Evaluation of LoRa Network Link Quality in Complex Urban Environments Based on Static and Dynamic Parameters

Abstract

With the continuous development of IoT technology, the application of Long Rang wireless link transmission in urban environments has gradually increased, making reliable wireless link quality transmission particularly important. Traditional wireless link quality assessments in urban environments have not comprehensively considered the various factors affecting LoRa link quality, nor have they assessed link quality by integrating fundamental environmental factors with digital models. This paper proposes a LoRa link quality assessment model that combines static and dynamic environmental factors. First, this paper analyzes the factors influencing LoRa link quality transmission through extensive real-environment experiments. Then, the First Fresnel Zone concept is considered to simulate LoRa link transmission. This is done by constructing a three-dimensional model of the urban environment and a three-dimensional model of the First Fresnel Zone to measure the proportions of various factors affecting link transmission. This paper employs a Bidirectional Long Short-Term Memory method to establish a static environment evaluation model. The reliability of this model is demonstrated by comparing the Mean Squared Error between the scores of the natural environment and the predicted scores. By comparing with three other methods, it was found that the static parameter evaluation model used in this paper is optimal. Subsequently, the static parameter evaluation model scores were combined with dynamic parameters to establish a dynamic environment evaluation model through a CNN-GRU-Attention mechanism. This model considers the influencing factors of time series. The MSE comparison with natural environmental conditions revealed that the dynamic parameter evaluation model scores are sufficient to indicate the link quality in the current urban environment.

Keywords

• Bidirectional long short-term memory (BiLSTM)
• CNN-GRU-attention
• first fresnel zone (FFZ)
• Internet of Things (IoT)
• LoRa
• mean squared error (MSE)