as simple as docker compose up

common feautures:

• latest binary releases from osmocom repo
• configuration with single file ./configs/config.yml (for most common parameters)
• automatic interaction with newly connected users
• egprs with routing
• active users monitoring and other useful scripts (see ./helpers)

Normal setup, egprs network isolated inside osmo container

Security assesment for gprs-devices

• network_mode=host
• insisibly proxy device's requests to your http proxy

Test branch to support antsdr devices with ethernet connection

Poc for dual bts setup

In ./config/config.yml set values in radio section:

...
radio:
  ...
  device-type: lime
  tx-path: BAND2
  rx-path: LNAW
  ...

BAND2 and LNAW do work on LimeSDR USB and LimeSDR Mini

In ./configs/config.yml set values in radio section:

...
radio:
  ...
  device-type: uhd
  tx-path: TX/RX
  rx-path: TX/RX
  clock-ref: external
  ...

TX/RX and TX/RX should work, but need testing. You can also set external/internal/gpsdo clock-ref for usrp devices:

• external - ext 10 mhz ref clock, such as leo godnar/octoclock gpsdo
• internal - default usrp's clock
• gpsdo - internal gpsdo that mounted in usrp board (if any)

Feel free to check offisial osmo-trx-uhd documentation, 1.10.4 clock-ref: https://ftp.osmocom.org/docs/osmo-trx/master/osmotrx-uhd-vty-reference.pdf

Place the custom firmware in ./configs/uhd_images/ and give it an appropriate name, such as usrp_b210_fpga.bin. It will be automatically placed in the /usr/share/uhd/images/ folder inside the osmocom container. It's useful for devices such as USRP B210 LibreSDR clones.

In ./configs/config.yml set values in radio section:

...
radio:
  ...
  device-type: uhd
  tx-path: TX/RX
  rx-path: TX/RX
  clock-ref: external
  ...

TX/RX and TX/RX do work on LibreSDR B220 mini (XC7A100T+AD9361). You can also set external/internal/gpsdo clock-ref for such devices:

• external - ext 10 mhz ref clock. Tested with leo bodnar gpsdo
• internal - default onboard clock, in my testbed it was too poor to get gsm work properly

This is not fully tested, but seems that network starts successfully, see godfuzz3r#1

Switch to antsdr_e200 branch and rebuild containers with docker compose build.

In this branch uhd driver and osmo-trx compiled from source to get it work with antsdr hardware.

DNS confirugration for default apn is in ./configs/dnsmasq/apn0.conf

You can check who is currently using gprs with ./helpers/show-pdp.sh or ./helpers/mon.sh

You can also check the traffic on the apn0 interface with ./helpers/wireshark.sh. This script will forward tcpdump output from container to wireshark.

By default android/ios devices send large amount of traffic, which results in network degradation. You can work around this behavior by responding to any dns request with localhost. Just uncomment rules in ./configs/dnsmasq/apn0.conf:

address=/#/127.0.0.1
address=/#/::1

Below are examples of ping timings with default DNS settings and DNS that responds with localhost to everything.

Default dns configuration, all mobile traffic goes through egprs:

Dns config resolves everything to localhost:

You can also enable/disable routing mobile traffic to the internet with routing-enabled: true/false in ./configs/config.yml file:

...
egprs:
  routing-enabled: true
  ...

This option will or will not setup iptables routing inside docker container.

Here is bunch of scripts to analyze UE's behaviour or to interact with the network.

wireshark.sh - sniff on apn0 interface and show it in wireshark

send-*.sh - do manual interaction

show-pdp.sh - show current egprs usage

show-subscribers-msc.sh - show subscribers on msc

show-subscribers-hlr.sh - show subscribers on hlr

mon.sh - show currently online subscribers in table

• if MS leaves the network without detaching IMSI, it will be "active" for up to 8 minutes. This is because the minimum value for T3212 timer (location update) is 6 minutes, and MSC requires some time gap between T3212 value in BSC and MSC. Without gap, MSC may expire active subscribers.
• IPv6 routing not configured
• sometimes devices may be not properly "passed" to container. Stop container, reconnect device and start container again may help

• probably network/firewall issues, restart building process may help

• sometimes devices may be not properly "passed" to container. Stop container, reconnect device and start container again may help. (I actually don't know how to solve this problem. If now how to solve this - don't hesitate to create pull requests!)

• silent call/paging interaction
• setup default values if some fields in ./configs/config.yml does not exists (for now deleting field from config.yml will cause crash)