSensing