SS7 stack recap
SIGTRAN transports the SS7 protocols, so it only makes sense against a quick recap of the SS7 stack. At the bottom, the Message Transfer Part (MTP levels 1–3) provides the physical link, reliable link-layer transfer, and network-level routing between nodes identified by point codes.
Above MTP, SCCP (Signalling Connection Control Part) adds global-title addressing so a message can be routed to a logical element — an HLR, for example — without knowing its point code. TCAP (Transaction Capabilities Application Part) carries transactional dialogues, and MAP (Mobile Application Part) is the mobile application protocol expressing operations like SRI-SM and MT-ForwardSM.
The key point for SIGTRAN is separation of concerns: the application protocols (SCCP, TCAP, MAP) are indifferent to what carries the bytes underneath. SIGTRAN replaces only the transport, leaving those upper layers untouched — which is exactly why an operator can migrate to IP without rewriting its signalling applications.
Why SIGTRAN
Classic SS7 ran over dedicated TDM circuits — 56/64 kbit/s links engineered for a low-bandwidth, tightly controlled signalling network. As core networks became IP end to end, maintaining a parallel TDM signalling plane became expensive, hard to scale, and operationally awkward.
SIGTRAN — defined by the IETF SIGTRAN working group — is the family of protocols that carries SS7 signalling over IP. It preserves the SS7 service model (the same reliability, sequencing, and addressing guarantees applications expect) while replacing narrow TDM links with commodity IP bandwidth and standard data-centre networking.
The payoff is capacity, cost, and convergence: far higher signalling throughput, redundancy using IP-network techniques, and one converged transport for both media and signalling. For a messaging platform, SIGTRAN is the layer at which many direct operator interconnects are implemented today, so signalling security becomes an IP-network discipline as well as a telecom one.
SCTP — the transport beneath SIGTRAN
SIGTRAN does not run over TCP. It uses SCTP (Stream Control Transmission Protocol, RFC 4960), a transport designed for exactly this signalling workload where TCP falls short. SCTP provides reliable, sequenced delivery like TCP but adds features signalling depends on.
Two features matter most. Multi-streaming lets a single SCTP association carry multiple independent message streams, so a lost message in one stream does not head-of-line-block the others — critical when many independent signalling transactions share a link. Multi-homing lets an association bind multiple IP addresses at each endpoint, so if one network path fails, traffic continues on another without tearing down the association.
SCTP is also message-oriented rather than byte-stream-oriented (it preserves message boundaries), and its four-way handshake with a signed cookie resists the SYN-flood style attacks that plague TCP. Together these give SIGTRAN the resilience TDM links provided physically, now achieved in software over IP.
Adaptation layers: M3UA, M2PA, SUA
On top of SCTP sit user-adaptation layers, each replacing a specific SS7 layer so the layers above it are unaware anything changed. Which one you use depends on where in the stack you are bridging.
M3UA (MTP3 User Adaptation, RFC 4666) is the most common: it presents an MTP3-equivalent service to SCCP/ISUP, connecting an IP node (an Application Server) to the SS7 network through a Signalling Gateway. M2PA (MTP2 Peer-to-Peer Adaptation) instead replaces MTP2 to build a true IP signalling link between two MTP3 peers, effectively extending the SS7 link set over IP.
SUA (SCCP User Adaptation) goes one layer higher, carrying SCCP-user traffic (TCAP/MAP) directly over SCTP without an intervening MTP3/SCCP stack — useful for IP endpoints that only need SCCP-user services and global-title routing. Choosing among them is an architecture decision: M3UA and SUA for gateway-connected application servers, M2PA when you are relaying links between signalling points.
Deployment architecture and migration
A SIGTRAN deployment centres on the Signalling Gateway (SG), which terminates TDM-based SS7 on one side and SIGTRAN over IP on the other, bridging legacy peers into the IP world. IP endpoints are Application Servers (AS), each a logical function served by one or more Application Server Processes (ASPs), while a legacy-side counterpart is a Signalling Gateway Process (SGP).
Resilience comes from running ASPs in active/standby or loadshare modes within an AS, combined with SCTP multi-homing across separate IP paths — so a process, a link, or a whole path can fail without dropping signalling. This is how SIGTRAN reproduces, and improves on, the redundancy that TDM link sets provided.
Migration is typically incremental: introduce Signalling Gateways so IP nodes can reach existing TDM peers, move traffic onto M3UA/M2PA associations, and retire TDM links once the IP plane is proven. Because the SCCP/TCAP/MAP applications are unchanged throughout, the migration is a transport swap beneath a stable application layer — the recurring theme of SS7-over-IP.
- SIGTRAN carries the unchanged SS7 upper layers (SCCP/TCAP/MAP) over IP, replacing only the TDM transport.
- It runs over SCTP (RFC 4960), whose multi-streaming and multi-homing give signalling resilience TCP cannot.
- M3UA replaces MTP3 for gateway-connected app servers; M2PA extends MTP2 links; SUA carries SCCP-user traffic.
- Signalling Gateways bridge TDM and IP, letting operators migrate incrementally beneath a stable application layer.