Network interconnection over VPN/SIP
Hello everybody, I hope I'm in the right place here, for my question, if not please ignore or maybe point me to a better place. For start, I'm quit new to the whole GSM stack and still in the process of understanding it, so please if you have the feeling I maybe misunderstood something horrible, I probably have! Feel free to point that out :D. I working together with other peoble on a project witch aims to connect refugee camps in the middle east, to enable easy, safer and non profit communication between them, this is often a big need for example to find relatives, but also to communicate with in the camps because sometimes they are huge etc.. One part (an at the moment my part) of this is to build up independent GSM Networks per camp witch are interconnected with the other camps. The idea is to build GSM networks with in the camp witch are autonomous. If there is power they run, if there is internet peoble can call the other camps (if they have internet). Every thing need's to be done without any central entity (like an HLR for all camps, run on a Server in Frankfurt) so that when, let say government xy decides to only allow nation wide networking an cut's the internet (happened since I'm here 2 times, it's annoying as f**k), the camp's within the country and in the camp it self can still communicate. The question is how to do this interconnection? As fare as I understand in classical GSM infrastructure, this is done by SS7, but since we are not interested (at least by know) to interconnect with "big" providers, and also I did not found an open source implementation (but maybe I missed it) this seems not an option, so the plan now is to do this over VPN's and SIP/VOIP Server (probably asterisk). We like to be able to scale the camps number, and to be relative easy to maintain so we can enable people to run this them self (after we learned to run this our self of course ;)). And these protocols are fare better documented then SS7 so also much more esay to learn and to debug. We are still testing and researching and so I thought it is maybe a good time to ask her if this is a valid way or I missing something, or maybe you can point me towards some documentations where people did some similar things. We are all not professionals and do this more or less with learning by doing, but we don't need to do all the mistakes other did before us :D (maybe some but not all) thx in advance and for this great project witch enabled us to do stuff like this naif
Hi, you can find a related thread started a few days ago which may be of interest to you here: https://lists.osmocom.org/pipermail/openbsc/2018-September/012149.html Regards, Pau -- - Pau Espin Pedrol <pespin@sysmocom.de> http://www.sysmocom.de/ ======================================================================= * sysmocom - systems for mobile communications GmbH * Alt-Moabit 93 * 10559 Berlin, Germany * Sitz / Registered office: Berlin, HRB 134158 B * Geschaeftsfuehrer / Managing Director: Harald Welte
Hi naif, your requirements sound a lot alike those that Rhizomatica are solving in Oaxaca. Running independent GSM networks is the easy part, the hard part is to connect them to each other in a failure-tolerant way. The basic concept is, usually you would have one MSC+HLR for all of your cells, and calls would easily get routed between the different cells. But if the connection to the MSC fails, then all is lost. So you want completely separate network infrastructure for each location. I'm not closely familiar with this aspect, but I imagine using SIP call routing between the otherwise independent networks is one solution (think osmo-sip-connector and an external call router program). The other problem space is that if you want to manage a common subscriber database (HLR), so that e.g. one SIM (phone) can travel freely around and use any of the network cells and always be reachable by the same number wherever it happens to be, then you need some synchronization of the HLR databases. To illustrate, if I want to be reachable by one given phone number in all the cells, something has to know which cell to route that number to. If I show up in a cell, we need to tell the previous cell that I am now over here instead... You would have an easier time if each network is managed separately with its own HLR on-location, in the sense that a SIM card can have a different phone number in each of them. I'm making this up as I go, I hope Rhizomatica folks can flesh out some more details. Note, we're currently working on improving the network setup for Rhizomatica, and you're likely to benefit from that, sooner or later. Either way it would be excellent if you stay in touch and contribute documentation and/or patches back to Osmocom. Also possible would be to book support hours from a professional support vendor, if you have some grant that allows you to, so that your specific needs can be addressed with higher priority than volunteer work can provide (which would also benefit Osmocom at large, to pay for maintenance etc.). None of this is required, of course, but Free Software lives by contribution. All kinds of contribution are welcome. Looking forward to hearing more about your progress! ~N
Hello naif, There is an open source project here: https://github.com/co-cell/ccm that attempts to provide a system for doing the kind of coordination between independent community networks. It is based on Osmocom under the hood, but uses the older osmo-nitb stack (before components were split). The networks are able to operate independently offline, and use a combination of sip call routing as suggested by Neels above, SMPP, and a web api to synchronize usage when connected to the cloud for centralized billing and to interconnect through the cloud for community to community, or community to public voip network communication. The project was originally run by the Telecom Infra Project, but was abandoned by them last year after they moved to a new (unfortunately private) internal tool https://github.com/facebookarchive/CommunityCellularManager . CCM is still actively used by the PCARI Vbts project ( https://www.up.edu.ph/index.php/up-globe-sign-moa-for-village-base-station-p...) with a number of local communities in the Philippines, and they run and maintain their own instance of the CCM cloud. It might be something to look at as a potential starting point, or something to learn from if you want to make different architectural decisions when connecting the networks! Regards, -Matt J. On Wed, Oct 17, 2018 at 6:06 AM Neels Hofmeyr <nhofmeyr@sysmocom.de> wrote:
Hi naif,
your requirements sound a lot alike those that Rhizomatica are solving in Oaxaca.
Running independent GSM networks is the easy part, the hard part is to connect them to each other in a failure-tolerant way.
The basic concept is, usually you would have one MSC+HLR for all of your cells, and calls would easily get routed between the different cells. But if the connection to the MSC fails, then all is lost. So you want completely separate network infrastructure for each location. I'm not closely familiar with this aspect, but I imagine using SIP call routing between the otherwise independent networks is one solution (think osmo-sip-connector and an external call router program). The other problem space is that if you want to manage a common subscriber database (HLR), so that e.g. one SIM (phone) can travel freely around and use any of the network cells and always be reachable by the same number wherever it happens to be, then you need some synchronization of the HLR databases. To illustrate, if I want to be reachable by one given phone number in all the cells, something has to know which cell to route that number to. If I show up in a cell, we need to tell the previous cell that I am now over here instead... You would have an easier time if each network is managed separately with its own HLR on-location, in the sense that a SIM card can have a different phone number in each of them.
I'm making this up as I go, I hope Rhizomatica folks can flesh out some more details.
Note, we're currently working on improving the network setup for Rhizomatica, and you're likely to benefit from that, sooner or later.
Either way it would be excellent if you stay in touch and contribute documentation and/or patches back to Osmocom. Also possible would be to book support hours from a professional support vendor, if you have some grant that allows you to, so that your specific needs can be addressed with higher priority than volunteer work can provide (which would also benefit Osmocom at large, to pay for maintenance etc.). None of this is required, of course, but Free Software lives by contribution. All kinds of contribution are welcome.
Looking forward to hearing more about your progress!
~N
Hi Matthew, On Mon, Oct 22, 2018 at 11:51:32AM -0700, Matthew Johnson wrote:
to public voip network communication. The project was originally run by the Telecom Infra Project, but was abandoned by them last year
It's sad to hear that it has been officially abandoned :( It was more or less apparent to me that priorities had somewhat changed/shifted and that none of the more or less expected steps (like porting it to the post-NITB osmocom stack) have happened.
after they moved to a new (unfortunately private) internal tool
This is the first time I'm hearing about some private internal tool that replaces CCM. Are you able to share more information about this? I've asked the OpenCellular project to clarify this in their forum, see http://ocforum.telecominfraproject.com/t/status-of-communitycellularmanager/... Unfortunately the Forum doesn't [yet] seem to be very active. Regards, Harald -- - Harald Welte <laforge@gnumonks.org> http://laforge.gnumonks.org/ ============================================================================ "Privacy in residential applications is a desirable marketing option." (ETSI EN 300 175-7 Ch. A6)
Some clarifications: On Mon, Oct 22, 2018 at 1:06 PM Harald Welte <laforge@gnumonks.org> wrote:
Hi Matthew,
On Mon, Oct 22, 2018 at 11:51:32AM -0700, Matthew Johnson wrote:
to public voip network communication. The project was originally run by the Telecom Infra Project, but was abandoned by them last year
It's sad to hear that it has been officially abandoned :(
It was more or less apparent to me that priorities had somewhat changed/shifted and that none of the more or less expected steps (like porting it to the post-NITB osmocom stack) have happened.
Couple clarifications, CCM was never part of TIP and was instead an FB project. Abandoned is also maybe the wrong term; internal development at FB ended and development was transferred out of the company. There isn't a lot of activity now though, aside from the ongoing deployments.
after they moved to a new (unfortunately private) internal tool
This is the first time I'm hearing about some private internal tool that replaces CCM. Are you able to share more information about this?
As is the nature of private internal tools, no. It's not really a replacement for CCM though, interests have moved in a different direction (or had when I was still there).
I've asked the OpenCellular project to clarify this in their forum, see
http://ocforum.telecominfraproject.com/t/status-of-communitycellularmanager/...
Unfortunately the Forum doesn't [yet] seem to be very active.
Regards, Harald
-- - Harald Welte <laforge@gnumonks.org> http://laforge.gnumonks.org/
============================================================================ "Privacy in residential applications is a desirable marketing option." (ETSI EN 300 175-7 Ch. A6)
-- Public Key: https://flowcrypt.com/pub/kheimerl@cs.washington.edu
Unfortunately I have no info about new development, only speculation from the site being archived and things heard thirdhand about people getting reassigned with no details. I should be more conservative/specific when putting assertions onto public mailing lists : ) Kurtis is right, that CCM was never officially brought into TIP, and was just a Facebook incubated project too. There are a couple of us using CCM in the community, but we haven’t had the time or resources to do much work on it over the last few months. People are open do doing code review and accepting patches on the co-cell/ccm repo, but facebookincubator/CommunityCellularManager has been officially archived with no plan for future work. -Matt J. On Mon, Oct 22, 2018 at 14:41 Kurtis Heimerl <kheimerl@cs.washington.edu> wrote:
Some clarifications:
On Mon, Oct 22, 2018 at 1:06 PM Harald Welte <laforge@gnumonks.org> wrote:
Hi Matthew,
On Mon, Oct 22, 2018 at 11:51:32AM -0700, Matthew Johnson wrote:
to public voip network communication. The project was originally run by the Telecom Infra Project, but was abandoned by them last year
It's sad to hear that it has been officially abandoned :(
It was more or less apparent to me that priorities had somewhat changed/shifted and that none of the more or less expected steps (like porting it to the post-NITB osmocom stack) have happened.
Couple clarifications, CCM was never part of TIP and was instead an FB project. Abandoned is also maybe the wrong term; internal development at FB ended and development was transferred out of the company. There isn't a lot of activity now though, aside from the ongoing deployments.
after they moved to a new (unfortunately private) internal tool
This is the first time I'm hearing about some private internal tool that replaces CCM. Are you able to share more information about this?
As is the nature of private internal tools, no. It's not really a replacement for CCM though, interests have moved in a different direction (or had when I was still there).
I've asked the OpenCellular project to clarify this in their forum, see
http://ocforum.telecominfraproject.com/t/status-of-communitycellularmanager/...
Unfortunately the Forum doesn't [yet] seem to be very active.
Regards, Harald
-- - Harald Welte <laforge@gnumonks.org> http://laforge.gnumonks.org/
============================================================================ "Privacy in residential applications is a desirable marketing option." (ETSI EN 300 175-7 Ch. A6)
-- Public Key: https://flowcrypt.com/pub/kheimerl@cs.washington.edu
Hi folks, Thanks for your clarifications Kurtis and Matt. The project was not abandoned, but we transitioned the project to community-driven development a few weeks ago. It's quiet now but at least a couple of us have discussed how to continue driving development forward. Adopting the new post-NITB stack is probably highest on the priority list after supporting ongoing deployments (given limited resources, support of existing deployments has meant no new feature work). There is no "internal tool" replacing CCM, not sure where that info came from. Shaddi sent from a phone On Tue, Oct 23, 2018, 11:01 Matthew Johnson <matt9j@cs.washington.edu> wrote:
Unfortunately I have no info about new development, only speculation from the site being archived and things heard thirdhand about people getting reassigned with no details. I should be more conservative/specific when putting assertions onto public mailing lists : )
Kurtis is right, that CCM was never officially brought into TIP, and was just a Facebook incubated project too. There are a couple of us using CCM in the community, but we haven’t had the time or resources to do much work on it over the last few months. People are open do doing code review and accepting patches on the co-cell/ccm repo, but facebookincubator/CommunityCellularManager has been officially archived with no plan for future work. -Matt J. On Mon, Oct 22, 2018 at 14:41 Kurtis Heimerl <kheimerl@cs.washington.edu> wrote:
Some clarifications:
On Mon, Oct 22, 2018 at 1:06 PM Harald Welte <laforge@gnumonks.org> wrote:
Hi Matthew,
On Mon, Oct 22, 2018 at 11:51:32AM -0700, Matthew Johnson wrote:
to public voip network communication. The project was originally run by the Telecom Infra Project, but was abandoned by them last year
It's sad to hear that it has been officially abandoned :(
It was more or less apparent to me that priorities had somewhat changed/shifted and that none of the more or less expected steps (like porting it to the post-NITB osmocom stack) have happened.
Couple clarifications, CCM was never part of TIP and was instead an FB project. Abandoned is also maybe the wrong term; internal development at FB ended and development was transferred out of the company. There isn't a lot of activity now though, aside from the ongoing deployments.
after they moved to a new (unfortunately private) internal tool
This is the first time I'm hearing about some private internal tool that replaces CCM. Are you able to share more information about this?
As is the nature of private internal tools, no. It's not really a replacement for CCM though, interests have moved in a different direction (or had when I was still there).
I've asked the OpenCellular project to clarify this in their forum, see
http://ocforum.telecominfraproject.com/t/status-of-communitycellularmanager/...
Unfortunately the Forum doesn't [yet] seem to be very active.
Regards, Harald
-- - Harald Welte <laforge@gnumonks.org> http://laforge.gnumonks.org/
============================================================================ "Privacy in residential applications is a desirable marketing option." (ETSI EN 300 175-7 Ch. A6)
-- Public Key: https://flowcrypt.com/pub/kheimerl@cs.washington.edu
participants (7)
-
Harald Welte -
Kurtis Heimerl -
Matthew Johnson -
naif8 -
Neels Hofmeyr -
Pau Espin Pedrol -
Shaddi Hasan