Soil Channel Estimation Using An UWB Radar Sensor

Monitoring soil information is essential to the development of agriculture. Therefore, it is necessary to develop a high-performance, low-cost technique for soil information collection. The Ultra-wide band(UWB) radar sensor has many technology advantages, such as low power density, strong material-penetrating quality, low cost and low power implementation. Furthermore, radar sensor networks(RSNs) formed by multiple radar sensors, can achieve regional soil information monitoring. In our paper, we mainly studied the soil channels based on the UWB radar sensor. First of all, the propagation characteristics of UWB radar signal in soil environment were analyzed. According to these features, the soil channel models were established. Secondly, CIRs in soil with different moisture/p H value was estimated. And the relationship of CIRs and p H or moisture value was investigated.The primary tasks and conclusions are listed as follows.1. The methods of soil channel modeling and measurements are described. The statistical modeling approach is selected for the soil channel after investigating several kinds of UWB channel models. And the data measurement in time domain is chosen as the method for collecting soil channel data.2. The soil environment and measurement equipment are selected. The echoes from soil with different p H values or humidity values were obtained through the UWB radar.3. The channel impulse response extracted from measurements using the CLEAN algorithm. The parameters of several statistical models for the amplitudes, cluster time delays and multipath components time delays data have been evaluated by using MLE algorithm. Also the standard deviation of each parameters was also calculated.4. The amplitudes and time delays of the CIRs are analyzed from a statistical view. The RMSE between density curve and statistical histogram are calculated. It is proposed that the amplitude of soil multipath channels can be more accurately characterized by a generalized extreme value(GEV) model other than commonly used clutter models such as lognormal, Weibull or Rayleigh. The arrival of clusters can be modeled also as a GEV process other than Poisson process, while within each cluster, the arrival of subsequent multipath contributions or rays can be described as either a log-logistic or a lognormal process. Poisson distribution is not suitable for the time delay under this paper experimental conditions.5. The relation between the p H values and the model parameters are finally disclosed. The p H and humidity value almost do not impact on statistical parameters for CIR amplitude models, and also little change in parameters for cluster time delays models. But it play a much more important role in multipath time delays as the statistical parameters vary along with the change of p H value. However, there is not a liner relation betweent soil p H value and the statistical parameter.