Investigating the effects of tuning parameters on the orthogonal clustering algorithm in time delay estimation

Localization systems are most often based on time delay estimation (TDE) techniques. TDE techniques based on channel impulse response (CIR) are effective in reverberant environment such as indoors. A recently developed algorithm called Orthogonal Clustering (OC) algorithm is one such algorithm that estimates the CIR utilizing a sparse signal reconstruction approach. OC is based on low complexity Bayesian method utilizing the sparsity constraint, the sensing matrix structure and the a priori statistical information. In practical systems several parameters affect the performance of a localization system based on OC TDE. Therefore, it is necessary to analyze the performance of an algorithm when certain parameters vary. In this paper we investigate the effect of variations in different parameters on the performance of the OC algorithm used in an impulsive acoustic source localization (IASL) system.