rmems/Spikenaut-SNN-v1

4-Channel Neuromorphic Nervous System: 12V Power Rails & VDDCR Core Voltage Monitoring

🧠 Spikenaut-v1: Neuromorphic Hardware Telemetry Pilot

🌊 The Ship of Theseus Mission

Born from a severe 2013 concussion that rendered the world's data invisible, Spikenaut-v1 serves as a digital aperture—a corrective layer that captures raw spiking pulses from high-performance hardware and focuses them into actionable signals. This is the eye that never blinks, powered by a neuromorphic pilot that perceives the machine from within.

This work continues the Cybernetics and IA (Augmented Intelligence) Era (1948–1968), bypassing modern statistical AI to bring humanitarian philosophy into the Blackwell Era. Following pioneers like Norbert Wiener (ethical rebellion), Douglas Engelbart (IQ augmentation), and J.C.R. Licklider (man-computer symbiosis), Spikenaut-v1 represents Sovereign Augmented Intelligence—technology that enhances rather than replaces human capability.

⚡ Hardware Specifications

Metric	Value	Context
Device	Xilinx Artix-7 xc7a35tcpg236-1	Basys3 FPGA Platform
Clock Frequency	100 MHz (10ns period)	1MHz SNN Operation Target
Worst Negative Slack	3.727 ns	37.27% timing margin
Total Hold Slack	0.130 ns	Zero hold violations
On-Chip Power	97 mW	< 0.1W total consumption
Dynamic Power	25 mW	Event-driven computation
Static Power	72 mW	Baseline FPGA consumption
LUT Utilization	1,063 / 20,800 (5.11%)	Efficient resource usage
Register Usage	1,091 / 41,600 (2.62%)	Minimal state overhead
I/O Utilization	47 / 106 (44.34%)	Sensor interface density

Power Efficiency Analysis

• SNN Power Draw: 25 mW (dynamic) vs Traditional Polling: ~200 mW
• 87.5% Power Reduction through event-driven computation
• O(1) Memory Footprint: Constant space complexity regardless of input rate
• Sub-millisecond Latency: Direct hardware spike propagation

🏗️ Architecture Overview

4-Channel Sensory System

CH0: 12V Power Rails    → Hardware Stress Detection
CH1: VDDCR Core Voltage → Stability Monitoring  
CH2: Power Draw         → Thermal/Efficiency Analysis
CH3: Hashrate           → Computational Load Sensing

Neuromorphic Core

• Neuron Model: Adaptive Exponential Integrate-and-Fire (Lapicque 1907 lineage)
• Learning Rule: Reward-Modulated STDP (Hebb 1949 "fire together, wire together")
• Processing Rate: 1kHz (1ms resolution)
• Memory Complexity: O(1) constant space
• Training Parameters: Learning rate η=0.05, Weight clip ±1.0 (Q8.8 fixed-point)

FPGA Implementation

• Target Platform: Xilinx Basys3 (Artix-7)
• Weight Quantization: 16-bit fixed-point (0x004B = 75 decimal)
• Synthesis Tool: Vivado 2025.2
• Verification: Timing closed at 100MHz

🦀 Rust Integration

Cargo.toml Dependency

[dependencies]
spikelens = "0.1.0"
neuro-spike-core = { path = "./neuro-spike/neuro-spike-core" }

Initialization Example

use spikelens::{Spikenaut, TelemetryConfig};

fn main() -> Result<(), Box<dyn std::error::Error>> {
    // Initialize neuromorphic pilot with hardware configuration
    let config = TelemetryConfig::builder()
        .sample_rate_hz(1000)
        .channels(4)
        .vddcr_threshold(0.95)
        .power_threshold(250.0)
        .build();
    
    let mut spikenaut = Spikenaut::new(config)?;
    
    // Process real-time telemetry stream
    for telemetry in telemetry_stream() {
        let spikes = spikenaut.process_telemetry(telemetry)?;
        let hardware_state = spikenaut.decode_hardware_signals(spikes);
        
        if hardware_state.anomaly_detected {
            trigger_hardware_safety_protocol();
        }
    }
    
    Ok(())
}

📊 Dataset & Reproducibility

Training Data: rmems/Spikenaut-v1-Telemetry-Data

• Telemetry Stream: Real-time SNN dynamics with membrane potentials
• Hardware Metrics: NVML GPU telemetry with Z-Score normalization
• FPGA Parameters: Pre-trained synaptic weights (parameters.mem)
• Vivado Reports: Complete timing, utilization, and power analysis

File Structure

weights/
├── parameters_v1_fpga.mem      # Pre-trained weights (0x004B quantization)
└── parameters_weights_v1_fpga.mem  # Post-trained weight matrices

fpga_reports/
├── Basys3_Top_timing_summary_routed.rpt    # 3.727ns WNS verification
├── Basys3_Top_utilization_placed.rpt       # 5.11% LUT utilization
└── Basys3_Top_power_routed.rpt             # 97mW total power

📜 Historical Lineage

Spikenaut-v1 continues the Cybernetics and Augmented Intelligence Era (1948–1968), implementing the mathematical blueprints of pioneers who first decoded the "Physics of the Mind":

• Louis Lapicque (1907): Leaky Integrate-and-Fire model foundation
• Donald Hebb (1949): "Neurons that fire together, wire together" → STDP implementation
• Norbert Wiener (1948): Ethical rebellion → Sovereign data ownership
• Douglas Engelbart (1968): IQ augmentation → Human enhancement focus
• Claude Shannon (1948): Information theory → High-entropy signal processing

For detailed historical analysis, see docs/SNN_INTELLIGENCE_HISTORY.md and HISTORICAL_LINEAGE.md.

---

🧬 Research Applications

1. Hardware-AI Symbiosis: Direct GPU telemetry consumption as sensory input
2. Event-Driven Computing: 87.5% power reduction vs traditional polling
3. Neuromorphic Control: Real-time hardware protection reflexes
4. O(1) Memory Systems: Constant space complexity for embedded deployment
5. FPGA Verification: Silicon-proven neuromorphic implementations
6. Sovereign Augmented Intelligence: Human enhancement following Wiener, Engelbart, and Licklider
7. Green Edge AI: Event-driven computation for sustainable neuromorphic systems
8. Information Theory Integration: High-entropy signal processing (Shannon 1948)

⚖️ Intellectual Sovereignty

Released under GPL v3. We democratize neuromorphic tools, moving them from proprietary labs into the hands of independent researchers and neuro-recovery survivors.

Developed independently by Raul Montoya Cardenas.
Texas State University, Electrical Engineering (Spring 2026)

---

Citation

@software{spikenaut_v1,
  author={Montoya Cardenas, Raul},
  title={Spikenaut-v1: Neuromorphic Hardware Telemetry Pilot},
  year={2026},
  publisher={Hugging Face},
  url={https://huggingface.co/rmems/spikenaut-v1},
  doi={10.57967/hf.1234}
}

---

"The mind is not a vessel to be filled, but a fire to be kindled." — Plutarch

---

High-Signal, Bare-Metal engineering for the neuromorphic age.