Environmental Monitoring Benchmarks

 

Sensor networks monitoring a real environment are prone to faults or aging phenomena, whose impact affects the overall system performance. In fact, permanent or transient faults can influence the sensors, the analog electronics, and the digital part of the embedded system inducing, in the best case, functional errors in the processing chain. In turn, erroneous information generates a strong side effect on the subsequent control chain, leading to wrong decisions and inappropriate control actions.
Our research introduces a novel cognitive fault diagnosis system (FDS) for distributed sensor networks that takes advantage of spatial and temporal relationships among sensors.

The Rock Collapse Forecasting System scenario is an experiment which refers to a real-world dataset provided by a real-time monitoring system for rock fall forecasting designed by Politecnico di Milano group and deployed in the Alps. The benchmarks considered in the next sections are related to measurements coming from a new generation of intelligent clinometer sensors which have an internal thermal sensor to correct and compensate online the measurements.

In the following you can find :

  1. Benchmark #1: Stuck-at fault
  2. Benchmark #2: missing data






Stuck-at fault Benchmark

 

The Stuck-at fault Benchmark has been acquired by the fully-wireless monitoring system for landslide forecasting that has been deployed in Northen Italy (Towers of Rialba). The system is composed by 3 units and 14 sensors: Unit 1 and 2 are endowed with 5 sensors (Int./Ext. Temperature, Biaxial Clinometer, Crackmeter), while Unit 3 is endowed with 4 sensors (Int./Ext. Temperature, Biaxial Clinometer).
The length of the benchmark is 2100 samples (from August 1, 2011 to October 31th, 2011), while the sampling frequency is 1 sample/hour. 
The type of fault which is present in the benchmark is a stuck-at fault affecting all the units of the network. 


Download the Benchmark #1: Stuck-at fault


 

 

Dataset Description

Number of units:  3

  • Unit 1: 5 sensors (Int./Ext. Temperature, Biaxial Clinometer, Crackmeter)
  • Unit 2: 5 sensors (Int./Ext. Temperature, Biaxial Clinometer, Crackmeter)
  • Unit 3: 4 sensors (Int./Ext. Temperature, Biaxial Clinometer)

Length of the dataset: August 1, 2011 until October 31th, 2011 (2100 samples)
Sampling frequency: 1 sample/hour
Type of fault: Stuck-at fault (between samples 1280-1305) affecting all the sensors

 

References

C. Alippi, S. Ntalampiras, M. Roveri, ”A Cognitive Fault Diagnosis System for Sensor Networks”, IEEE Transactions on Neural Networks and Learning Systems, Vol. 24, No. 8, pp. 1213-1226, Aug. 2013.

C. Alippi, S. Ntalampiras, M. Roveri, "An HMM-based change detection method for intelligent embedded sensors", in Proc. IEEE International Joint Conference on Neural Networks (IEEE IJCNN 2012), Brisbane, Australia, June 10-15, 2012.

Alippi, C., Roveri, M., Trovò, F.,  “A self-building, cluster-based cognitive fault diagnosis system for sensor networks” submitted to IEEE Transactions on Neural Networks and Learning Systems, 2013

 


Missing data Benchmark


The Missing data Benchmark has been acquired by the same wireless-wired monitoring system for rock-collapse forecasting that has been deployed in Northern Italy (Towers of Rialba). The number of units of the system is 3. In this benchmark only temperature measurements have been considered. The length of the dataset is 12000 samples (from Jul. 25, 2012 to Oct. 17, 2012), while the sampling frequency is 6 samples/hour.
The type of existing fault is missing data induced by communication problems.

Download the Benchmark #2: missing data

 

 

 

Dataset Description

 

Number of units:  3

Sensors per unit:

  • Temperature sensors (INCLUDED IN THE BENCHMARK)
  • MEMS accelerometer/Geophone/Inclinometer/Crackmeter (NOT INCLUDED IN THE BENCHMARK)

Length of the dataset: Jul. 25, 2012 until Oct. 17, 2012 (12000 samples)
Sampling frequency: 6 sample/hour
Type of fault: Missing Data

 

 

 

We are committed to maintaining a public repository of i-Sense benchmarks in the spirit of cooperative scientific progress as promoted by digital open access guidelines to EU funded research oriented towards the dissemination of results and outcomes.

You are free to download a portion of the datasets for non-commercial research and educational purposes.  Work based on the dataset should cite our references papers listed with each benchmark.

All datasets and benchmarks on this page are published under the Creative Commons Attribution-NonCommercial-ShareAlike 3.0 License.

Creative Commons License