Page history

From OpenCellID wiki
Revision as of 14:02, 30 March 2016 by Msemm (Talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search
OpenCellID banner history.png



March 25th, 2016

Mozilla Location Service (MLS) and OpenCellID agreed back in 2014 to share their data. MLS at that time was able to implement a regular import of OpenCellID cell locations in short time whereas OpenCellID at that time was busy with rewriting the entire code for the OpenCellID server software including a serious clean-up of the existing cell locations data (see below).
Early 2016 the OpenCellID developers finally started implementing an automatic regular import of the MLS cell location data.
The initial import has been completed and from now on new data arriving on MLS servers is replicated once per hour to the OpenCellID database..

Since MLS currently has approx. 19 Million unique GSM/CDMA cells in its database which OpenCellID does not have - this merge of data is a great benefit for all users of the OpenCellID database.

Thanks to the MLS CC-0 license we can merge the MLS data into the OpenCellID database which makes it seamless for all OpenCellID users to access the database: no changes of any code are required.
Since the MLS team has decided not to publish the measurements which were used for calculating their cell tower locations you will not see any increase of the OpenCellID measurements database whereas a significant growth of the cell locations database is reflected in all related OpenCellID statistics.

We are very happy that this collaboration of two Open Source communities works so smooth and cooperative for the benefit of all being interested in having access to free cell location information. We would like to use this opportunity saying a warm "THANK YOU" to Hanno and the entire MLS team.

January 10th, 2016

There are a few (less than 10) APIkeys which are heavily misusing the OpenCellID API in one or the other way. Last week this raised to a total of 100-160 requests every second, so more than 10 million unauthorized requests per day, and caused a noticeable drop of the performance of the OpenCellID servers.

On top of that we constantly receive ten thousands of requests every day without any APIkey or with a fake APIkey.

Therefore it was decided that these APIkeys and requests are blocked in our load balancers and will not get ANY feedback from the platform any more.

November 13th, 2015

The long awaited data clean-up has been completed!
After taking the responsibility for OpenCellID we first have rewritten the entire software because former Ruby on Rails implementation was poor and could not handle the expected amount of users and data. This was completed in summer 2014.

Then we approached all issues regarding the improvement of the quality of the data. As a first step we implemented a variety of filters to ensure filtering out all bad data from newly arriving measurements. This was completed early 2015.

Then we started to clean up all existing data.
In order to do so we have applied a number of filters to all 1.5 Billion records in the database and removed all records which are obviously wrong (e.g. a cell with German MCC located in Africa).
Due to the complex computations execution of this filtering took several weeks despite the fact that we are using several powerful servers with 128GB RAM, 16 processors etc.
After completing this batch filtering process the first time back in June we found, that we missed to execute one intermediate step in the middle of the process so we had to start over again.
Our 2nd attempt worked as expected so now we have a clean and trustworthy database with measurements.

Then we recreated all computations of the positions of all cell IDs. This has been completed today.

This means for you that in case you are working with the OpenCellID measurements_??.csv.gz files you should download ALL existing files again in order to work with the latest cell tower position tables.

Those of you using the OpenCellID API already enjoy the updated data. Nothing to be done in this regards.

November 19th, 2014

Version 3.5.0 has been launched with the following new features:

Backend:

  • Enhanced geo-location filtering by improving the country boundaries database using OpenStreetMap data.
    This filter checks if the GPS position of a measurement is inside the country defined by MCC or near by.
  • A cell data range filter was fixed in order to accept only 28 bit values of the cell ID for UMTS
  • Enhanced website insights by introducing Alexa's tools
  • Daily statistics for admins have been enhanced in order to improve detection of sources of invalid measurements

UI:
no changes

Wiki:
CLF3 file format documentation has been updated

Next steps: cleanup of the existing data; we made some progress but this task is not yet finished; completion expected in Q1/2015

September 29th, 2014

1 Million unique cells have been reached in Germany:

1-million-cells-in-germany.png

Here you can find the number of cells by network operator:

September 1st, 2014

Version 3.4.0 has been launched with the following new features:

Backend:

  • Export of hourly diff files with cells has been added. Diff files older than 3 days will be deleted. The diff files have the same format as the full export file. Each user can download a diff file max 3x per day. Details about the format of diff files can be found here
  • An additional mechanism has been implemented that limits the API read requests to 1000 read operations per day for all users without a special agreement with OpenCellID.
  • psc/pci values have been added to the cell data. This data is now also added to the file of exported cells and in the affected API requests. Details can be found here
  • The range (radius) of each cell has been added to the cell export CSV file as well as to the affected API requests. Currently the data provided for most of the cells is not valid. This column will be filled with accurate data after the database clean-up which is planned later this year.
  • Duplicate measurements are filtered out. If two consecutive measurements for the same cell have the same coordinates, then second measurement will be filtered out.
  • Collaboration with the Mozilla Location Service project. The format of export cells files is unified for the both projects. The same filters for newly incoming measurements are applied in OpenCellID as in MLS, in particular filters based on the geographical location. Currently the cells data provided by OpenCellID is not fully compatible with MLS' data, because of missing range values. It will be fixed after the database clean-up which is planned later this year. It is planned later this year to import MLS' cells into the OpenCellID database for enhancing the number of known cells. Details about the format of export cells files can be found here
  • Newly incoming measurements are assumed to be suspect if they are more than 150km away from current average cell position of the related cell. Before the DB clean-up, the filter will be enabled only for precise cells, because calculated cells might currently have wrong coordinates. After the DB clean-up, we will apply the filter for all cells.
  • MCC of incoming data is verified now according to geographical location: in case the MCC of a measurement does not match with the country the measurements GPS coordinate is pointing to, then the measurement is treated to be suspect. For now a square around each country is used with a distance to the country border of 40 km to 100 km depending on the latitude.
    It is planned to improve this validation in 2015 by using accurate country border data.

UI:
no changes

Wiki:
API documentation was updated

August 23rd, 2014

OpenCellID last night exceeded the number of 1.000.000.000 (1 Billion) measurements thanks to its great community of contributors!

1 Billion measurements on Aug 23rd 2014.png

August 22nd, 2014

Version 3.3.9.2 has been launched with the following new features:

Backend:
- support for 'device name' field (brand and model of the device)
- support for 'TX power' and 'temperature' fields
- backend support for hours based statistics; not yet supported by UI
- improved error message in case an apiKey is not white listed
- IP address is stored of the server which requested a new OpenCellID user account
- blocking of "new user" requests if a new user has already been created from the same IP within the last 24 hours

UI:
no changes

Wiki:
API documentation was updated

August 1st, 2014

Mozilla Location Service announced its collaboration with OpenCellID here

July 29th, 2014

Version 3.3.9.1 has been launched.

Backend:
Full support for GSM, UMTS, LTE, CDMA network types including
- enhancement of import files
- enhancement of export files
- enhancement of public API
- enhanced filter for incoming measurements

UI:
not yet updated for supporting enhanced data structure

Wiki:
Wiki documentation was updated

July 28th, 2014

Business Geomatics read here (German)

July 20th, 2014

Open Source Software Guide: read here

July 19th, 2014

Liberales Forum: read here

July 18th, 2014

Important German "Linux Magazin" publishes OpenCellID article.

July 16th, 2014

German press release: Android library has been released: Download PDF file

July 7th, 2014

Version 3.3.8 was launched;

  • Italian language added to UI - thank you Lorenzo for the translation!
  • powerful new filter options in the statistics section
  • permalink in upper left map screen corner:
    allows users to create a link with a given zoom level and area which makes it easy to return to this area later; this can, for instance, be used to supervise a select area.
  • detect hex MCC/MNC numbers and convert them to decimal: needed for some Blackberry data
  • method added for checking the online status of OpenCellId: http://opencellid.org/ocid/status
  • enhancement of features that allows users to avoid misuse of the community servers
    Information about white listing can be found here.
    • introduction of white list for accounts that do not contribute data
    • apiKey must be entered for database download
    • APIkeys of new users are blocked automatically if they do not contribute data for a certain period of time
  • "I forgot my apiKey" feature added to main menu in database section
  • API enhancements
    • JSON format option added to various API calls
    • network type added to various API calls
  • OpenCellID test environment http://test-ocid.enaikoon.de/ is now public. The software on this MongoDB cluster is frequently updated, typically showing the latest achieved status. The version number does NOT change with each update. Warning: this is NOT a stable release and may be down or contain corrupted data at any time. It is also slower than the production system due to artificially limited hardware resources that makes it easier to find inefficient code.
  • CSV files with the most up-to-date measurements will always be named 'last_measurements.csv.gz' in the download section from now on
  • various bug fixes

June 23rd, 2014

English press release: Android library has been released: Press release OpenCellID Android library

June 20th, 2014

German press release published for finding additional telematics service providers as contributors: Pressemeldung OpenCellID Telematik-Anbieter

June 2nd, 2014

Version 3.3.5.2 was launched; this is the fastest OpenCellID version to date.

  • all statistics are now calculated in real time, typically in less than 5 seconds
  • more than 20 million records can be received per day; this is approx. 8 times more than before
  • Kafka waiting queues have been activated in case too many new measurements are contributed at the same time
  • the tables for measurements and cell towers were separated into different databases to avoid collisions