RangingNet: A convolutional deep neural network based ranging model for wireless sensor networks (WSN)

Huafeng Wu, Weijun Wang, Jun Wang, Prasant Mohapatra

Abstract: Distance measurement based on signal strength indication (RSSI) remains a challenge in wireless sensor network. Traditional theoretical models cannot accurately describe the signal propagation in actual ground environment. Existing intelligent solutions such as BP neural network ranging model and adaptive neural fuzzy inference system work well in a specific measuring environment by effectively reducing the calculation error. However, they cannot work in different environments. In this paper, we develop a deep convolutional neural network named ranging model to address the problem. Our solution automatically extracts high-level semantics features of signal propagation in different ground environments according to both image data and actual signal attenuation data. The experimental results indicate that this model outperforms current schemes in terms of overall measurement accuracy in different test environments.