Multi-Path Encrypted Transport Daemon

Bond Hybrid Links.
Gradual failover. Real time.
Mesh. Offline.

Nexus Atlas bonds any combination of network links — satellite, cellular, microwave, radio, WiFi — into a single encrypted tunnel. Transport-agnostic. Protocol-agnostic. Completely transparent to applications. Real-time link monitoring with gradual failover — not binary up/down. Multi-path routing across a self-healing mesh network. No controller. No cloud. Works fully offline.

4+
Simultaneous Links
<1s
Failover Time
100%
Encrypted Traffic
$0
Controller Cost
0
Cloud Dependencies
Works Offline
No Internet Required
Near Real-Time
Continuous Monitoring
Single Binary
No Dependencies
Any Transport
Satellite · Cell · Radio · WiFi
Self-Healing
Automatic Rerouting
Network Topology

Multi-Path Architecture

Traffic flows through multiple physical links simultaneously. Kill links below to watch traffic automatically redistribute in real time.

Core Capabilities

Engineered for Resilience

Purpose-built for environments where connectivity is critical and single points of failure are unacceptable.

🔗

Multi-Path Bonding

Aggregate bandwidth across VSAT, LTE, microwave, WiFi, and Ethernet links simultaneously. True link diversity, not just failover.

Per-packet scheduling across all links
🌐

Heterogeneous Links

Bond any combination of transport types — satellite, cellular, microwave, fibre, WiFi, serial, LoRa. Each link monitored independently with its own quality profile.

Any transport · Any speed · Any latency
🔐

Military-Grade Encryption

Noise IK handshake with Curve25519 ECDH. ChaCha20-Poly1305 AEAD with automatic rekeying every 120s or 1 GiB. 2048-bit anti-replay window.

44 bytes protocol overhead
📊

Adaptive Link Monitoring

250ms probe intervals with EWMA-smoothed RTT, jitter, loss, and bandwidth estimation. Detects degradation before failure — not after.

Continuous quality scoring · No binary up/down

Intelligent Scheduling

Three strategies: lowest delivery time, weighted round-robin, and broadcast mode. Quality-aware path selection in real-time, per-packet granularity.

Scheduler adapts every 250ms
🧩

Smart Fragmentation

Automatic packet fragmentation and out-of-order reassembly across MTU boundaries. Zero-copy design for high throughput. Applications never know.

Transparent to applications
📢

Gossip Config Propagation

Push a configuration change from any single node and it automatically propagates to every other node in the mesh via gossip protocol over existing encrypted tunnels. No central server, no manual sync — one command reaches the entire fleet in under a second.

<1s convergence across 20+ nodes
🔀

Multi-Hop Mesh Routing

Traffic doesn't just bond across transports on a single node — it routes through other nodes in the network. If a drone is destroyed or a relay goes down, packets automatically find alternative multi-hop paths through surviving nodes. This is a second layer of resilience beyond link bonding: resilience at the network topology level.

Link bonding + topology routing = two layers of resilience
Intelligent Scheduling

Three Strategies, One Goal

Dynamically route packets based on real-time link quality. Choose the strategy that fits your deployment.

0
Packets Sent
0
Delivered
0%
Link Utilization
Effective Throughput
Lowest Delivery Time
Each packet is sent on the link with the lowest predicted delivery time — calculated from current RTT, queue depth, and EWMA quality score. The fastest link gets most traffic, but if it degrades, the scheduler seamlessly shifts to the next best option within one probe cycle (250ms).
When to use
VoIP, video conferencing, real-time control systems, remote desktop sessions, drone telemetry — any application where latency matters more than raw throughput.
Real-world example
Offshore platform: operator runs VoIP over 3 links. WiFi (2ms) carries voice, microwave (5ms) carries backup, VSAT (600ms) is deprioritised automatically. If WiFi goes down, voice switches to microwave in 250ms — no call drop.
System Design

Architecture Overview

From application layer through encryption and scheduling to physical transport links.

NEXUS ATLAS SYSTEM ARCHITECTURE APPLICATION LAYER Applications Any IP Traffic Virtual Interface Tunnel Endpoint NEXUS ATLAS ENGINE Encryption Fragmentation Scheduling Security Peer Management Link Intelligence TRANSPORT LAYER Socket Pool Per-link binding Wire Protocol Minimal overhead · Encrypted datagrams Session Management Automatic rekeying · Session rotation PHYSICAL LINKS VSAT Satellite Link High latency · Global reach ONLINE LTE / 5G Cellular Link Mobile · Wide coverage ONLINE Microwave Point-to-Point Low latency · High bandwidth ONLINE WiFi / Ethernet Local Network Lowest latency · Highest throughput ONLINE REMOTE PEER(S) Remote Nexus Atlas Peer Mirror architecture · Fully encrypted · Bidirectional NEXUS ATLAS · Multi-Path Encrypted Transport
Real-Time Monitoring

Operations Command Center

Live operations dashboard with real-time telemetry across all links, mesh control plane status, and scheduler analytics.

154.8
Aggregate Mbps
12,847
Packets / Sec
8/8
Nodes Synced
99.97%
Mesh Uptime
Throughput by Link (Real-Time)
Link Health
Mesh Control Plane
Link Quality Scores (EWMA)
RTT Distribution
Scheduler Decisions
Controller-Less Architecture

No Controller. No Cloud. No Single Point of Failure.

Nexus Atlas is a distributed mesh transport daemon. Every node holds a full replica of the configuration database. Changes propagate via gossip protocol + CRDTs over existing encrypted tunnels. No centralised controller, no cloud dependency, no internet connection required.

🔄

Gossip-Based Sync

Config changes spread via epidemic gossip over existing tunnels. Version vectors ensure convergence across 20+ nodes in under 1 second.

🧬

CRDT Conflict Resolution

Simultaneous edits at different nodes? CRDTs merge automatically — LWW-Registers for scalars, OR-Sets for peers and links.

🚀

Zero-Touch Onboarding

New node joins with minimal config. Connects to any existing node, receives full state via ConfigBulk. 30 seconds from power-on.

The Problem with SD-WAN

SD-WAN Has a Single Point of Failure

SD-WAN does connection-level switching — when a link fails, entire flows move. Nexus Atlas works at the packet level, splitting and fragmenting data across all links transparently. Applications never know.

SD-WAN

Centralised controllerRequired
Cloud dependencyOrchestrator + licensing
Controller cost$30-75K/yr
Failover granularityPer-flow
Failover speedSeconds
Degradation modelBinary up/down
Config propagationOnly if controller reachable
WAN outage behaviourIsland mode — stale config
Network partitionMinority loses management
Simultaneous config editsSerialised through controller
Works without InternetNo
Mesh routingNo
App awarenessFlows disrupted
VS

Nexus Atlas

Distributed meshEvery node
Cloud dependencyZero — works on isolated networks
Controller cost$0
Failover granularityPer-packet
Failover speed<1 second
Degradation modelGradual — continuous quality scoring
Config propagation<1s gossip across mesh
WAN outage behaviourNormal — sync over surviving links
Network partitionEach partition operates independently
Simultaneous config editsAuto-resolved — deterministic merge
Works without InternetFully autonomous
Mesh routingMulti-hop mesh
App awarenessTransparent
Distributed Mesh Control Plane

Controller Without a Controller

Every node holds a full replica of the configuration database. Push config from any node, kill links — gossip finds alternative paths.

0/8
Nodes Synced
Propagation Time
v1
Config Version
Gradual Degradation

Not Binary. Continuous.

SD-WAN sees links as up or down. Nexus Atlas measures continuous quality scores — detecting degradation before failure and shifting load proportionally.

t=0s — Link healthy
VSAT link: RTT 580ms, loss 0.2%, jitter 12ms. Quality score: 92/100. Scheduler assigns 12% of traffic.
SD-WAN: Link UP. Full traffic eligible.
t=30s — Early degradation detected
Storm approaching. EWMA detects RTT creeping to 640ms (+10%). Quality: 78/100. Scheduler reduces VSAT weight by 15%.
SD-WAN: Still sees link as UP. No action taken.
t=2min — Moderate degradation
RTT 850ms, loss 3.2%. Quality: 41/100. 80% traffic already on other links. Only bulk on VSAT.
SD-WAN: Link still UP (hasn't hit threshold). Applications stuttering.
t=5min — Link near failure
RTT 2400ms, loss 18%. Quality: 8/100. Only probes on VSAT. All workload on LTE + microwave. Zero user impact.
SD-WAN: Link finally DOWN. All flows dropped and re-established. Seconds of disruption.
t=6hr — Recovery
Storm passes. EWMA smoothly recovers: 8 → 25 → 52 → 78 → 91. Traffic gradually returns. No flapping.
SD-WAN: Link UP. All flows immediately shifted — potential for flapping.
Defense Applications

Military Use Cases

Built for contested, degraded, and operationally-limited environments where connectivity resilience is mission-critical.

🎯

Drone Control (1:1)

Single operator controlling one drone over bonded LTE + radio links. Seamless handoff between base stations. Zero-latency failover for critical control streams.

🐝

Drone Swarms

Multi-hop mesh routing across 6-50 drones. Self-healing when drones are destroyed. RF jamming mitigation through frequency diversity. 300ms failover.

⚔️

Tactical Operations

FOB-to-HQ communications bonding VSAT, HF radio, and cellular. EW mitigation through automatic link diversity. Works in SIGINT-contested environments.

🤖

Field Robots (EOD)

Mine clearance and explosive ordnance disposal robots operating over bonded radio links. Redundant control channels ensure operator never loses contact with asset.

🏗️

Warehouse Automation

Autonomous logistics robots loading/unloading cargo in forward bases. Bonded WiFi + mesh radio ensures continuous operation across large warehouse environments.

📡

Base-to-Base Comms

Multi-path encrypted tunnels between forward operating bases, bonding satellite, microwave, and tactical radio. Survives partial infrastructure destruction.

🚁

ISR Feed Distribution

Intelligence, surveillance, reconnaissance video flowing from airborne assets to ground stations via multi-hop mesh. Bandwidth aggregation for HD video streams.

🛡️

Electronic Warfare Resilience

Automatic frequency hopping across bonded links when RF jamming is detected. EWMA-based quality scoring shifts traffic away from jammed bands in under 300ms.

Deployment Scenarios

Built for Mission-Critical

Where connectivity failures are not an option.

🎖️

Defense / Military

FOB communications, tactical networks, and electronic warfare mitigation through link diversity and encryption.

🛢️

Offshore Oil & Gas

Bond VSAT + LTE + microwave on offshore platforms. Maintain operations even when satellite weather fades.

🚛

Mobile / Vehicle

Convoy and fleet connectivity with automatic cellular failover. Seamless handoff between towers and WiFi zones.

🚑

Public Safety

Emergency services, disaster response, and first responder networks requiring always-on redundant connectivity.

🏭

Industrial IoT / SCADA

Telemetry and control systems over diverse WAN paths. Critical infrastructure that demands zero downtime.

🏗️

Remote Sites

Construction, mining, and branch offices with diverse ISP uplinks for guaranteed connectivity.

Airborne Mesh Routing

Drone Swarms as an Encrypted Transport Fabric

10 drones + ground station across 4 frequency bands. Destroy drones, jam frequencies — watch the swarm self-heal in 300ms.

Active:10/10
Best Path:
Route:
Jammed:None
Frequencies:900M · 2.4G · 5G · 4G/5G
Offline-First Architecture

Designed to Work Without the Internet

No cloud. No licensing server. No phone-home. Every node is fully autonomous.

🎖️

Air-Gapped Military Networks

SIPR/JWICS environments where internet connectivity is forbidden. The mesh runs entirely over local encrypted tunnels. No external dependency.

🏗️

Isolated Construction Sites

Remote mining or construction with local radio mesh. No cellular, no satellite. Nodes discover each other and sync config over whatever links exist.

🌊

Offshore During Storms

8 platforms managing themselves when VSAT degrades. Config changes propagate via the mesh — through whichever links survive.

🚑

Disaster Response

Cell towers down, internet gone. Drop Nexus Atlas nodes with local radios. They self-organise into a mesh — no infrastructure needed.

🛩️

Contested Airspace

GPS denied, RF jammed, ground station intermittent. Drones form autonomous mesh, cache data when partitioned, flush when reconnected.

🏭

Industrial SCADA / OT

Air-gapped operational technology networks. Mesh transport bonds redundant serial/Ethernet links between PLCs and HMIs.

Core Capabilities

Built for the Impossible

Where connectivity is critical, environments are hostile, and traditional solutions collapse.

🔗

Distributed Control Plane

Gossip-based config sync via SWIM over existing tunnels. CRDTs for conflict-free merging. Every node is a full replica.

Convergence: <1s across 20 nodes
🛩️

Airborne Mesh Routing

Next-hop-aware path selection for drone swarms. Bellman-Ford with EWMA-weighted link cost and mobility prediction.

100ms beacon · 300ms failover
📊

Gradual Degradation

EWMA-smoothed quality replaces binary up/down. Traffic shifts proportionally as links degrade — not after they fail.

250ms probe · Continuous scoring
🔐

Military-Grade Crypto

Noise IK + Curve25519 + ChaCha20-Poly1305 AEAD. 2048-bit anti-replay. Auto rekey every 120s or 1 GiB.

44 bytes overhead
📡

Works Fully Offline

No internet, no cloud, no licensing. Every node is autonomous. Works on isolated networks and air-gapped environments.

Zero external dependencies

Per-Packet Aggregation

True bandwidth bonding — one TCP flow uses all links simultaneously. Fragment, encrypt, schedule across every available path.

3 scheduling strategies
Deployment & Operations

Configure, Monitor, Operate

Every node ships as a single binary. Configure it your way — through a web-based interface or a plain text config file. Monitor everything in real time.

🌐

Web-Based Management Interface

Each node exposes a local web interface for configuration and monitoring. Add or remove links, change scheduling strategy, manage peers, and push config changes across the mesh — all from a browser. No SSH required, no CLI knowledge needed. Ideal for field operators and NOC teams who need point-and-click control.

Browser-based · No client install · Responsive design
📄

Plain Text Configuration

Every setting lives in a human-readable config file that can be version-controlled, templated, and deployed with standard automation tools. Edit it with any text editor, push it via your existing CI/CD pipeline, or generate it programmatically. The daemon watches for changes and applies them without restart.

Human-readable · Version-controllable · Automation-friendly
📊

Live Diagnostics Dashboard

The web interface provides real-time visibility into link health, throughput per link, packet loss, latency trends, scheduler decisions, encryption session status, and mesh synchronisation state. Live charts update continuously — the same telemetry the daemon uses internally is exposed to operators.

Real-time charts · Link quality · Mesh status · Session info
🖥️

CLI Logging & Diagnostics

For headless deployments, embedded systems, or operators who prefer the terminal — the binary outputs structured logs with configurable verbosity levels. Pipe logs to syslog, journald, or any log aggregator. Run diagnostic commands to inspect link states, peer status, routing tables, and encryption sessions from the command line.

Structured logs · Configurable verbosity · CLI inspection tools
Technology

What Makes It Possible

Predictive Link Awareness

Every link is probed several times per second with lightweight heartbeats. Quality trends are smoothed mathematically so the system detects degradation early and reacts before humans notice — shifting traffic away from a failing link while it still has some capacity left.

Adaptive Path Discovery

Every node continuously discovers the best route to every other node, factoring in real-time link quality, congestion, and mobility. When topology changes, new paths are computed instantly — not configured manually.

Self-Healing Configuration

Change a setting on any node and it spreads to every other node automatically — even if some links are down. When nodes disagree, conflicts resolve themselves without human intervention. No central database, no coordinator.

Engineered for the Edge

Ships as a single lightweight binary with no external dependencies. Runs on everything from a data centre server to a small embedded board on a drone. No runtime overhead, no background services, no cloud phone-home. Just the binary and a config file.