Deadline-constrained content upload from multihomed devices: Formulations and algorithms
This work originates from the practical requirements of video surveillance in public transport systems, where security cameras store video onboard, and a central operator occasionally needs to access portions of the recordings. When this happens, the selected video portions must be uploaded within a...
| Autores: | , , , |
|---|---|
| Tipo de recurso: | artículo |
| Fecha de publicación: | 2018 |
| País: | España |
| Institución: | IMDEA Networks Institute |
| Repositorio: | IMDEA Networks Institute Digital Repository |
| Idioma: | inglés |
| OAI Identifier: | oai:dspace.networks.imdea.org:20.500.12761/529 |
| Acceso en línea: | http://hdl.handle.net/20.500.12761/529 https://dx.doi.org/https://doi.org/10.1016/j.comnet.2018.06.008 |
| Access Level: | acceso abierto |
| Palabra clave: | Adaptive scheduling Wireless network Multihomed devices Content upload Smart city |
| Sumario: | This work originates from the practical requirements of video surveillance in public transport systems, where security cameras store video onboard, and a central operator occasionally needs to access portions of the recordings. When this happens, the selected video portions must be uploaded within a given deadline, using (multiple) wireless interfaces, with different costs (which correspond to, e.g., tariffs). We study this video upload problem as a scheduling problem with deadline, where our goal is to choose which interfaces to use and when, so as to minimize the cost of the upload while meeting the given deadline. Our study gives rise to adaptive schedulers that require only a very coarse knowledge of the wireless interfaces bandwidth. In this paper, we first assume an oracle has the perfect knowledge about the available bandwidth of wireless interfaces at each time, and we formulate an optimization problem to minimize the upload cost within the given deadline. Second, we propose greedy oracle-based heuristics that perform very close to optimal, and that can provide a simple baseline for performance. Third, we formulate a stochastic optimization problem, assuming only the knowledge of the distribution of available bandwidth, and, fourth, we propose adaptive schedulers, that we simulate and also implement and test in a real testbed. Simulation results demonstrate that the proposed adaptive solutions can effectively leverage the fundamental trade-off between upload cost and completion time, despite unpredictable variations in the available bandwidth of wireless interfaces. Experiments with real mobile nodes provided by the MONROE platform confirm the findings. |
|---|