AIに何でも質問してください

EC3320
AIエッジコンピュータ

The EC3320 is a high-performance AI edge inference computer based on the RK3588 platform, designed for smart manufacturing, smart cities, and energy applications, enabling real-time intelligent decision-making.

PHOTO

High-Performance Industrial AI Edge Inference Platform

Powerful AI computing platform

Powered by the RK3588 ARM multi-core processor with integrated NPU, supporting multiple AI frameworks and edge intelligence applications.

Rich Industrial Interfaces

Supports Ethernet, serial ports, CAN, USB, HDMI, and DI/DO for diverse industrial applications.

Stable multi-network connectivity

Supports wired, cellular, and Wi-Fi connectivity with dual SIM and link redundancy.

Advanced security design

Supports Secure Boot, TPM 2.0, and system hardening for enhanced device and data security.

Industrial-grade reliability

Fanless design with industrial EMC compliance and wide temperature operation.

Cloud-based device management

Integrated with InHand DeviceLive for remote device, application, and container management.

  • Core Technical Specifications
    Cellular Network 5G NR / 4G Cat6 (model-dependent)
    Cloud Management DeviceLive, HTTP/HTTPS/SSH remote management
    CPU 4×Cortex-A76 + 4×Cortex-A55, up to 2.4GHz
    Data Acquisition Modbus RTU/TCP, EtherNet/IP, OPC UA, DNP3.0, BACnet, CNC
    GPU Mali-G610 MC4
    Interfaces 3×GE, 2×RS232 + 2×RS485, CAN, DI/DO, USB3.0, HDMI, MIC/SPK
    Memory / Storage 8GB / 64GB eMMC
    Network Features Dual SIM backup, multi-level link detection, auto-redial
    NPU Up to 6 TOPS (INT8), Hailo expansion up to 26 TOPS
    Open Platform Linux with edge app/container management
    Power Input 9~36V DC
    Security (TPM2.0 Optional) Secure Boot, TPM2.0
  • Hardware Platform
    CPU Quad Cortex-A76 + Quad Cortex-A55, up to 2.4GHz
    FLASH 64GB eMMC
    GPU Quad-core Mali-G610 MC4 high-performance GPU
    NPU Up to 6 TOPS (INT8)Support INT4/INT8/lNT16 mixed operations and framework switching of TensorFlow / MXNet / PyTorch / Caffe
    RAM 8GB
  • Connectivity & Interfaces
    Bluetooth BLE5.0
    Buttons 1×Power, 1×Reset
    CAN 1×CAN2.0
    Ethernet Ports 3×10/100/1000Mbps GE
    Expansion Interfaces 1x M.2(Supports SATA3.0 SSD, supports 2242 size)1x M.2(Supports Hailo-8 M.2 module, up to 26TOPs computing power)
    GPS Supported (requires cellular module support)
    HDMI 2×HDMI2.0
    I/O Ports 10pin industrial green terminal (4IN, 4OUT, 2GND)
    LED Indicators Power, Status, 4G, USER
    MIC Supported
    Serial Ports 2×RS232 + 2×RS485, industrial terminal
    SIM Card Holders 2×Nano SIM
    SPK Supported
    USB 4×USB3.0
    WiFi STA, 802.11ac/a/b/g/n, 2.4G/5G dual band
  • Power & Power Consumption
    Input Voltage 9~36V DC (2pin green industrial terminal, with flange)
    Power Interface Industrial DC terminal input
  • Mechanical Specifications
    Enclosure & Heat Dissipation Metal enclosure, fanless design
    Mounting Method DIN-rail / wall mounting
    Product Dimensions 180×136×54mm
    Protection Rating IP40
    TPM (Optional) TPM2.0
  • Environment & Certifications
    EMC Standard EN61000-4-2, level 3, StaticEN61000-4-3, level 3, Radiation Electric FieldEN61000-4-4, level 3, Pulsed Electric FieldEN61000-4-5, level 3, SurgeEN61000-4-6, level 3, Conducted Disturbance ImmunityEN61000-4-8, Power Frequency Field Resistance, horizontal / vertical 400A/m (>level 2)EN61000-4-12, level 3, Shock Wave Resistance
    Environmental Humidity 5~95% RH (non-condensing)
    Operating Temperature -20~60℃
    Physical Characteristics IEC60068-2-27 shock resistanceIEC60068-2-6 vibration resistanceIEC60068-2-32 drop resistance
    Storage Temperature -40~85℃
  • Operating System
    Operating System Linux
  • Network Features
    Network Access 5G NR / LTE Cat6 (model-dependent), Ethernet
  • 安全
    セキュアブート Supported
  • 信頼性
    Built-in Watchdog Embedded watchdog
    Dual SIM Switchover Supported
    Link Detection Multi-level link detection with auto-redial
  • Data Acquisition Protocols (DSA)
    Electricity Protocols DLT645-2007, IEC101/104, DNP3.0
    Industrial Protocols Modbus RTU Master/Slave, Modbus TCP Master/Slave, EtherNet/IP, ISO on TCP, OPC UA Client/Server, Mitsubishi MC 3C/3E/3C OverTCP, Mitsubishi CPU Port, FINSUDP, HostLink, PPI
    Other Protocols BACnet, CNC
  • Network Management
    DeviceLive Cloud Supports cloud-based parameter configuration, container management, application and firmware management
    Log Functions Local/remote logs
    Remote management DeviceLive / HTTP / HTTPS / SSH remote management
    Upgrade Method Supports patent upgrade mechanism, local or remote firmware upgrade
  • AI Expansion Module (Optional)
    None
    H8 Hailo-8, M.2 Key B+M 2280
Fetching resources...

Standard package*

  • EC3320 AI Edge Computer *1
  • 4Gアンテナ *1
  • 5G Antenna *4
  • Wi-Fi Antenna *2
  • イーサネットケーブル *1
  • Optional accessories
  • Power Adapter *1
  • AC Cable *1
  • Panel Mounting Kit *1

* Standard package contents vary by standard order codes.

Contact us directly to learn more about our packaging options.

DeviceLive

IoTデバイス管理プラットフォーム

Device management, remote monitoring, edge app management, and remote maintenance to help industrial enterprises build intelligent edge networks.

Frequently Asked Questions

What AI workloads is the EC3320 designed for?

The EC3320 is an AI edge computer with a quad-core ARM Cortex-A76 + quad-core Cortex-A55 CPU @ up to 2.4 GHz, Mali-G610 MC4 GPU, and a built-in 6 TOPS (INT8) NPUexpandable to up to 26 TOPS via an optional Hailo-8 M.2 module ( -H8 suffix). Designed for industrial AI inference at the edge — real-time data acquisition, protocol conversion, and AI application hosting. Rich industrial I/O: 3× Gigabit Ethernet, 2× RS-232 + 2× RS-485, CAN, 4× DI + 4× DO, Wi-Fi 5, BLE 5.0, dual Nano SIM. Target deployments include smart manufacturing, smart energy, smart city, smart healthcare, and public utilities. Cellular options: LTE Cat 6 (FQ09), 5G NR (NRQ3), or no-cellular (EN00).

The EC3320 is a Linux-based edge computer with 2× HDMI outputs and 4× USB 3.0 ports — three access methods are supported.

Direct (primary): Connect HDMI display + USB keyboard/mouse. Power on, select the System User account on the login screen, and enter the password printed on the device nameplate. Use the Linux desktop directly.

Remote — SSH: Connect to the default ETH2 IP 192.168.4.100 using an SSH client (MobaXterm, PuTTY, etc.). Log in with the System User credentials from the nameplate.

Remote — IEOS web: Open a browser to https://192.168.4.100:9100 (HTTPS only, port 9100). Log in as username adm with the nameplate password.

The EC3320 has a built-in 6 TOPS NPU and an on-board M.2 slot that accepts a Hailo-8 AI accelerator module (selected with the -H8 SKU suffix). With the Hailo-8 installed, AI compute is expandable to up to 26 TOPS for higher-density vision and inference workloads.

The EC3320 uses InHand DeviceLive for cloud management. Registration is done in the DeviceLive portal — no configuration is needed on the device’s web GUI. Log in to your DeviceLive tenant, click Add Device, and enter the EC3320’s serial number and credentials from the device label. Once the device has internet, it auto-connects to DeviceLive. From there, manage the fleet via remote container management and batch firmware/application updates; Linux security patches are also delivered via DeviceLive.

こちらもおすすめ