(672e) Using Channel State Information for Estimating Moisture Content in Wood Via 5 GHz Wi-Fi

Kerul Suthar, Forest Dewberry, Zhihua Jiang, Q. Peter He

Moisture content monitoring is a very important factor in manufacturing for forest industry as well as pulp and paper industry. For a manufacturing process that uses wood as a raw material, it is important to have real-time knowledge about the moisture content of wood so that the process can be optimized and/or controlled to achieve satisfactory product quality. Besides the manufacturing industry, the energy exploitation, storage suitability, and delivery price also depend on the moisture content of wood as well. Both destructive and non-destructive methods have been developed for moisture detection and/or estimation in wood. Destructive methods such as oven drying are accurate, but they also lead to destructive changes in the properties of wood. In addition, they are usually off-line and time consuming and therefore cannot be used for online applications due to the time constraint [1]. On the other hand, several non-destructive techniques that are less time consuming and require less man power have been developed, which include capacitance method, resistance method, microwave method, NIR method, etc. But these techniques are relatively complex and costly to install and operate [2].

To address the above-mentioned limitations of existing techniques, we propose a non-destructive and economic approach based on 5 GHz Wi-Fi, and use Channel State Information (CSI) to predict the moisture content in wood chips. We extract CSI by modifying the open source device drivers for Intel Wi-Fi link 5300 Network Interface Card based on CSITool [3]. CSI contains information about the channel in form of individual data sub-carriers capturing indoor channel features like effect of scattering, fading and power decay with distance. Modern Wi-Fi systems are equipped with Orthogonal Frequency Division Multiplexing (OFDM), dividing the data in multiple orthogonal subcarrier groups which solves the issue of selective frequency fading [4] . CSI has been used for indoor localization, device free sensing including fall detection, activity recognition and heart rate monitoring [5]. In addition, CSI and phase difference data have been successfully used for multi-class moisture detection in wheat [5].

In this work, we collect CSI and phase difference data using IWL5300 NIC by configuring the transmitter and receiver in injection and monitor mode respectively. We use Lenovo ThinkPad systems equipped with Linux based OS 14.02 and kernel version 4.2 due to the version specific selectivity of CSI tool. Both systems are equipped with IWL5300 NIC with modified driver and firmware for data collection. Our work includes CSI data collection, preprocessing, outlier detection, offline training and online testing. The CSI data is collected for wood chips at different moisture levels for offline training of models. The statistics pattern analysis (SPA) framework [6,7] is adapted in this work to build a multivariate statistical model for predicting the moisture content in wood. The most significant difference between SPA based approach and other multivariate statistical methods, such as partial least squares (PLS), is that instead of extracting correlations between the variables and moisture content measurements, we extract correlations between the statistics of the variables and the moisture content measurements. The proposed method is compared with other multivariate statistical and machine learning methods, including PLS, long short-term memory (LSTM) network, etc. We also discuss some challenges of the project and potential solutions to address them.

References:

1. Govett, R., Mace, T., & Bowe, S. (2010). A practical guide for the determination of moisture content of woody biomass. University of Wisconsin, 20.

2. John Bogart, Kett US - Learn the six methods for determining moisture: Advantages, Deficiencies and who uses each method.

3. Halperin, D., Hu, W., Sheth, A., & Wetherall, D. (2011). Tool release: Gathering 802.11 n traces with channel state information. ACM SIGCOMM Computer Communication Review, 41(1), 53-53.

4. Wang, X., Mao, S., Pandey, S., & Agrawal, P. (2014). CA2T: Cooperative antenna arrays technique for pinpoint indoor localization. Procedia Computer Science, 34, 392-399.

5. Yang, W., Wang, X., Cao, S., Wang, H., & Mao, S. (2018, July). Multi-class wheat moisture detection with 5GHz Wi-Fi: A deep LSTM approach. In 2018 27th International Conference on Computer Communication and Networks (ICCCN)(pp. 1-9). IEEE.

6. Wang and Q. P. He, “Multivariate process monitoring based on statistics pattern analysis,” Ind. Eng. Chem. Res., vol. 49, pp. 7858–7869, 2010.

7. P. He and J. Wang, “Statistics Pattern Analysis - A New Process Monitoring Framework and Its Application to Semiconductor Batch Processes,” AIChE J., vol. 57, pp. 107–121, 2011.