Statistical Optimization of WARP Radio Board Parameters for Frugal Spectrum Estimation Using AR Model

Accurate as well as an efficient Spectrum Estimation (SE) is the fundamental requirement of any practical wireless communication system for seamless connectivity. In this work an attempt has been undertaken to accurately estimate spectrum through Wireless Open Access Research Platform (WARP) Radio Board using Autoregressive (AR) signal modelling based on Time Series analysis. A sequential optimization technique of the baseband (BB) and radio frequency (RF) gain parameters defined as BB_TX (dB), BB_RX (dB), RF_TX (dB) and RF_RX (dB) of the Transmitter and Receiver radio board parameters of the WARP board has been adopted to achieve the frugal SE. The technique involves incremental improvements employing Yule Walker (YW) method using Levinson Durbin Algorithm (LDA) along with the concept of Confidence Interval (CI). The proposed statistical approach based on Regression Analysis for the design optimization of WARP board parameters has also been compared and established over the Akaike Information Criteria (AIC), Bayesian Information Criterion (BIC) and Minimum Description Length (MDL) approaches of the Machine Learning (ML) tools for the best model selection. The frugalities in terms of Data length, Lag order along with other radio gain parameters, received signal power and antenna distance have been achieved by exhaustive practical experimentation on the WARP radio boards. The proposed Statistical Analysis (SA) shows a significant reduction of Minimum Data Length (80%), Lag Order (96%) and RF/BB Transmitter Power (30%) compared to classical methods employing YW technique.