Two backward elimination based approaches for low order model selection

This paper addresses the problem of model selection. Three different backward elimination based approaches for low order model selection will be presented; a classic hierarchical MDL/AIC based approach and two low complexity approaches, a MDL/AIC based approach and F-statistic based approach. The performance of each method will be calculated over two different linear models scenario; a moving average filter and a recursive filter. First, using the techniques of least square estimator (LSE) , the model parameters is estimated, in the time domain. Based on these estimates, the proposed approaches are then applied to identity the true model, in other words the model corresponding to the true non-zero coefficients. Simulation results demonstrate the power of using each technique for model selection with low complexity in a low SNR environment. A comparison between the proposed schemes will also be presented.