Ultra wide band (UWB) communication is expected to play a key role in next generation broadband intra vehicle wireless applications. The car compartment differs significantly from other well studied indoor or outdoor environments. Hence, channel sounding experiments are crucial for gaining a thorough knowledge of the UWB signal propagation characteristics in such a medium. Time domain channel sounding campaigns often employ some sort of deconvolution during measurement post processing as the measured signal in these experiments is the convolution of the channel response and the probing pulse which violates the Nyquist criterion. In this paper, a comparison of two variants of time-domain serial subtractive deconvolution algorithm, popularly known as CLEAN, is presented. Appropriate statistical metrics for assessing the relative merit of a deconvolution technique are identified in the context of intra vehicle UWB transmission, and the better algorithm is selected based on its performance over a standard IEEE channel simulation testbed. The chosen method is then applied to extract power delay profile and delay parameters from an empirical time domain sounding experiment performed inside a passenger car. The effects of passenger occupancy, transmitter receiver separation, and absence of direct transmission path are studied.