Real-time simulation of biophysically accurate neuron models is essential for advanced neuromorphic computing and neuroprosthetic applications. The Hodgkin–Huxley (HH) model provides high biological fidelity but is computationally expensive, making large-scale FPGA implementations challenging. Existing hardware implementations typically trade biological accuracy for scalability or require substantial hardware resources to achieve real-time performance.

This thesis presents BrainLUT, a resource-efficient FPGA architecture that implements the nonlinear ion-channel computations of the HH model using neural network-generated lookup tables. The proposed approach leverages Quantisation-Aware Training (QAT) and adapted NeuraLUT and ReducedLUT methodologies to approximate complex HH functions with compact logic-based lookup tables, eliminating the need for expensive arithmetic evaluation while preserving model fidelity. Multiple hardware architectures and optimisation techniques, including structural decomposition, lookup table compression, derivative-based prediction, neural-network-based LUT generation, and reduced numerical precision, are explored to identify an optimal design.

The resulting proof-of-concept implementation simulates 1,250 Hodgkin–Huxley neurons in real time on a Xilinx Artix-7 XC7A200T FPGA operating at 125 MHz, utilising only 6,307 LUTs (4.71%), 15 BRAMs (3.01%), and 4 DSP blocks (0.54%). Although not achieving the highest absolute neuron count reported in the literature, the proposed architecture demonstrates substantially improved resource efficiency and neuron density compared with existing FPGA implementations of the Hodgkin–Huxley model. These results establish BrainLUT as a scalable and resource-efficient foundation for future large-scale, biophysically accurate neuromorphic systems. ... Read more

As technologies in embedded systems are rapidly advancing, possibilities are opening up in many research fields. Neuroscience is one of those fields. Research on the brain is crucial for medial applications such as seizure prevention, understanding neurological disorders like Parkinson's disease and developing Brain Machine Interfaces for prosthetics. In order to make accurate neural recordings, invasive techniques like Intracortical electrophysiology (IC-Ephys) are often required. Using tiny probes, neural signals are measured directly from the brain tissue. These experiments are almost always performed on animals because of the health risks that they pose.

The Neuroscience department at the Erasmus Medical Center (EMC) has been conducting these sorts of experiments on mice using either wired or battery-based neural recording devices (Neurologgers). These devices simply make neural recordings without processing the data in any other way. To get the most realistic neural response in the brain, the mouse should be free to move and not be limited by the recording device. Ideally, signal processing and spike detection is done immediately on the device, so that it outputs useful data instead of raw data.

For these reasons our group was given the task by the Neuroscience department to make a battery-based neural-logging device containing an FPGA for real-time signal processing and spike detection. In this thesis, the design process, implementation and validation of the hardware of this so called Neurologger is discussed.

We were able to make a compact design that weighs 4.139 g without battery. It has an FPGA which performs real-time signal processing and spike detection. Two neural probes can be attached to perform neural recordings in multiple parts of the brain. Finally, 64 channels can be recorded simultaneously at a sampling rate of 20 kHz.

This thesis also contains a proposal for an even smaller Neurologger design. Making specific design choices, more channels can be recorded with a device that has a significantly smaller surface area than the current design. ... Read more

Neural recorders capture electrical signals from brain tissue to study neural activity, but multi-channel systems generate large data volumes that challenge portable applications with limited storage and battery life. This project presents version three of the Neurologger, an ultra-lightweight FPGA-based neural recording system for real-time data acquisition and on board signal processing. The system uses specialized analog front-end chips to acquire neural signals from multiple channels. The FPGA implements processing pipelines for spike detection and sorting across all channels concurrently, achieving deterministic low-latency operation that sequential microcontrollers cannot match. By processing and classifying spikes in real-time, the system can substantially reduce data volume before storage, making it suitable for portable and closed-loop neural applications. ... Read more