Estimation of Reflection Coefficient on Microwave Transmission Lines Using Autoregressive Signal Processing

Microwave energy has been extensively studied and used in various medical applications. Some of them involves the use of generator to generate, transmit, and emit the energy into human body as a mean of therapy. In this case, accurate measurement and/or manipulation of the microwave is necessary in order to ensure safety to the patients. However, this is a challenging task due to the interferences from the reflected waves caused by mismatch at each connection. Time domain analysis (TDA) using a vector network analyzer (VNA) is a commonly employed technique to separate the reflected waves of the antenna from those of the antenna. However, its requirement for sweeping over a broad frequency band renders it impractical for integration within a generator. To address this limitation, we propose a novel method for estimating reflection coefficients utilizing an autoregressive (AR) model. The proposed method offers the advantage of estimating the distance and reflection coefficient of each reflector simultaneously by sweeping over a narrow frequency band and measuring the reflection amplitude and phase. To validate the efficacy of this approach, simulations were conducted, demonstrating that the implemented AR model works correctly. Furthermore, experimental validation was performed with 10 sets of Devices Under Test (DUTs) fabricated. The reflection coefficients at 2.45 GHz were estimated for coaxial-slot antennas in air, water, and isopropyl alcohol (IPA). The error rates of the estimated reflection coefficients were found to be −1.38 ± 3.69% for air, −7.85 ± 17.8% for water, and −0.551 ± 8.29% for IPA, respectively. In the case of water, where even with an error of +2σ from the mean value, the difference between the mean radiated power of DUTs and the estimated one was only 1.37%, indicating that the estimation could be performed with sufficient accuracy. In conclusion, our proposed estimation method is promising for accurate estimation radiated power and detection abnormal conditions in microwave medical devices.