Updating the Channel Estimation in a Flat-Fading Channel

1556-6072/09/$25.00©2009IEEE ||| 71 Vehicular networks require accurate channel state information (CSI) to decode the received signal. Such knowledge is usually estimated using short training sequences. However, in vehicular networks, the channel coherence time is very small due to the high speeds of the nodes, therefore the channel estimate from the training is likely to become inaccurate as the decoding proceeds. Using shorter packets can improve the performance at the cost of increased overhead. In this article, we introduce a novel channel tracking algorithm for vertical Bell Labs layered spacetime (VBLAST) in vehicular networks with no change in the overhead. The algorithm uses a set of first order Kalman filters, therefore it has less complexity compared to existing tracking methods that use higher order filters. The developed algorithm uses the decoded symbols and the received signal after the VBLAST decoding process to improve the channel estimate. Simulation results show considerable improvement in mean square error (MSE) and bit error rate (BER) when using this algorithm compared to channel estimation by training only with a small increase in hardware complexity. The capacity of multipleinput, multiple-output (MIMO) systems was shown to increase with the number of antennas [1]. Several algorithms have been developed to achieve part of this capacity, including space-time block codes (STBC), space-time trellis codes (STTC) and Bell Labs layered space-time (BLAST) algorithms. Spacetime codes increase the reliability of the link making it possible to use higher modulations to achieve higher data rates. BLAST systems, on the other hand, assume the receive antennas are in a rich Rayleigh fading environment causing each antenna to receive an independent signal. VBLAST makes use of the channel state matrix (H) to decode the signal recursively. It starts decoding the signal that has the highest signal-to-noise ratio (SNR) then cancels its contribution (interference) from the received signal vector, thus achieving better performance than the zero forcing receiver. Other BLAST algorithms exist such as Ghassan Abdalla, Mosa Ali Abu-Rgheff, and Sidi-Mohammed Senouci Updating the Channel Estimation in a Flat-Fading Channel