The authors analyze the cause of bias in a fast Fourier transform (FFT)-based frequency domain signal-to-noise ratio (SNR) estimator by deriving the upper bound of the bias. The analysis is then used to propose a new frequency domain estimator using discrete cosine transform (DCT), which has lower bias. In addition, a criterion is proposed for use when maximum Doppler frequency information is available; it is based on the derived upper bound and can be used to improve the mean squared error (MSE) performance of the proposed DCT-based estimator. Simulation results show that the proposed estimator reduces the MSE remarkably by diminishing the bias.
All Science Journal Classification (ASJC) codes
- Computer Science Applications
- Electrical and Electronic Engineering
- Applied Mathematics