M. Josie Ammer and Jan Rabaey

ISSSTA 2004

Abstract:

An approach to simultaneously improve the energy consumption and convergence time (given the input SNR and required estimation variance) of feed-forward data-aided frequency estimation is presented. Four well-known frequency estimation algorithms are compared using actual ASIC hardware implementations to verify the results. It is demonstrated how a modification to the algorithms can simultaneously achieve lower energy consumption and improved convergence time. For example, for an input SNR of 12dB and required estimation variance of 2*10-5, convergence time is decreased by a factor of 4 while decreasing the energy consumption by a factor of 4.3. Directions on how to apply these algorithms to spread spectrum systems are provided.