A Spiking Neural Network with Visible Light Communications

Hulea, Mircea; Ghassemlooy, Zabih; Rajbhandari, Sujan

doi:10.1109/csndsp.2018.8471811

Cited by 3 publications

(6 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…At the receiving end, we used an optical filter to reduce the ambient light and therefore improve the signal-to-noise ratio. In [12,38], WDM systems with filters for parallel transmissions between optical neurons were reported, whereas a less costly version of WDM without filters was recently introduced in [54].…”

Section: Methodsmentioning

confidence: 99%

“…A number of different forms of neuromorphic hardware with bioinspired plasticity rules have been reported in the literature [49], including the neuron model used in this work [47]. The SOMA models the integration of input stimuli detection of the activation threshold and refractory period as in [12], and the spike generation and the plasticity mechanisms are modeled by the SYN [47]. Note that in our previous work [12] the optical channel was implemented between the SYN output and the postsynaptic SOMA input, whereas in the current research the optical connection is between the SOMA output and the SYN input.…”

Section: Electro-optical Spiking Neuronmentioning

confidence: 99%

“…The SOMA models the integration of input stimuli detection of the activation threshold and refractory period as in [12], and the spike generation and the plasticity mechanisms are modeled by the SYN [47]. Note that in our previous work [12] the optical channel was implemented between the SYN output and the postsynaptic SOMA input, whereas in the current research the optical connection is between the SOMA output and the SYN input. On activation of the SOMA, V IN reaches the neuron activation threshold voltage V TH , which for this neuron model is the transistor T 1 base-emitter voltage V BE .…”

Section: Electro-optical Spiking Neuronmentioning

confidence: 99%

“…Typically, the communication between SNs is performed using the discrete stimuli, in which the intensity of the postsynaptic response depends on w s [11]. Implementing the SNs in the analogue hardware domain enables parallel transmission of multiple colors (i.e., wavelength division multiplexing (WDM)) between neurons [12]. Optical neural networks (ONN) have been investigated offering a number of features including (i) the computational power of the neuromorphic hardware; (ii) high processing speed and energy efficiency [13][14][15][16]; (iii) use of the license-free optical spectrum (i.e., ultraviolet, visible, and infrared bands); and (iv) lower crosstalk and immunity to radio-frequency induced electromagnetic interference [17].…”

Section: Introductionmentioning

confidence: 99%

“…Similarly, in free space-based ONNs, the weighted signal is affected by the light intensity fluctuations during propagation, which depends on the transmission span. One possible option to mitigate the effect of intensity variation is to use electro-optical synapses, in which w s is encoded within the optical pulse width [12]. However, the stray capacitance of transistors affects transmission of lower values w s .…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Optical Axons for Electro-Optical Neural Networks

Hulea

Ghassemlooy

Rajbhandari

et al. 2020

Sensors

Self Cite

View full text Add to dashboard Cite

Recently, neuromorphic sensors, which convert analogue signals to spiking frequencies, have been reported for neurorobotics. In bio-inspired systems these sensors are connected to the main neural unit to perform post-processing of the sensor data. The performance of spiking neural networks has been improved using optical synapses, which offer parallel communications between the distanced neural areas but are sensitive to the intensity variations of the optical signal. For systems with several neuromorphic sensors, which are connected optically to the main unit, the use of optical synapses is not an advantage. To address this, in this paper we propose and experimentally verify optical axons with synapses activated optically using digital signals. The synaptic weights are encoded by the energy of the stimuli, which are then optically transmitted independently. We show that the optical intensity fluctuations and link’s misalignment result in delay in activation of the synapses. For the proposed optical axon, we have demonstrated line of sight transmission over a maximum link length of 190 cm with a delay of 8 μs. Furthermore, we show the axon delay as a function of the illuminance using a fitted model for which the root mean square error (RMS) similarity is 0.95.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Electro-optical Spiking Neuronmentioning

confidence: 99%

Section: Electro-optical Spiking Neuronmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Optical Axons for Electro-Optical Neural Networks

Hulea

Ghassemlooy

Rajbhandari

et al. 2020

Sensors

Self Cite

View full text Add to dashboard Cite

show abstract

Deep Learning Based Signal Detection for OFDM VLC Systems

Amran

Soltani

Yaghoobi

et al. 2020

2020 IEEE International Conference on Communications Workshops (ICC Workshops)

View full text Add to dashboard Cite

Visible light communication (VLC) has become increasingly popular and has sparked a wide interest from various research areas. In order to fully realize the potential of VLC and to provide seamless connectivity to users, the underlying channel model of the system must be carefully understood. However, in a practical environment which considers specific geometrical configurations of the network and user behavior effects, the optical channel can often become too complex to be modelled mathematically. In this work, we apply deep learning (DL) to design an effective signal detection scheme for an indoor VLC communication system. With the aid of DL, our system is able to learn directly from the transmitted and received symbols and can reliably detect the original transmitted signals during the real time implementation with only a limited instantaneous knowledge of the channel. The simulation results confirm that our model offers very close performance to the optimal maximum likelihood (ML) detection with perfect channel state information (CSI). We also consider specific indoor environments (e.g., using a hot spot model) to confirm the robustness of the learning based schemes in different indoor scenarios.

show abstract