{
  "id": 238658,
  "title": "Noisy data filtering examples by the data provider",
  "url": "/competitions/google-smartphone-decimeter-challenge/discussion/238658",
  "author_name": "jeong",
  "post_date": "2021-05-13T01:24:28.074000",
  "votes": 8,
  "comment_count": 0,
  "views": 0,
  "content": "<p>In addition to the filtering rules shared by <a href=\"https://www.kaggle.com/sohier\" target=\"_blank\">@sohier</a> <a href=\"https://www.kaggle.com/c/google-smartphone-decimeter-challenge/discussion/238583\" target=\"_blank\">here</a>, we can find additional filtering rules from the <a href=\"https://www.kaggle.com/google/android-smartphones-high-accuracy-datasets?select=ION+GNSS+2020+Slides+Android+Raw+GNSS+Measurement+Datasets+for+Precise+Positioning.pdf\" target=\"_blank\">slides</a> shared by the data provider (new ones are highlighted):</p>\n<blockquote>\n  <p>Measurements from GNSS chipsets of mobile phones are often noisier and more erroneous. Example of filters your can apply (to exclude) are:</p>\n  <ol>\n  <li>Full bias nanoseconds is zero or invalid</li>\n  <li><code>BiasUncertaintyNanos</code> too large</li>\n  <li>Arrival time is negative or unrealistically large</li>\n  <li><strong>Unknown constellation</strong></li>\n  <li><code>TimeNanos</code> is empty</li>\n  <li><code>State</code> is not <code>STATE_TOW_DECODED</code></li>\n  <li><code>ReceivedSvTimeUncertaintyNanos</code> is high</li>\n  <li><strong><code>AdrState</code> violating this condition: <code>ADR_STATE_VALID == 1 &amp; ADR_STATE_RESET == 0 &amp; ADR_STATE_CYCLE_SLIP == 0</code></strong></li>\n  <li><strong><code>AdrUncertaintyMeters</code> is high</strong></li>\n  </ol>\n</blockquote>\n<p>Hope it helps.</p>",
  "messages": [
    {
      "id": 1304903,
      "postDate": "2021-05-13T01:24:28.073Z",
      "content": "<p>In addition to the filtering rules shared by <a href=\"https://www.kaggle.com/sohier\" target=\"_blank\">@sohier</a> <a href=\"https://www.kaggle.com/c/google-smartphone-decimeter-challenge/discussion/238583\" target=\"_blank\">here</a>, we can find additional filtering rules from the <a href=\"https://www.kaggle.com/google/android-smartphones-high-accuracy-datasets?select=ION+GNSS+2020+Slides+Android+Raw+GNSS+Measurement+Datasets+for+Precise+Positioning.pdf\" target=\"_blank\">slides</a> shared by the data provider (new ones are highlighted):</p>\n<blockquote>\n  <p>Measurements from GNSS chipsets of mobile phones are often noisier and more erroneous. Example of filters your can apply (to exclude) are:</p>\n  <ol>\n  <li>Full bias nanoseconds is zero or invalid</li>\n  <li><code>BiasUncertaintyNanos</code> too large</li>\n  <li>Arrival time is negative or unrealistically large</li>\n  <li><strong>Unknown constellation</strong></li>\n  <li><code>TimeNanos</code> is empty</li>\n  <li><code>State</code> is not <code>STATE_TOW_DECODED</code></li>\n  <li><code>ReceivedSvTimeUncertaintyNanos</code> is high</li>\n  <li><strong><code>AdrState</code> violating this condition: <code>ADR_STATE_VALID == 1 &amp; ADR_STATE_RESET == 0 &amp; ADR_STATE_CYCLE_SLIP == 0</code></strong></li>\n  <li><strong><code>AdrUncertaintyMeters</code> is high</strong></li>\n  </ol>\n</blockquote>\n<p>Hope it helps.</p>",
      "rawMarkdown": "In addition to the filtering rules shared by @sohier [here](https://www.kaggle.com/c/google-smartphone-decimeter-challenge/discussion/238583), we can find additional filtering rules from the [slides](https://www.kaggle.com/google/android-smartphones-high-accuracy-datasets?select=ION+GNSS+2020+Slides+Android+Raw+GNSS+Measurement+Datasets+for+Precise+Positioning.pdf) shared by the data provider (new ones are highlighted):\n\n> Measurements from GNSS chipsets of mobile phones are often noisier and more erroneous. Example of filters your can apply (to exclude) are:\n1. Full bias nanoseconds is zero or invalid\n2. `BiasUncertaintyNanos` too large\n3. Arrival time is negative or unrealistically large\n4. **Unknown constellation**\n5. `TimeNanos` is empty\n6. `State` is not `STATE_TOW_DECODED`\n7. `ReceivedSvTimeUncertaintyNanos` is high\n8. **`AdrState` violating this condition: `ADR_STATE_VALID == 1 & ADR_STATE_RESET == 0 & ADR_STATE_CYCLE_SLIP == 0`**\n9. **`AdrUncertaintyMeters` is high**\n\nHope it helps.",
      "votes": 7
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "1304903": "In addition to the filtering rules shared by @sohier [here](https://www.kaggle.com/c/google-smartphone-decimeter-challenge/discussion/238583), we can find additional filtering rules from the [slides](https://www.kaggle.com/google/android-smartphones-high-accuracy-datasets?select=ION+GNSS+2020+Slides+Android+Raw+GNSS+Measurement+Datasets+for+Precise+Positioning.pdf) shared by the data provider (new ones are highlighted):\n\n> Measurements from GNSS chipsets of mobile phones are often noisier and more erroneous. Example of filters your can apply (to exclude) are:\n1. Full bias nanoseconds is zero or invalid\n2. `BiasUncertaintyNanos` too large\n3. Arrival time is negative or unrealistically large\n4. **Unknown constellation**\n5. `TimeNanos` is empty\n6. `State` is not `STATE_TOW_DECODED`\n7. `ReceivedSvTimeUncertaintyNanos` is high\n8. **`AdrState` violating this condition: `ADR_STATE_VALID == 1 & ADR_STATE_RESET == 0 & ADR_STATE_CYCLE_SLIP == 0`**\n9. **`AdrUncertaintyMeters` is high**\n\nHope it helps."
  }
}