Work/FleetMatrix
FLEET TECH · IOT · 2024

Orchestratingautonomous fleetsat millisecond precision.

// ClientFleetMatrix Inc.
// Timeline16 weeks · Q2 2024
// Our RoleIoT Systems · Mobile Engineering · Distributed Architecture
SCREEN_02 // TELEMETRY_STREAM|LATENCY: 18ms
STATUS: LIVE
LOC: US_CENTRAL_01
ENC: TLS-1.3 / AES-256
FLEETMATRIX

Real-time CAN-bus Ingestion & Dynamic Route Dispatch

01 / The Problem

Thousands of long-haul haulers. Dropped packets in rural dead zones. Zero dispatch visibility.

Managing 8,400+ commercial transport units across trans-continental corridors requires continuous situational awareness. Legacy cellular telemetry hardware dropped connection in mountainous terrain, leaving dispatchers blind to critical engine temperature alerts and driver fatigue states.

When connectivity was restored, simultaneous batch packet dumps overwhelmed centralized databases, triggering cascading lock contention and reporting delays of up to three hours.

FleetMatrix engaged Codedway to design a resilient edge-to-cloud telemetry mesh that buffers offline vehicular data safely and streams updates instantaneously the moment a cellular tower is acquired.

3.2 hrs
average telemetry lag during rural highway transits pre-FleetMatrix
14%
excess fuel burned due to unoptimized dispatch routing
1.2M
CAN-bus data points ingested per minute without packet drop
02 / Our Approach

Edge buffering first, guaranteed eventual consistency second.

Phase 01

Hardware Protocols (Wk 1–3)

Reverse-engineering J1939 CAN-bus protocols and embedded gateway firmware buffering limits.

// CANBUS_PARSE
Phase 02

Edge Broker Mesh (Wk 4–6)

Deploying geo-distributed MQTT brokers with deterministic topic partitioning and compaction.

// MQTT_CLUSTER
Phase 03

Mobile Console (Wk 7–13)

Building 60 FPS driver application and desktop dispatcher consoles with dynamic route graphs.

// FLUTTER_CANVAS
Phase 04

Stress Testing (Wk 14–16)

Simulating nationwide network partitions with chaos engineering and 100k synthetic haulers.

// CHAOS_VERIFIED
Codedway solved our dead-zone synchronization problem in two weeks after two other agencies had spent six months failing at it.
— VP of Engineering, FleetMatrix
03 / What We Built

Edge MQTT Ingestion Broker

Ultra-low overhead publish-subscribe topology handling 1.2M messages/min with sub-millisecond serialization.

Go · VerneMQ · AWS IoT

Dynamic Dispatch Optimization

Real-time heuristic vehicle routing recalculating delivery windows based on live traffic, weather, and hours-of-service.

Python · OR-Tools · Redis

Offline-First Mobile Driver App

Robust Flutter client caching telemetry locally on SQLite with automatic cryptographic reconciliation on reconnect.

Flutter · SQLite · Dart

Time-Series Telemetry Warehouse

Hypertable-partitioned storage capturing high-frequency vehicular sensor readings with 90-day retention compression.

TimescaleDB · PostgreSQL

Collision & Anomaly Alerting

Sub-200ms threshold triggers detecting sudden deceleration, hard cornering, and engine fault codes.

Apache Flink · Kafka

Dispatcher Fleet Command

High-density multi-screen operations dashboard rendering thousands of live assets smoothly on interactive vector maps.

React · Mapbox GL · WebSockets
04 / Results

Numbers don't lie.

<220ms
GLOBAL DISPATCH SYNC WINDOW
1.2M
EVENTS PROCESSED PER MINUTE
34%
REDUCTION IN IDLE FUEL CONSUMPTION
99.999%
DATA INGESTION RELIABILITY
We went from guessing where our haulers were to having live telemetry and automated dispatch routing that saved us seven figures in diesel.
— Chief Operations Officer, FleetMatrix
05 // ARCHITECTURAL DEEP DIVE & BENCHMARKS

FleetMatrix powers real-time operations across massive commercial fleets. The client needed a distributed edge telemetry pipeline capable of processing millions of vehicular telemetry signals without packet drop, even through rural dead zones.

[METRIC_AUDIT]Telemetry Scale

Ingesting over 1.2 million CAN-bus events every minute, the edge clustering architecture maintains a sub-second dispatch sync window with zero message loss during intermittent connectivity.

Engineering Objectives

  1. Edge-to-Cloud Resilience: Offline-first caching on vehicle gateway units with deterministic conflict resolution upon signal re-acquisition.
  2. Dynamic Route Dispatch: Graph-based re-routing algorithms calculating real-time traffic congestion, weather models, and driver duty hours.
  3. High-Frequency Ingestion: Scalable MQTT clusters connected to Amazon Kinesis and time-series datastores.

Solution Highlights

We implemented a dual-tier architecture:

  • On-device local buffering using embedded SQLite with cryptographic transaction logs.
  • Edge MQTT brokers distributing geo-sharded payloads directly into partition-keyed event streams.
  • Mobile dispatch driver console built with Flutter featuring 60 FPS real-time map clustering and telemetry visualization.
Next →

CortexSearch — Cognitive Search Engine & RAG Core

NEXT.JS · OPENAI · QDRANT
View case study →

Have a project? Let's scope it.

We'll tell you honestly if we're the right fit.