Technical Blueprint v4.0

One Score. Two Jobs.

Proof-of-Unity is dual-purpose: the same score elects block proposers AND weights FL gradient updates at consensus layer. Most "AI + blockchain" projects bolt smart contracts on top β€” they only see a hash. Our consensus sees the weights.

Three Node Types

dns

Masternode

High-performance computational anchors responsible for transaction sequencing and Multi-Group DAG management.

CPU: Consumer 8-Core CPU
RAM: 8GB RAM
NET: 100Mbps
Primary Authority Tier
verified_user
security

Guardian

Security watchdogs that validate Proof-of-Unity scores and penalize malicious behavior through rapid slashing protocols.

CPU: Consumer 8-Core CPU
RAM: 8GB RAM
NET: 50Mbps
Consensus Security Tier
devices

Lightnode

Edge-computing nodes providing data availability and state proofs for mobile and IoT devices across the ecosystem.

CPU: 4-Core ARM v9
RAM: 4GB RAM
NET: 1Mbps
Edge Participation Tier

Proof-of-Unity
Deep Dive

Unlike PoS, Proof-of-Unity (PoU) weights validator influence based on performance, uptime, and network integrity scores. This creates a meritocratic security layer that optimizes for the fastest and most reliable actors.

Network Availability 99.98%
Data Integrity 94.2%
Propogation Speed 88.5%
SCORE_ENGINE_PROMPT_v1.0.4

/* Unity Score Calculation */

const calculateUnityScore = (node) => {

const avail = node.getAvailability(); // 25%

const integrity = node.getIntegrity(); // 25%

const rep = node.getReputation(); // 20%

const part = node.getParticipation();// 15%

const lat = node.getLatencyScore(); // 15%


return 0.3 * prevScore + 0.7 *

(avail*0.25 + integrity*0.25 +

rep*0.20 + part*0.15 + lat*0.15);

};


>> RUNNING LIVE VALIDATION...

>> SCORE: 0.948271

Multi-Group DAG Flow

Parallel transaction execution through asynchronous vertex propagation and dynamic grouping.

A B C Aβ‚€ A₁ Aβ‚‚ A₃ Aβ‚„ Bβ‚€ B₁ Bβ‚‚ B₃ Bβ‚„ Cβ‚€ C₁ Cβ‚‚ C₃ Cβ‚„ BFT FINAL t=0 t=1 t=2 t=3 t=4 finality
Group A
Group B
Group C
Cross-group causality (multi-parent)
STEP 01

Local Grouping

Transactions are assigned to localized shards based on state-access affinity.

STEP 02

Vertex Casting

Nodes cast vertices to neighboring groups to establish causality across the DAG.

STEP 03

Parallel Consensus

Each shard achieves consensus independently before global ordering.

STEP 04

Asymptotic Finality

Vertices consolidate into a singular global state with zero-latency confirmation.

Benchmarks

Real-world performance simulation under varying load conditions.

Current Environment

Small Cluster

  • Active Shards 04
  • Total Node Count 128
  • Simulation Time 3,600s
91,650
Avg. TPS
12ms
Inter-Node Lag
0.00%
Fail Rate

Security Model

Threat: Sybil

Identity Masking

PoU requires hardware-attested identity and minimum staked collateral per node, making multi-identity creation economically unfeasible.

check_circle Mitigated via PoU Staking
Threat: Eclipse

Network Isolation

Dynamic neighbor selection and periodic peer-rotations ensure that no node can be isolated by a cluster of malicious actors.

check_circle Mitigated via Peer Rotation
Threat: Long-Range

Chain Re-writing

Checkpointing via Guardian nodes creates cryptographic anchors that prevent historical state manipulation beyond the finality window.

check_circle Mitigated via Checkpointing

Scale the Future

Start your node today and contribute to the kinetic security of the Savitri Network.