Noise power estimation for broadcasting OFDM systems

Abstract

The estimation of the noise power is a core issue in wireless communication systems. In broadcasting, every OFDM frame starts with a preamble symbol, which facilitates the noise power estimation. However, the performance of preamble-based noise estimation schemes worsens in fast-changing environments and cannot efficiently track the noise variation well. In order to track the noise variation symbol to symbol, pilot-based schemes would be more fruitful. Accordingly, in this paper, we propose a novel pilot-based noise power estimation scheme for OFDM systems that estimate the noise power for each data symbol. It makes use of the pilot subcarriers reserved for channel estimation in the data symbol. The proposed algorithm computes the circular correlation between received signal and comb type sub-pilot subsequence twice to generate the noise power. Compared with the conventional preamble-based scheme, the proposed pilot-based noise algorithm generates near-ideal accuracy at different values of signal-to-noise ratio (SNR). Without the additional overheads of the preamble, the scheme offers the significant saving of computation over a conventional time domain noise estimator.