This paper deals with channel estimation for orthogonal frequency-division multiplexing (OFDM) in time-variant wireless propagation channels. We particularly consider the challenges of the IEEE 802.11p standard, which is the worldwide dominant system for vehicular communications. For historic reasons, 802.11p uses a pilot pattern that is identical to the pattern used in 802.11a, which was initially designed for the estimation of indoor channels with little or no time variations. Therefore, this pilot pattern violates the sampling theorem for channels with both large delay spread and large Doppler spread, as often occurs in vehicular communications. To remedy this problem, we design a robust iterative channel estimator based on a 2-D subspace spanned by generalized discrete prolate spheroidal sequences. Due to the tight subspace design, the iterative receiver is able to converge to the same bit error rate (BER) as a receiver with perfect channel knowledge. Furthermore, we propose a backward compatible modification of the 802.11p pilot pattern such that the number of iterations sufficient for convergence can be reduced by a factor of 2-3, strongly reducing implementation complexity.
- Enabling Digital Technologies