A Fast and High Quality Multilevel Scheme for Partitioning Irregular Graphs
SIAM Journal on Scientific Computing
An Efficient Path Computation Model for Hierarchically Structured Topographical Road Maps
IEEE Transactions on Knowledge and Data Engineering
SEA'11 Proceedings of the 10th international conference on Experimental algorithms
Efficient data management in support of shortest-path computation
Proceedings of the 4th ACM SIGSPATIAL International Workshop on Computational Transportation Science
Towards a Flexible and Scalable Fleet Management Service
Proceedings of the Sixth ACM SIGSPATIAL International Workshop on Computational Transportation Science
Optimizing Landmark-Based Routing and Preprocessing
Proceedings of the Sixth ACM SIGSPATIAL International Workshop on Computational Transportation Science
Mobility and social networking: a data management perspective
Proceedings of the VLDB Endowment
Hi-index | 0.00 |
Crowdsourcing road network data, i.e., involving users to collect data including the detection and assessment of changes to the road network graph, poses a challenge to shortest-path algorithms that rely on preprocessing. Hence, current research challenges lie with improving performance by adequately balancing preprocessing with respect to fast-changing road networks. In this work, we take the crowdsourcing approach further in that we solicit the help of users not only for data collection, but also to provide us their computing resources. A promising approach is parallelization, which splits the graph into chunks of data that may be processed separately. This work extends this approach in that small-enough chunks allow us to use browser-based computing to solve the pre-computation problem. Essentially, we aim for a Web-based navigation service that whenever users request a route, the service uses their browsers for partially preprocessing a large, but changing road network. The paper gives performance studies that highlight the potential of the browser as a computing platform and showcases a scalable approach, which almost eliminates the computing load on the server.