README.md

ℱ fft - perform FFT anomaly analysis

Monitors attached microphone using record service and provides FFT functionality as micro-service, processing WAV data and producing JSON and PNG output. This container may be run locally using Docker, pushed to a Docker registry, and published to any Open Horizon exchange.

Status

Service discovery

• org - github@dcmartin.com
• url - com.github.dcmartin.open-horizon.fft
• version - 0.0.1

Service variables

• FFT_ANOMALY_GROUP - group for MQTT topics; default fft
• FFT_ANOMALY_CLIENT - client identifier; defaults to HZN_DEVICE_ID or hostname
• FFT_ANOMALY_TYPE - type of anomaly; default: "motor"
• FFT_ANOMALY_LEVEL - level indicating anomaly; default: 0.2 for motor
• FFT_ANOMALY_MOCK - generate mock anomaly; default: false; options: true, false
• FFT_INCLUDE_RAW - include raw results; default: false; options: true, false
• FFT_INCLUDE_PNG - include image results; default: false; options: true, false
• FFT_INCLUDE_WAV - include audio; default: false; options: true, false
• FFT_PERIOD - interval to poll for new recording; default: 5; range: (0,+)
• MQTT_HOST - default: mqtt
• MQTT_PORT - default: 1883
• MQTT_USERNAME - default: ""
• MQTT_PASSWORD - default: ""
• LOG_LEVEL - specify level of logging; default info; options include (debug and none)
• DEBUG - default: false

Required Services

record

• RECORD_DEVICE - device to record sound; default: default system device
• RECORD_PERIOD - interval to poll audio device; default: 10.0 seconds
• RECORD_SECONDS - amount of time to record; default: 5.0 seconds

mqtt

• MQTT_PERIOD

Description

This service provides a variety of FFT based anomaly detectors to provide data that may be used for temporal analysis of sound. The service makes use of the record service to poll the device's microphone and collect audio. The fft service polls the record service output and when updated, performs the specified anomaly detector (n.b. FFT_ANOMALY_TYPE), updating its output to include both the BASE64 encoded audio, but also the analysis data in both BASE64 encoded Python numpy array as well as a JSON array.

The fft service provides a ReStful API on its designated port and returns a JSON payload; see EXAMPLE below.

Options for anomaly analysis:

• motor - perform Welch power spectral density and compare between readings for variation in top signal and signal strength.
• butter - perform Butterworth order 3 signal filter
• welch - perform Welch power spectral density

How To Use

Copy this repository, change to the fft directory, then use the make command; see below:

% mkdir ~/gitdir
% cd ~/gitdir
% git clone http://github.com/dcmartin/open-horizon
% cd open-horizon/fft
% make
...

The fft value will initially be incomplete until the service completes its initial execution. Subsequent tests should return a completed payload, see below:

% make check

EXAMPLE

{
  "mqtt": {
    "date": 1556054659,
    "pid": 71,
    "version": "mosquitto version 1.4.15",
    "broker": {
      "bytes": {
        "received": 16894449,
        "sent": 2049
      },
      "clients": {
        "connected": 0
      },
      "load": {
        "messages": {
          "sent": {
            "one": 82.43,
            "five": 23.85,
            "fifteen": 8.5
          },
          "received": {
            "one": 105.6,
            "five": 32.46,
            "fifteen": 11.72
          }
        }
      },
      "publish": {
        "messages": {
          "received": 15,
          "sent": 46,
          "dropped": 0
        }
      },
      "subscriptions": {
        "count": 0
      }
    }
  },
  "fft": {
    "date": 1556054668,
    "type": "motor",
    "level": 0.2,
    "id": "davidsimac.local",
    "record": {
      "mock": "mixer_1",
      "date": 1556054666,
      "audio": true
    },
    "motor": {
      "data": [
        6890.625,
        4976.8681640625,
        1205.859375,
        1966.3599853515625,
        "true",
        "true",
        "true",
        "true"
      ],
      "image": true
    }
  },
  "date": 1556054571,
  "hzn": {
    "agreementid": "0001b487674e63737d84fbcdc011eaf365a39477a31f5feb20960040aaad154d",
    "arch": "amd64",
    "cpus": 1,
    "device_id": "davidsimac.local",
    "exchange_url": "http://exchange:3090/v1",
    "host_ips": [
      "127.0.0.1",
      "192.168.1.27",
      "192.168.1.26",
      "9.80.94.82"
    ],
    "organization": "github@dcmartin.com",
    "ram": 1024,
    "pattern": null
  },
  "config": {
    "log_level": "info",
    "debug": true,
    "period": 5,
    "type": "motor",
    "level": 0.2,
    "mock": true,
    "raw": false,
    "wav": true,
    "services": [
      {
        "name": "mqtt",
        "url": "http://mqtt"
      }
    ]
  },
  "service": {
    "label": "fft",
    "version": "0.0.1"
  }
}

SAMPLE

A provided square.wav file, which may also be created using the rootfs/usr/bin/mksqwave.sh script is used as one of the mock data sets.

raw

butter

welch

Changelog & Releases

Releases are based on Semantic Versioning, and use the format of MAJOR.MINOR.PATCH. In a nutshell, the version will be incremented based on the following:

• MAJOR: Incompatible or major changes.
• MINOR: Backwards-compatible new features and enhancements.
• PATCH: Backwards-compatible bugfixes and package updates.

Authors & contributors

David C Martin (github@dcmartin.com)