Flash-based Computing in-Memory Scheme for IOT

A novel flash-based computing in-memory scheme is presented to accelerate the multiply-accumulate (MAC) operation. Based on the novel scheme the hardware implementation and optimization object-oriented deep leaning neuron networks (DNN) is explored to achieve computation power with high efficiency and low power. Based on a 65nm node flash process, a 5-layer DNN is designed and evaluated by simulation to verify the feasibility of the presented flash-based computing in-memory scheme and the corresponding hardware implementations for the low power edge computing applications such as internet of things (IoT).