== Building the cellmgr_ng ==
=== Requirements ===
==== OpenBSC ====
The result of "make install" of OpenBSC is required. The cellmgr_ng is using
the osmocomm and sccp library provided by OpenBSC.
==== NexusWare C7 =====
The NexusWare C7 library must be available. It is used to communicate with
MTP up to Level3.
==== NexusWare Uniporte ====
The NexusWare Uniported library must be available. It is used to handle
the configuration of the MGW.
=== Configuring & Building ===
The cellmgr_ng is using autoconf. At first the configure script must be
generated, after this ./configure can be called. The configure script will
use pkg-config to find the CFLAGS and LIBS for the Osmocomm, SCCP, NexusWare
C7 and NexusWare Uniporte libraries.
The NexusWare libraries naturally do not come with pkg-config files. There
are two example files in the pkg-config directory of the cellmgr_ng that can
be changed to point to the proper paths. Alternatively the NEXUSWARE_C7_CFLAGS,
NEXUSWARE_C7_LIBS, NEXUSWARE_UNIPORTE_CFLAGS, NEXUSWARE_UNIPORTE_LIBS environment
variables can be set instead.
$ autoreconf --install --force
$ export PKG_CONFIG_PATH=/openbsc/install/lib/pkg-config:$PWD/pkgconfig
$ . /stuff/NexusWare/SETUP.SH
$ ./configure --host=ppc-linux
== Reset handling ==
=== Loss of the TCP connection to the MSC ===
* All open SCCP connections need to be closed
* All circuits allocated for voice calls need to be closed
* On reconnect the cellmgr needs to generate the reset messages
* The SCCP link is considered to be up
=== Loss of the MTP link/SLTM timeouts ===
* We will have to generate a SCCP reset to the network
* We will have to close ever voice call/mgcp and such
== Filtering ==
=== Filtering reset acks ===
For the above reset handling we filter the reset on both sides. Whenever
we connect to the MSC or lose the BSC we send it a reset.
Whenever the BSC is sending us a reset we directly respond with a reset act
=== Filtering RLSD messages ===
We are using the RLSD from the network and immediately reply with a RLC
if we don't know about this combination of src and dest reference. If we
do have a pair we set a flag and will send a RLC when we receive a RLC from
If we receive a RLSD from the BSC we do have a bug as we didn't respond
within a short enough timeout. We will close the connection. At which point
we will handle the above.
The reason to handle RLSD is that the BSC has a rather short timeout from
cleanup complete to wanting a RLSD and then the MSC and BSC will have some
== Header Rewriting ==
=== POI ===
We do not want to have the point code indicator in the SCCP header. Currently
we are removing this from the CR and DT1 messages.