Abstract
A novel scheme for sampling graph signals is proposed. Space-shift sampling can be understood as a hybrid scheme that combines selection sampling -- observing the signal values on a subset of nodes - and aggregation sampling - observing the signal values at a single node after successive aggregation of local data. Under the assumption of bandlimitedness, we state conditions and propose strategies for signal recovery in different settings. Being a more general procedure, space-shift sampling achieves smaller reconstruction errors than current schemes, as we illustrate through the reconstruction of the industrial activity in a graph of the U.S. economy.
Original language | English |
---|---|
Title of host publication | 2016 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) |
Subtitle of host publication | Proceedings |
Editors | Min Dong, Thomas Fang Zheng |
Place of Publication | Danvers, MA |
Publisher | IEEE |
Pages | 6355-6359 |
Number of pages | 5 |
ISBN (Electronic) | 978-1-4799-9988-0 |
DOIs | |
Publication status | Published - 19 May 2016 |
Event | 2016 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2016 - Shanghai International Convention Center, Shanghai, China Duration: 20 Mar 2016 → 25 Mar 2016 |
Conference
Conference | 2016 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2016 |
---|---|
Abbreviated title | ICASSP |
Country/Territory | China |
City | Shanghai |
Period | 20/03/16 → 25/03/16 |
Bibliographical note
Accepted Author ManuscriptKeywords
- Reconstruction
- Graph signal processing
- Space-shift sampling
- Bandlimited signal