SNR estimation in decibel domain over communication channels

The statistics of the signal estimator for estimating the signal value in decibel (dB) domain proposed in [1] is analyzed. The basic idea of the proposed estimator combines the linear averaging and log domain. Using the most frequently used averaging filter, i.e., sample mean averaging filter, we prove that the performance of the proposed method is very close to the one of the optimum estimator with an assumption of independent identically distributed signal samples under Nakagami fading channel model. Moreover, the proposed method does not need to explicitly estimate Nakagami parameter, m, which is required for the optimum estimator. For the correlated samples to be used for the estimation, which is more practical situation, we show from the simulation that the proposed estimator outperforms significantly the optimum minimum variance unbiased estimator obtained under the Rayleigh fading channels. We also show from the simulations that the proposed estimator works well for the multipath fading case and the low signal strength.