Optimizing Automotive MIMO Radar Detection With Regression Modeling-Based Correlation Test

Persistent automation of driving functions results in development of advanced driver assistance systems (ADAS) into fully autonomous driving systems. The environmental sensing capabilities of these systems have a significant impact on their performance, reliability, and safety. Accordingly, the automotive radar is an indispensable technology in the evolution of modern vehicles. In this paper, the signal model is reformulated to be a linear model (LM). Under the framework of regression analysis, Breusch-Godfrey test (BGT)-based target detection technique is investigated. The BGT is an efficient test for higher-order autocorrelation in regression residuals. Powerful statistical tools, such as ordinary least squares (OLS) and Yule-Walker equations are utilized to estimate the parameters of the regression models. Akaike information criterion (AIC) and Bayesian information criterion (BIC) are applied to select the model that best explains the data with the least number of parameters. They achieve a balance between goodness-of-fit (GoF) and model complexity. The proposed BGT-based detection technique outperforms the state-of-the-art detection techniques in different scenarios of simulations. In addition, the proposed detector is able to discriminate between the target vehicle, interfering vehicle, and Weibull-distributed clutter. Furthermore, closed-form expression for theoretical threshold is derived, which it also exhibited a full compatibility with the simulated one.