⥠RNG Comparison Test Suite v1.1âĄ
Compare ntrpRNG (Behavioral) vs cgRNDV (Pure Crypto) - Enhanced with Weighted Event Tracking
đĄ ntrpRNG requires 500 weighted events (~5-10s interaction)
âšī¸
Event Weights:
âĸ Keyboard: Ã3 (high entropy)
âĸ Mouse Down/Touch/Motion: Ã2
âĸ Mouse Move/Scroll: Ã1
Example: 100 keyboard + 200 mouse = 500 weighted events
đ¯ Collecting Entropy - Interact With Page!
KEYBOARD
0
weight: Ã3
MOUSE
0
weight: Ã1-2
TOUCH
0
weight: Ã1-2
SCROLL
0
weight: Ã1
OTHER
0
weight: Ã2
0
Weighted Events
0
Events/sec
~0s
Est. Time
đĄ Tip: Use keyboard for fastest collection (Ã3 weight)
RNG Types:
âĸ ntrpRNG v1.3.0: Behavioral entropy (weighted events) + crypto.getRandomValues() + salt + 5000 iterations
âĸ cgRNDV v1.1.0: Pure crypto.getRandomValues() only
âĸ Compare Both: Side-by-side statistical comparison
Weighted Event System (ntrpRNG):
âĸ Keyboard events: Ã3 weight (highest entropy: timing + key choice)
âĸ Mouse down/Touch/Motion: Ã2 weight (medium entropy)
âĸ Mouse move/Scroll: Ã1 weight (basic entropy)
âĸ Target: 500 weighted events (hardcoded for security)
âĸ Example: 100 keyboard + 100 mouse down + 100 mouse move = 100Ã3 + 100Ã2 + 100Ã1 = 600 weighted events â
Quick Settings:
âĸ FAST: 1K iterations (testing/dev)
âĸ BALANCED: 5K iterations (recommended)
âĸ SECURE: 10K iterations (production)
Parameter Ranges:
âĸ Dataset Size (KB): 1 - 10240
âĸ Hash Iterations: 100 - 50000
âĸ Salt Size (bytes): 16 - 128
Test Quality Scales:
âĸ Shannon Entropy: 7.9-8.0=Excellent | 7.5-7.9=Good | 7.0-7.5=Fair | <7.0=Poor
âĸ Monobit/Runs: 0-1.0=Excellent | 1.0-2.0=Good | 2.0-3.0=Fair | >3.0=Poor
âĸ Chi-Square: <250=Excellent | 250-293.25=Good | 293.25-350=Fair | >350=Poor
âĸ Serial Correlation: <0.05=Excellent | 0.05-0.1=Good | 0.1-0.2=Fair | >0.2=Poor
âĸ Avalanche Effect: 45-55%=Excellent | 40-60%=Good | 35-65%=Fair | else=Poor
Security Notes:
âĸ ntrpRNG v1.3.0 uses hardcoded 500 minimum events (cannot be overridden)
âĸ Dual-path XOR fortification mixes behavioral + CSPRNG entropy
âĸ Asynchronous batching prevents UI blocking during 5000 iterations