Hello Osmocom community,

I have a question about terminology that is not really specific to Osmocom, but is specific to GSM/3G/etc cellular industry, or whatever is left of it.

An MSC in a GSM or 3G or any other circuit-switched cellular network has two sides: one side facing the RAN (A interface in GSM, IuCS in 3G), the other side facing PSTN or ISDN or BICC or SIP-I or whatever the fixed telephone network may be. My question: is there a general industry-wide agreement as to which side should be called the front and which should be called the back?

To me it seems intuitive to refer to the A/IuCS side of an MSC as the front and the PSTN/ISDN/BICC/SIP-I side of the same MSC as the back. However, I am just one person, and I have no idea whether or not my view is intuitive to others. And it would even worse if the opposite convention happens to be widely used... Hence I am asking this community, in the hope that someone with more industry experience would know how telco network engineers casually refer to RAN-facing and fixed-network-facing sides of an MSC.

For a long time I had the idea of building ThemWi MSC as a modular system (multiple processes talking to each other via sockets) with a stripped-down OsmoMSC derivative as the "central" component, and with a radically different placement of MSC-MGW. However, as I started working on more detailed design, I realized a major flaw in my originally envisioned architecture: my original approach to MSC-MGW would break TFO for AMR, specifically the dialogue between TFO-aware BTS and the TRAU equivalent in the MGW.

I am now spinning a new architecture in my head, one which tentatively includes two dissimilar MGWs in the MSC. If my front vs back terminology is valid, then the front MGW (talking to AoIP user plane) would perform speech transcoding but become a transparent pass-through for CSD (basically a software-only, IP-based equivalent of a hardware TCSM colocated at the MSC site), while the back MGW would be a transparent pass-through for speech calls, but turn into a complex IWF for CSD. This tentative architecture exactly matches the placement of speech transcoding and CSD IWF functions in traditional hardware MSCs of TDM-based, before-AoIP world. But I need to know if I should keep these front and back terms, or swap them the other way around.

M~