How Digital Twins and Edge Computing Power the Next IoT Wave

Technology Waves Overview

The digital era can be divided into three successive waves. The first wave was the PC‑based Internet, which lowered content creation costs and gave rise to portals and search engines. The second wave was mobile Internet, driven by smartphones and tablets, enabling personalized services. The current wave is the Internet of Everything (IoE), where devices, people, and services are digitally linked to form comprehensive digital twins of physical environments.

1. Device Twin and Connectivity

A device twin is a digital representation of a physical IoT device. Establishing a bidirectional link between the twin and the device enables the device to join a network and be managed programmatically.

Key practical elements:

JoyLink protocol – JD’s in‑house protocol that provides lightweight, secure onboarding and control for smart‑home devices.

HiLink – Huawei’s comparable protocol; both illustrate the fragmentation of IoT networking standards.

Industry alliances – CHIP (Connected Home over IP) and OLA (Open Lighting Alliance) are working toward unified device‑level specifications to break “IoT islands.”

By inserting a unified device‑twin layer above heterogeneous protocols, applications can address devices through a common API (e.g., getProperty(), setProperty(), invokeAction()), effectively translating dialects into a standard language.

2. Scene Twin and Cross‑Scene Interoperability

A scene twin extends the device‑twin concept to an entire physical space—home, hotel, factory, vehicle, etc. It models all devices, occupants, and services within the environment and the relationships among them.

Typical use case: when a resident returns home, the scene twin triggers a workflow that adjusts temperature, sets lighting levels, and starts media playback without explicit user commands. Because the scene logic is encoded in the twin, the same workflow can be reused across different locations (e.g., office → conference room) with minimal re‑configuration.

3. IoT Application Framework

IoT applications differ from traditional mobile or web apps in three dimensions:

They must run on heterogeneous hardware (sensors, actuators, gateways).

Interaction modes include voice, vision, gestures, and location‑based triggers.

Runtime can be distributed across cloud, edge, and device nodes.

A flexible framework therefore provides:

Device‑twin and scene‑twin APIs for uniform access.

Low‑code visual designers that generate the underlying service mesh.

Dynamic workload placement: compute‑intensive tasks (e.g., video analytics) run on edge boxes, while lightweight control logic runs on devices or in the cloud.

4. Edge Computing for Physical‑Site Digitization

Placing compute resources close to sensors reduces latency and bandwidth consumption. JD’s edge‑intelligent boxes illustrate this pattern:

Multiple camera streams are ingested locally, and computer‑vision algorithms (object detection, occupancy estimation) are executed on the edge, avoiding raw video upload to the cloud.

Edge nodes aggregate hundreds of electric‑meter connections, enabling incremental upgrades for national‑grid deployments without a massive centralized rollout.

In electric vehicles, the shift from distributed ECUs to a centralized high‑performance compute unit creates a natural edge node that handles sensor fusion, OTA updates, and real‑time control.

Edge deployment steps typically include:

# Example: Deploying a containerized vision service to an edge box
docker pull registry.example.com/vision-service:1.2.0
docker run -d \
  --name vision \
  --restart unless-stopped \
  -p 8080:8080 \
  -v /data/models:/models \
  registry.example.com/vision-service:1.2.0

5. Mining IoT Data and Building a Data Infrastructure

Massive, time‑series IoT streams soon exceed the data volumes of the PC and mobile eras. Specialized time‑series databases are required to store, query, and analyze this data efficiently.

TDEngine – a high‑performance, column‑oriented TSDB optimized for billions of rows per day.

Apache IoTDB – an open‑source TSDB that supports hierarchical device schemas and edge‑side ingestion.

Typical pipeline:

# Ingestion (edge → cloud)
mosquitto_sub -t 'sensors/+/+' | \
  iotdb-cli -u admin -p password -s insert

# Query (real‑time dashboard)
SELECT avg(temperature) FROM root.sg1.d1 
WHERE time > now() - 5m;

When device twins and scene twins enrich raw measurements with semantic context (e.g., “living‑room temperature” vs. generic sensor ID), downstream services can provide precise, real‑time autonomous actions such as predictive HVAC control or safety alerts.

6. Outlook

IoT forms the foundational infrastructure for the third technological wave. By standardizing device connectivity (device twin), unifying environment models (scene twin), and enabling distributed computation and data analytics, the ecosystem can support digital twins of homes, factories, cities, and vehicles. The long‑term value lies in a robust, interoperable stack rather than short‑term hype, and it will underpin future business‑model transformations and public‑service innovations.