Uniflow Technical Overview

1. Executive Summary

Uniflow is a high-performance, industrial-grade middleware platform designed to bridge the gap between diverse hardware protocols, enterprise software, and cloud ecosystems. It runs as a cross-platform service on Windows, Linux, and ARM64, providing a robust, scalable, and secure "connective tissue" for mission-critical data orchestration.

Uniflow solves protocol fragmentation by providing a unified data normalization layer and a powerful, low-latency visual rule engine. It enables real-time logic execution, cross-protocol translation, and centralized telemetry monitoring without custom coding for every integration.

2. System Architecture: Distributed Client-Server Model

Uniflow utilizes a decoupled, independent client-server architecture powered by SignalR over WebSockets. This maximizes engine stability, execution performance, remote scalability, and ease of multi-node management.

2.1 Component Breakdown

2.2 System Component Architecture Diagram

+-------------------------------------------------------------------+
|               Uniflow Independent Administration GUI              |
|   +-------------------+  +-------------------+  +---------------+ |
|   |  WinForms/WPF UI  |  |  Topology View    |  | Rule & Source | |
|   |   Fluent Shell    |  |                   |  | Configuration | |
|   +-------------------+  +-------------------+  +---------------+ |
|                                |                                  |
|                     SignalR Client Manager                        |
+-------------------------------------------------------------------+
                                ||
                   SignalR WebSockets (RPC & Push)
                                ||
+-------------------------------------------------------------------+
|              Uniflow Background Service (Instance N)              |
|  +------------------+  +-------------------+  +-----------------+ |
|  |   SignalR Hub    |  | Rule Runtime Coord|  | Storage Registry| |
|  +------------------+  +-------------------+  +-----------------+ |
|                                |                                  |
|                    Plugin Manager (ALC Sandbox)                   |
|         +--------------+---------------+--------------+           |
|         |              |               |              |           |
|     Modbus/OPC UA  MQTT/Cloud      SNMP/Network    VMS/Security   |
+-------------------------------------------------------------------+

2.3 Network Communication Flow (SignalR)

The decoupled GUI and Service engine communicate over SignalR WebSockets, allowing transparent local or remote network connections:

• Command RPCs: Synchronous request/response calls for configuration changes (e.g., starting/stopping source adapters, updating graph rules, deploying plugin configs).
• Real-Time Push Streams: High-frequency asynchronous telemetry streams pushing events, diagnostic logs, node state changes, and rule execution traces out to subscribed clients with sub-millisecond precision.

3. Dynamic Plugin Architecture & Data Ingestion

3.1 Sandboxed Assembly Loading (`AssemblyLoadContext`)

Rather than relying on a monolithic adapter codebase, Uniflow uses a dynamically loaded Plugin Architecture. Each protocol adapter (Modbus, OPC UA, SNMP, MQTT, File Access, ONVIF, etc.) is built as an independent assembly isolated within custom .NET AssemblyLoadContext sandboxes.

• Hot-Swappable Runtime: Protocol plugins can be loaded, updated, or replaced at runtime without restarting the main service daemon or interrupting active rule streams.
• Fault Isolation: Each plugin runs in a protected scope. A network fault or exception inside one protocol adapter will never crash the core engine or affect other active sources.
• Unified Contract: All plugins implement standardized interfaces for reading, writing, address browsing, and state polling.

3.2 Unified Data Normalization

Every incoming telemetry signal—regardless of whether it originates from a Modbus PLC register, an SNMP OID, an MQTT JSON payload, or an Avigilon VMS alarm—is normalized into a strongly typed format. This allows visual rules to process data generically without protocol-specific branching.

3.3 Address Space Auto-Discovery

Plugins support dynamic address space browsing out of the box. Users can discover OPC UA node trees, query SNMP MIB OID structures, browse local and network file paths, or inspect Modbus address maps directly from the UI visual catalog.

4. Visual Rule Engine & Execution Safeguards

4.1 Directed Acyclic Graph (DAG) Execution

The Uniflow rule engine evaluates logic as a visual Directed Acyclic Graph (DAG):

• Atomic Nodes: Computational blocks covering Triggers, Logic, Math, String manipulation, Flow Control, Bitwise packing, Cryptography, and Output Sinks.
• Topological Ordering: Graph nodes evaluate strictly when all upstream inputs are fully resolved, guaranteeing deterministic computational output.

4.2 Sequential Determinism

4.3 Performance Safeguards & Protection Mechanisms

• Recursion Safety Breaker (Max Trigger Depth): To prevent CPU starvation from infinite feedback loops (e.g., Rule A updating Storage X, which triggers Rule A), Uniflow enforces a configurable trigger depth limiter. If execution depth exceeds the limit, the engine safely halts the loop and raises an execution fault.
• Hardware Flooding Protection: Outbound write nodes feature configurable Cooldown (ms) and Debounce (ms) controls to prevent hammering physical PLCs or external APIs with rapid control bursts.
• Event Storm Priority Scheduling: During high-traffic events, rules are dispatched according to assigned `Priority Levels` (10 to 100), ensuring critical safety rules execute ahead of background logging routines.

4.4 Dynamic Schema & JSON Payload Processing

To bridge legacy industrial hardware with modern cloud endpoints, Uniflow includes dynamic JSON object processing nodes. The engine automatically flattens deeply nested JSON structures into strongly typed visual properties, supporting full JSONPath expression extraction and canvas-level JSON serialization/deserialization.

5. Hardened Protocol Security Architecture

Security is built directly into Uniflow's protocol implementations to ensure compliance with enterprise and critical infrastructure standards:

6. Interactive Topology & Real-Time Visualization

Uniflow's independent GUI provides a live view of system topology using a high-performance **D3.js graph engine** embedded inside a **WebView2** host container:

• Visual Signal Flow: Active connections and rules pulse visually on screen as data frames flow through the system, giving operators instant awareness of active logic.
• Node Health State Indicators: Nodes change color dynamically (Green = Healthy, Yellow = Warning/Reconnecting, Red = Fault) based on protocol connection status and rule evaluation results.
• Deep-Dive Telemetry Snapshots: Clicking any node opens instant access to raw data buffers, live tag value tables, and execution timing logs.

7. Deployment & Integration Patterns

Uniflow supports two primary deployment topologies, which can be deployed individually or combined in enterprise environments:

Mode A: The Protocol Bridge (`System A -> System B`)

Translates data in real time between incompatible protocols and system layers without requiring an intermediate database or SCADA host.

Example: Converting Modbus PLC holding registers to MQTT JSON payloads and publishing them directly to AWS IoT SiteWise or Azure IoT Hub.

Mode B: The Unified Aggregator (`Multiple -> One`)

Ingests telemetry from dozens of disparate hardware sources, normalizes the data, and exposes the aggregated dataset under a single standard interface.

Example: Collecting telemetry from SNMP network switches, Serial sensors, and HTTP endpoints, and hosting all points under a centralized OPC UA Server node space for a master SCADA system.

8. Comprehensive Protocol & Plugin Catalog

Uniflow provides native out-of-the-box support for over 29 industrial, IT, IoT, building automation, and physical security plugins: