Skip to main content

Overview

Stream Analyze offers a lightweight streaming analytics and AI platform designed for edge devices including microcontrollers. It is an end-to-end platform enabling efficient development, training, deployment, and orchestration of analytics, computational, and AI models. The major goal of Stream Analyze Platform is to provide an end-to-end solution to edge analytics in areas such as manufacturing, smart products, autonomous vehicles, IIoT (Industrial Internet of Things), and IoV (Internet of Vehicles).

Stream Analyze Platform architecture diagram

What can I use Stream Analyze Platform for?​

Smartness on edge devices

Add smartness to any edge device. Stream Analyze Platform enables intelligence on all kinds of products, such as consumer products, vehicles, industrial machines, heavy-duty equipment, robots and production line devices. This facilitates new levels of insights and automation, such as predictive maintenance, operation mode insights, real-time usage-based cost models. SA Engine is the core component of the Stream Analyze Platform, distinguished by its ultra-lightweight memory requirement. SA Engine runs on a wide range of hardware such as industrial PCs, single board computers (SBCs), microcontrollers (MCUs) and telematic control units (TCUs), and it can coexist with existing software on the device either as a separate process or in a container, or even run bare bone on devices without an OS.

SA Engine includes a library of over 1000 predefined functions for mathematical/statistical computations, data stream filtering and transformation, signal processing, model and data management, and much more. The function library is continuously extended for new customer needs and it is easy to define and deploy new user functions on the fly.

Easy integration with enterprise systems

Built-in support for 3rd party communication protocols and the possibility to install SA Engine on a wide range of devices makes it easy to integrate into existing enterprise solutions.

AI deployment to large fleets of devices

Since SA Engine runs as a software engine on a device AI models and analytics code can be updated interactively on the fly. Deploying an AI model does not require a firmware update and can even be done while the system is running. This facilitates easy deployment of new AI models to large fleets of devices without time-consuming and costly stops in operation or manual interactions with the hardware.

Why Stream Analyze Platform?​

Revolutionary model development process

  • Stream Analyze Platform separates the analytical lifecycle from the software (firmware) lifecycle, thus accelerating and evolving model development validation, reducing a process that usually takes hours/days/weeks/years to minutes.
  • Models can be developed, validated, refined and deployed on the fly.
  • Analysts can experiment with data directly on equipment.
  • Productivity and time to market is vastly improved.

Supporting a broad range of analytical models

  • Supporting a broad range of analytics models suitable for a large variety of different use cases.
  • Including over a thousand predefined mathematical/statistical functions for traditional analytics.
  • Providing ML models for clustering, classification, vector quantization and more, including DBSCAN, DenStream, k-NN, Random Forest and k-means.
  • Reuse existing models, plugin custom models, algorithms and libraries – easily integrated into SA Engine.

Extraction and export of data

  • Easy to extract real time data streams from edge devices.
  • Easy to export both raw and filtered real time data streams to other systems.

Documentation and tutorials

  • Comprehensive reference documentation.
  • Extensive tutorial and example code to use as a starting point for your project.
  • Interactive notebooks to document models.

Supports all stages of edge analytics

Stream Analyze Platform supports all stages of edge analytics. Visualize the data, interact with the data to get an understanding of the data, develop models to predict and take actions on operational behavior, and automate operations based on data stream inputs.

Stages of edge analytics

Features​

  • Lightweight
    • Full version ~5MB
    • Nanocore version ~17kB
  • Cross-platform
    • CPU Arch: X86, AMD, ARM
    • Industrial PC, Raspberry Pi, Microcontrollers (MCUs), Telematic Control Units (TCUs)
  • Data analysis
    • Data extract, transform, load (ETL)
    • Data aggregation, grouping, windowing, multiple sources and combining streams
    • Large library of mathematical, statistical and AI/ML functions
  • Highly extensible
    • Easy integration of any sensor
    • Support extensions with C/C++, Python, Lisp, Java, JS, etc
  • Highly embeddable
    • Exposes call-in APIs for C/C++, Python, Lisp, Java, JS, etc
  • Management
    • Free web-based management GUI and IDE (in cloud or on proprietary servers)
    • VSCode extension (in development)
  • Integration with 3rd party products
    • Highly integrable with 3rd party communications and enterprise systems
    • An end-to-end solution for AIoT, IIoT, and IoV

Understand Stream Analyze Platform​

Learn about Stream Analyze Platform and its fundamental concepts.