Real-time neuronal networks reconstruction using hierarchical systolic arrays

The correlation network of neurons emerges as an important mathematical framework for a spectrum of applications including neural modeling, brain disease prediction and brain-machine interface. However, construction of correlation network is computationally expensive, especially when the number of neurons is large and this prohibits realtime applications. This paper proposes a hardware architecture using hierarchical systolic arrays to reconstruct the correlation network. Through mapping an efficient algorithm for cross-correlation onto a massively parallel structure, the hardware can accomplish the network construction with extremely small delay. The proposed structure is evaluated using Field Programmable Gate Array (FPGA). Results show that our method is three orders of magnitudes faster than the software approach using desktop computer. This new method enables real-time network construction and leads to future novel devices of realtime neuronal network monitoring and rehabilitation.