### Abstrakti

The increasing number and variety of IoT (Internet of Things) devices produce a huge amount of diverse data upon which applications are built. Depending on the specific use case, the sampling rate of IoT sensors may be high, thus leading

the devices to fast energy and storage depletion. One option to address these issues is to perform data reduction at the source nodes so as to decrease both energy consumption and used storage. Most of current available solutions perform data reduction only at a single tier of the IoT architecture (e.g., at gateways), or simply operate a-posteriori once the data transmission has already taken place (i.e., at the cloud data center). This paper proposes a multi-tier data reduction mechanism deployed at both gateways and the edge tier. At the gateways, we apply the PIP (Perceptually Important Point) method to represent the features of a time series by using a finite amount of data. We extend such an algorithm by introducing several techniques, namely interval restriction, dynamic caching and weighted sequence selection. At the edge tier, we propose a data fusion method based on an optimal set selection. Such a method employs a simple strategy to fuse the data in the same time domain for a specific location.

Finally, we evaluate the performance of the proposed filtering and the fusion technique. The obtained results demonstrate the efficiency of the proposed mechanism in terms of time and accuracy.

the devices to fast energy and storage depletion. One option to address these issues is to perform data reduction at the source nodes so as to decrease both energy consumption and used storage. Most of current available solutions perform data reduction only at a single tier of the IoT architecture (e.g., at gateways), or simply operate a-posteriori once the data transmission has already taken place (i.e., at the cloud data center). This paper proposes a multi-tier data reduction mechanism deployed at both gateways and the edge tier. At the gateways, we apply the PIP (Perceptually Important Point) method to represent the features of a time series by using a finite amount of data. We extend such an algorithm by introducing several techniques, namely interval restriction, dynamic caching and weighted sequence selection. At the edge tier, we propose a data fusion method based on an optimal set selection. Such a method employs a simple strategy to fuse the data in the same time domain for a specific location.

Finally, we evaluate the performance of the proposed filtering and the fusion technique. The obtained results demonstrate the efficiency of the proposed mechanism in terms of time and accuracy.

Alkuperäiskieli | Englanti |
---|---|

Otsikko | Proceedings of the Seventh International Conference on the Internet of Things |

Kustantaja | IEEE |

Sivut | 1-8 |

ISBN (elektroninen) | 978-1-4503-5318-2 |

DOI - pysyväislinkit | |

Tila | Julkaistu - 2017 |

OKM-julkaisutyyppi | A4 Artikkeli konferenssijulkaisuussa |

Tapahtuma | INTERNATIONAL CONFERENCE ON INTERNET OF THINGS - Johannes Kepler University Linz, Altenberger Strasse 69, 4040 Linz, Austria, Linz, Itävalta Kesto: 22 lokakuuta 2017 → 25 lokakuuta 2017 Konferenssinumero: 7 http://iot-conference.org/iot2017/ |

### Conference

Conference | INTERNATIONAL CONFERENCE ON INTERNET OF THINGS |
---|---|

Lyhennettä | IoT |

Maa | Itävalta |

Kaupunki | Linz |

Ajanjakso | 22/10/2017 → 25/10/2017 |

www-osoite |

## Sormenjälki Sukella tutkimusaiheisiin 'A Multi-tier Data Reduction Mechanism for IoT Sensors'. Ne muodostavat yhdessä ainutlaatuisen sormenjäljen.

## Siteeraa tätä

Feng, L., Kortoci, P., & Liu, Y. (2017). A Multi-tier Data Reduction Mechanism for IoT Sensors. teoksessa

*Proceedings of the Seventh International Conference on the Internet of Things*(Sivut 1-8). [6] IEEE. https://doi.org/10.1145/3131542.3131557