In vehicular ad hoc network (VANET), the accuracy of the channel estimation plays a key role toward obtaining considerable improvement to the physical layer performance. In this study, least square (LS) and minimum mean square error (MMSE) estimation techniques are discussed and their performances in different VANET transmission scenarios are thoroughly investigated. Furthermore, a modified scheme of these techniques to improve the performance of the system and reduce the complexity in such a way that suits VANET communications is developed. Discreet Fourier transform and singular value decomposition techniques are companied with LS and MMSE to develop a channel estimation scheme which suits VANET channel conditions and ensures performance improvement and remarkable complexity reduction. Comprehensive simulations results show that the developed scheme greatly improves the performance of the system compared with LS and substantially reduces the complexity of MMSE.