Blind mobile sensor calibration using an informed nonnegative matrix factorization with a relaxed rendezvous model

Dorffer, Clément; Puigt, Matthieu; Delmaire, Gilles; Roussel, Gilles

doi:10.1109/icassp.2016.7472216

Cited by 21 publications

(23 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This work opens many perspectives. We recently proposed some mobile sensor calibration techniques based on informed matrix factorization [12], [13], [14]. In future work, we will investigate some outlier-robust extensions of these approaches, using a similar low-rank modeling.…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

Outlier-robust calibration method for sensor networks

Dorffer¹,

Puigt²,

Delmaire³

et al. 2017

2017 IEEE International Workshop of Electronics, Control, Measurement, Signals and Their Application to Mechatronics (ECMSM)

Self Cite

View full text Add to dashboard Cite

Abstract-In this paper, we aim to blindly calibrate the responses of a sensor network whose outputs are possibly corrupted by outliers. In particular, we extend some well-known nullspacebased blind calibration approaches, proposed for fixed sensors with affine responses-i.e., with unknown gain and offset for each sensor-to that difficult case. These state-of-the-art approaches assume that the true data lie in a known lower dimensional subspace, so that in practice sensors can be calibrated by projection of the uncalibrated observations to this subspace. A robust extension was recently proposed in order to provide less sensitivity to noise. In this paper, we show that such methods (including the robust extensions) are very sensitive to outliers and we propose new extensions able to deal with such issues. For that purpose, we assume the outliers to be rare events, which can be modeled as a sparse contribution to the low-rank observed data. Using such an assumption, we separate sparse outliers from the low-rank data, so that we can perform calibration. We show that the proposed approach is able to handle up to 10% of outliers in the data without major impact on the calibration accuracy while state-of-the-art methods are already sensitive to the presence of one unique outlier.

show abstract

Section: Discussionmentioning

confidence: 99%

“…When the sensors are mobile, they can be in rendezvous, i.e., they are in the same spatio-temporal neighborhood, thus sensing the same phenomenon [6]. Such an assumption was recently used 1 in both micro- [8], [9], [10] and macro-calibration 2 [12], [13], [14]. However, in the case of fixed sensors, other assumptions are needed.…”

Section: Introductionmentioning

confidence: 99%

Outlier-robust calibration method for sensor networks

Dorffer¹,

Puigt²,

Delmaire³

et al. 2017

2017 IEEE International Workshop of Electronics, Control, Measurement, Signals and Their Application to Mechatronics (ECMSM)

Self Cite

View full text Add to dashboard Cite

show abstract

“…As a consequence, the blind calibration problem can be tackled by informed Nonnegative Matrix Factorization (NMF). We propose to that end four informed NMF techniques which were partially and preliminarily proposed in [17], [18]. In [17], we were showing how to write the calibration problem in an informed NMF framework while we extended it in [18] by adding some sparse assumptions in one column of one matrix factor.…”

Section: Introductionmentioning

confidence: 99%

Informed Nonnegative Matrix Factorization Methods for Mobile Sensor Network Calibration

Dorffer

Puigt

Delmaire

et al. 2018

IEEE Trans. on Signal and Inf. Process. over Networks

Self Cite

View full text Add to dashboard Cite

In this paper, we consider the problem of blindly calibrating a mobile sensor network-i.e., determining the gain and the offset of each sensor-from heterogeneous observations on a defined spatial area over time. For that purpose, we propose to revisit blind sensor calibration as an informed Nonnegative Matrix Factorization (NMF) problem with missing entries. In the considered framework, one matrix factor contains the calibration structure of the sensors-and especially the values of the sensed phenomenon-while the other one contains the calibration parameters of the whole sensor network. The available information is taken into account by using a specific parameterization of the NMF problem. Moreover, we also consider additional NMF constraints which can be independently taken into account, i.e., an approximate constraint over the mean calibration parameters and a sparse approximation of the sensed phenomenon over a known dictionary. The enhancement of our proposed approaches is investigated through more than 5000 simulations and is shown to be accurate for the considered application and to outperform a multi-hop micro-calibration technique as well as a method based on low-rank matrix completion and nonnegative least squares.

show abstract

“…When radiation changes after the coupling phase, the correction co-efficient does not fit the situation anymore and causes larger errors. However, as point (n) indicates, there is no clear difference between stations with open (Stations 1-5) and slightly obscuring surroundings (Stations [6][7][8][9][10][11][12][13][14][15][16]. In December, there are no significant differences between RMSE values at different station groups due to the low intensity of the SW radiation.…”

Section: Station Numbermentioning

confidence: 99%

“…Automatic calibration methods are thus used to eliminate the cumbersome and error-prone manual calibration [8]. One such automatic calibration method is the rendezvous model [9][10][11][12]. In the rendezvous model, observations by two or more sensors, mobile or stationary, are collected when the sensors are co-located, i.e.…”

Section: Introductionmentioning

confidence: 99%

Observing and forecasting road surface temperatures

Karsisto¹

View full text Add to dashboard Cite

Wintertime weather conditions can be hazardous for road traffic. Icy roads and poor visibility caused by snowfall increase the accident risk. Accurate forecasting of road conditions is important, because reliable and precise forecasts help the road maintenance personnel to plan their operations accordingly. Well timed maintenance operations increase safety and enable economical savings as unnecessary actions can be avoided. Drivers can also adjust their route plan and driving behaviour appropriately when warnings of hazardous conditions are given well beforehand. Road conditions are forecasted in the Finnish Meteorological Institute (FMI) with specialized road weather model. Before executing the actual forecast, the model is first initialized by feeding it with observation data. The quality of this data is essential for forecast accuracy, as the forecast is greatly dependent of the initial model state. Road weather stations have traditionally been one of the main sources of information, but their density is sparse especially in rural areas. Road surface temperature can vary considerably across the road network, so observations should be done in dense enough spatial scale. Nowadays it is possible to gather real time information from vehicles. Mobile sources provide observations with high spatial density and thus facilitate detecting the road stretches most prone to freezing. However, the quality of mobile observations should be assessed before implementing them to the road weather forecasting systems. This dissertation aims to answer to two research questions. Firstly, it has been studied how to best use available surface temperature observations in the road weather model initialization. Secondly , it has been studied how differences in two road weather models' physics affect to the surface temperature forecast accuracy. A method called coupling was implemented to the FMI road weather model. The main idea of the method is to adjust the incoming radiation flux so that the modelled surface temperature fits to the last observed value. The results show that this method improves considerably the short range surface temperature forecasts. Mobile surface temperature observations done with Teconer RTS411 were compared to road weather station measurements to assess the mobile data quality. According to the results, the mobile observations were on average 0.62 ○ C warmer than the road weather station measurements at 0 ○ C and in dry conditions. It was found out that the difference between mobile observations and road weather station measurements was dependent on the road status. A calibration equation for mobile observations was developed using linear mixed models to get mobile observations more in line with road weather station measurements. The effect of the mobile observations to the road surface temperature forecast accuracy was studied. According to the results, using the mobile observations calibrated with the developed equation improved the accuracy of road surface temperature forecasts compared to a theoretical situ...

show abstract

Blind mobile sensor calibration using an informed nonnegative matrix factorization with a relaxed rendezvous model

Cited by 21 publications

References 17 publications

Outlier-robust calibration method for sensor networks

Outlier-robust calibration method for sensor networks

Informed Nonnegative Matrix Factorization Methods for Mobile Sensor Network Calibration

Observing and forecasting road surface temperatures

Contact Info

Product

Resources

About