TY - GEN

T1 - Pulse Amplitude Modulation for Electro-optical Spiking Neural Networks

AU - Rajbhandari, Sujan

AU - Uleru, George-Iulian

AU - Hulea, Mircea

AU - Younus, Othman Isam

AU - Ghassemlooy, Zabih

PY - 2022/9/1

Y1 - 2022/9/1

N2 - Spiking neurons represent the most accurate modelof the neural cells by using pulses and timing for informationprocessing and adaptation. Visible light communication can beleveraged to establish a wireless link between neurons in spikingnetworks even when neural areas are in relative motions.Typically, parallel transmission in electro-optical spiking neuralnetworks is performed using wavelength division multiplexing,which is limited by the number of wavelengths used andmultiple bandpass optical filters. This paper explores thepossibility of using multi-level pulse amplitude modulation(PAM) in multi-input-optical-axons (MIOA) integrated by theparallel neural paths in a spiking neuron network (SNN). Toevaluate PAM-MIOA, we implement an electro-optical SNNthat controls the force of two anthropomorphic fingers actuatedby the shape memory alloy (SMA)-based actuators. The voltagethreshold level in PAM is automatically adjusted based on thereference optical power. Results show that the electro-opticalSNN is able to hold an object when using PAM-MIOA even withthe link misalignment.

AB - Spiking neurons represent the most accurate modelof the neural cells by using pulses and timing for informationprocessing and adaptation. Visible light communication can beleveraged to establish a wireless link between neurons in spikingnetworks even when neural areas are in relative motions.Typically, parallel transmission in electro-optical spiking neuralnetworks is performed using wavelength division multiplexing,which is limited by the number of wavelengths used andmultiple bandpass optical filters. This paper explores thepossibility of using multi-level pulse amplitude modulation(PAM) in multi-input-optical-axons (MIOA) integrated by theparallel neural paths in a spiking neuron network (SNN). Toevaluate PAM-MIOA, we implement an electro-optical SNNthat controls the force of two anthropomorphic fingers actuatedby the shape memory alloy (SMA)-based actuators. The voltagethreshold level in PAM is automatically adjusted based on thereference optical power. Results show that the electro-opticalSNN is able to hold an object when using PAM-MIOA even withthe link misalignment.

M3 - Conference contribution

BT - 2022 13th International Symposium on Communication Systems, Networks and Digital Signal Processing (CSNDSP)

ER -