Improving Prediction Accuracy for WSN Data Reduction by Applying Multivariate Spatio-Temporal Correlation

Abstract

This paper proposes a method based on multivariate spatial and temporal correlation to improve prediction accuracy in data reduction for Wireless Sensor Networks (WSN). Prediction of data not sent to the sink node is a technique used to save energy in WSNs by reducing the amount of data traffic. However, it may not be very accurate. Simulations were made involving simple linear regression and multiple linear regression functions to assess the performance of the proposed method. The results show a higher correlation between gathered inputs when compared to time, which is an independent variable widely used for prediction and forecasting. Prediction accuracy is lower when simple linear regression is used, whereas multiple linear regression is the most accurate one. In addition to that, our proposal outperforms some current solutions by about 50% in humidity prediction and 21% in light prediction. To the best of our knowledge, we believe that we are probably the first to address prediction based on multivariate correlation for WSN data reduction.

Application of wireless sensor networks for beehive monitoring and in-hive thermal patterns detection

“…(2)) proposed by Jurdak et al (2008). This power model was chosen for its comprehensive implementation and for being used in other WSN work of the research group to which this work is linked Carvalho et al (2011), Hermeto et al (2014 and Kridi et al (2014): where E t represents the energy cost in millijoules (mJ) for packet transmissions; P send is the quantity of packets sent; P size is the size in bytes of each packet (4 bytes referring to a float number that represents the temperature sent), considering payload only; T B represents the time required for the XBee module to send 1 byte (32); I t is the electrical current value in the wireless module in transmit mode (45 mA) and V is the voltage supplied to the device. Note that we do not analyze packet receive costs, since we are interested to see how much of these will be saved by reducing the node to send data.…”

A Survey About Prediction-Based Data Reduction in Wireless Sensor Networks

“…Most of the surveyed works use predictions models in their solutions with very low or absolutely no mathematical basis, i.e., authors usually ignore the existence of related works in statistics when deciding which prediction method can be the best one for their scenario, which decreases the reliability of their mechanisms. For example, most of the works ([Yann-Ael and Gianluca 2005;Tulone and Madden 2006;Deshpandem et al 2004;Chu et al 2006;Debono and Borg 2008;Jiang et al 2011;Min and Chung 2010;Askari Moghadam and Keshmirpour 2011;Stojkoska et al 2011;Carvalho et al 2011;Aderohunmu et al 2013a;Yin et al 2015;Raza et al 2015;Wu et al 2016]) are based on the dataset from the experiments described in [Madden 2004]. However, each work takes its own decision about which prediction method to use, i.e., none of them incorporate tools to properly analyze the data and find out its characteristics before choosing the prediction method that best fits to their requirements.…”

Avoiding Data Traffic on Smart Grid Communication System

“…Furthermore, the lossless compression [8] may also be impractical because that requires memory and processing devices which are already constrained computational resources. In our previous works [9] recommend that data modeling should be done by source node and sent to the sink node. That approach enables the sensor node make decisions instantly, regardless of the transmission delay of the model.…”

“…The first modification of the solution proposed in [9], which was used in a fixed samples window size in the calculation of the coefficients, consists of using an adaptive window of readings (w) and guided by the prediction error. This error (Error predictor ) is calculated by the sum of difference between the actual sample value (y i actual ) and the sample value predicted (y i predicted ).…”

“…Works, such as [9] apply data prediction to reduce traffic from sensor node to the sink. An algorithm is embedded within the sensor node to modeling data from readings.…”