CellSearch is open source software that searches a set of carrier frequencies for LTE basestation signals and returns a list of all cells found. A cell is considered identified if the master information block (MIB) can be successfully decoded (i.e. passes CRC check).
CellSearch -s start_frequency [optional_parameters] Basic options -h --help print help screen -v --verbose increase status messages from program -b --brief reduce status messages from program -i --device-index N specify which attached RTLSDR dongle to use Frequency search options: -s --freq-start fs frequency where cell search should start -e --freq-end fe frequency where cell search should end Dongle LO correction options: -p --ppm ppm crystal remaining PPM error -c --correction c crystal correction factor Capture buffer save/ load options: -r --record save captured data in the files capbuf_XXXX.it -l --load used data in capbuf_XXXX.it files instead of live data -d --data-dir dir directory where capbuf_XXXX.it files are located
If you don't know the center frequency of the LTE cells in your area, you will have to perform a search specifying the first and the last frequency of the band that you want the dongle to examine. See http://niviuk.free.fr/lte_band.php for a list of possible LTE bands.
In the US, the band between 715MHz and 768MHz often contains LTE carriers and the following command will search all possible LTE center frequencies there:
CellSearch --freq-start 715e6 --freq-end 768e6
On a dual core i7-2640 with the default PPM value of 120, this search takes about 6s per center frequency. With a PPM value of 10 (which requires a crystal calibration factor as described below), the search takes about 1s per center frequency.
If you know a frequency where LTE cells are located, perhaps from a previous call to CellSearch, you can specify just that frequency and the program will search for and report all cells detected on that frequency. In my location, in St. Louis, there are two LTE carriers, one for AT&T and one for Verizon. To scan for cells on those two frequencies:
CellSearch --freq-start 739000000
CellSearch --freq-start 751000000
These same frequencies are known to have LTE carriers in San Francisco, Chicago, and Portland.
During program execution, the programs displays a status reports indicating the current center frequency being probed and also prints out messages each time a cell has been detected. After the program finishes executing, it prints a summary of all the carriers where LTE signals were present and also a list of all detected cells. For example:
Detected the following cells:
A: #antenna ports C: CP type ; P: PHICH duration ; PR: PHICH resource type
CID A fc foff RXPWR C nRB P PR CrystalCorrectionFactor
277 2 739M -2.05k -4.62 N 50 N one 0.99999722639394528212
271 2 739M -2.05k -5.21 N 50 N one 0.99999722679001878944
256 2 739M -2.05k -9.01 N 50 N one 0.9999972314719194344
259 2 739M -2.07k -9.91 N 50 N one 0.99999720228692212665
150 2 751M -2.07k -6.07 N 50 N one 0.99999721052901307361
In the above example program output, five cells were detected. An LTE cell with ID 277 was found on carrier frequency 739 MHz. This cell has 2 transmit antennas and the frequency offset between this cell and the dongle was -2.05 kHz. The received power was -4.62 dB and it had a normal cyclic prefix. This cell occupies 50 resource blocks (10MHz), and the PHICH has normal duration and resource type one.
The last column in the program output is the crystal correction factor which provides information about the frequency error of the crystal in the dongle. Suppose this column has the value C. This means that although the crystal in the dongle may be stamped with a label of 28.8MHz, its actual frequency is C*28.8MHz. In calculating this value the basestation is considered as an absolute reference since it has a frequency error of only 0.050 PPM.
The very first time an RTL2832 dongle is powered on, the frequency uncertainty is very high at around 100 PPM. This means that if 1000 dongles are randomly chosen and measured, each of them will have a different frequency error from -100 PPM to +100 PPM. However, for any one dongle, the frequency error does not change significantly from hour to hour or day to day.
Furthermore, the frequency error due to temperature drift does not seem to be very large. For one particular dongle there is a drift of about 4kHz at 739MHz as the dongle warms from room temperature to operating temperature (5 minutes). This is a frequency uncertainty of about 5 PPM.
The --ppm parameter to the program specifies the remaining frequency uncertainty after the frequency correction factor has been applied. The first time that a dongle is used, the default value of 120 should be used until a cell is identified and the program returns a new correction factor. Subsequently, when a correction coefficient is supplied while re-running the program, the PPM value can be reduced, resulting in a much faster cell search.
For example, if the program just exited and reported a correction factor of 1.000052 and the program is re-run later, one can use this correction factor to improve execution speed by reducing the PPM value. If one immediately re-runs the program, it is possible to reduce the PPM value all the way down to zero:
CellSearch --correction 1.000052 --ppm 0 --freq-start 739000000
In general, there will always be some remaining frequency uncertainty and it is not recommended to set the ppm value less than 10:
CellSearch --correction 1.000052 --ppm 10 --freq-start 739000000
If you get the following error message:
usb_open error -3
This means that you need to be 'root' to access the device. Use:
sudo CellSearch XXXXX
See also: LTE Tracker