Implementation of cross correlation with stochastic computation in FPGA

Stochastic computation is an alternative method for computing where numbers are encoded as probabilistic values. In this paper, we investigate the use of stochastic computation in the general cross corellation operation which finds various applications such as in interaural time difference in sound localization algorithms. In this work, we evaluate the performance of stochastic computation in performing cross correlation between two phase-shifted signals through experimental results on an FPGA. The two analog signals are converted into binary format using an on-board ADC of the Altera DE-10 Nano board, while all conversion into stochastic form and cross-correlation are performed within the FPGA. The system has been implemented in an Altera Cyclone V FPGA and able to distinguish signal phase delays in the audio frequency range.