Starter-Kit “SmartEnergy”
From an end-2-end solution to a reference Industrial
IoT pipeline
Lorenzo Maiorfi
Naples, November 24th 2017
Any Industrial IoT platform/reference-app should be…
● Reliable
● Reliable
● Reliable (yes, reliable)
● Based on (actual) standards
● Built on a (technology) stack entirely made of open-source layers
● Nice to have, but by far less important, fast, flexible, scalable, cool, etc
Our «stack»
● mbedOS (on STM32 microcontroller here, but plenty of targets are available)
● ZigBee
● MQTT (running on mosquitto broker here)
● NodeJS (with PM2 scheduler), running a custom ZigBee/MQTT «soft gateway»
● InfluxDB (time-series database)
● Grafana (web visualization front-end)
● (…many other adapters, exposing/consuming grpc/http/amqp/raw-tcp/wamp
endpoints…)
Focus on mbedOS 1/2
● Powered by ARM
● 13.090 git commits
● 351 active contributors, 1230 forks
● 450MB source repository
● ~100 targets (starting from €1 MCUs, such as STM32F1, based on Cortex-M3)
● Amazing community web site, with plenty of samples, libraries,
questions+answers
Focus on mbedOS 2/2
● Powerful RTOS
● Based on C++ from ground up
● Amazing tooling (multi-toolchain)
● Testing platform
● Astonishing base services (networking, cryptography, file-system, BLE, etc)
● Great for edge-computing also! (take a look at uTensorFlow sample)
Focus on InfluxDB
● Samples with nanosecond timestamp, with O(1) «insert» complexity
● Each sample is made of fields (numeric values) and tags (filtering, grouping)
● Builtin high-performance time-based math operators (down-sampling)
● Retention Policies (hi-res recent data to lo-res past data…step-by-step)
● Continuous Queries (built-in scheduler runs maintainance data manipulation
tasks)
● Available as-a-service (e.g. aiven.io)
Thanks !
Questions ?

IoT Day 2017 - Starter Kit SmartEnergy

  • 1.
    Starter-Kit “SmartEnergy” From anend-2-end solution to a reference Industrial IoT pipeline Lorenzo Maiorfi Naples, November 24th 2017
  • 3.
    Any Industrial IoTplatform/reference-app should be… ● Reliable ● Reliable ● Reliable (yes, reliable) ● Based on (actual) standards ● Built on a (technology) stack entirely made of open-source layers ● Nice to have, but by far less important, fast, flexible, scalable, cool, etc
  • 4.
    Our «stack» ● mbedOS(on STM32 microcontroller here, but plenty of targets are available) ● ZigBee ● MQTT (running on mosquitto broker here) ● NodeJS (with PM2 scheduler), running a custom ZigBee/MQTT «soft gateway» ● InfluxDB (time-series database) ● Grafana (web visualization front-end) ● (…many other adapters, exposing/consuming grpc/http/amqp/raw-tcp/wamp endpoints…)
  • 5.
    Focus on mbedOS1/2 ● Powered by ARM ● 13.090 git commits ● 351 active contributors, 1230 forks ● 450MB source repository ● ~100 targets (starting from €1 MCUs, such as STM32F1, based on Cortex-M3) ● Amazing community web site, with plenty of samples, libraries, questions+answers
  • 6.
    Focus on mbedOS2/2 ● Powerful RTOS ● Based on C++ from ground up ● Amazing tooling (multi-toolchain) ● Testing platform ● Astonishing base services (networking, cryptography, file-system, BLE, etc) ● Great for edge-computing also! (take a look at uTensorFlow sample)
  • 7.
    Focus on InfluxDB ●Samples with nanosecond timestamp, with O(1) «insert» complexity ● Each sample is made of fields (numeric values) and tags (filtering, grouping) ● Builtin high-performance time-based math operators (down-sampling) ● Retention Policies (hi-res recent data to lo-res past data…step-by-step) ● Continuous Queries (built-in scheduler runs maintainance data manipulation tasks) ● Available as-a-service (e.g. aiven.io)
  • 8.