Real-time DSP Based Time Delay Estimation Using Two Sensors

There are few available approaches for time delay estimation (TDE) which offer with a potential to be implemented as real-time embedded systems. Most of these lose accuracy in reverberant and highly noisy environments. In this paper, we initially proposed a 2D signal based on the cross-correlation. Then we proposed two approaches for TDE, based on this 2D signal. There are two main advantages for using the proposed approaches. The first merit is that using a large amount of information about the correlation between the received signals allows a more robust TDE against noise and reverberation, when compared to the other possible techniques. The second one is that the number of samples to calculate the TDE algorithms are decrease significantly, which is reduce the computational complexity. This paper also proposes a feasible hardware structure to implement the proposed approaches on digital signal processor (DSP) embedded systems in real-time.

Time delay estimation; Cross-correlation; Bufferable cross-correlation matrix; Source localization; Digital signal processor (DSP)