Energy-efficient multipath ring network for heterogeneous clustered neuronal arrays

Simulating large spiking neural networks (SNN) with a high level ofrealism in a field programmable gate array (FPGA) requires efficientnetwork architectures that satisfy both resource and interconnect constraints, as well as changes in traffic patterns due to learning processes.Based on a clustered SNN simulator concept, in this thesis, an energy-efficient multipath ring network topology is presented for the neuron-to-neuron communication. It is compared in terms of its mathematicalproperties with other common network topology graphs after whichthe traffic distributions across it and a two dimensional torus network are estimated and contrasted. As a final characterization step, the energy-delay product of the multipath topology is estimated and compared with other low power architectures. In addition, a simplified binary tree is suggested as a network layer for handling configuration and input/output data that uses a custom channel protocol without the need for routing tables