Session Initiation Protocol (SIP) is the signaling protocol used by most IP phone systems, VoIP services and SIP trunks to set up, change and end calls. It is how a phone tells the system it is available, how one endpoint invites another to a call, how the two agree on the audio or video format to use, and how calls are transferred, held and hung up. SIP is an open standard published by the Internet Engineering Task Force (IETF), which is why phones, phone systems and carriers from different vendors can often work together.
At a glance
- SIP handles call signaling; the audio and video travel separately, usually over RTP, or SRTP when encrypted.
- It is used by IP desk phones, softphones, PBXs, UCaaS platforms, contact centers and carrier SIP trunks.
- It is an open standard, but vendors implement it differently, so interoperability still needs testing.
- Firewalls, NAT and SIP ALG settings cause many call quality and one-way audio problems.
- Encryption (TLS for signaling, SRTP for media) is available but depends on the provider and configuration.
What problem it solves
Traditional phone systems used proprietary signaling between handsets and the PBX, and telephone-network signaling between the PBX and the carrier. That tied businesses to one vendor’s phones and to dedicated circuits.
SIP gives voice and video a common, text-based way to set up sessions over any IP network. A phone from one manufacturer can often register with a phone system from another (where both support the needed features), a PBX can connect to a carrier over the internet instead of a PRI, and cloud platforms can connect to each other. It also makes it practical to add features such as forwarding a call to several devices at once or moving a call from a desk phone to a mobile app.
How it works
SIP works by exchanging requests and responses between endpoints (called user agents) and servers:
- REGISTER tells the phone system where a user or device can be reached.
- INVITE starts a call, carrying a description of the media the caller can use (in SDP, the Session Description Protocol), such as codecs and IP addresses.
- Responses such as “180 Ringing” and “200 OK” report progress and acceptance, followed by an acknowledgment.
- BYE ends the call; other methods handle cancellations, transfers and mid-call changes.
Once the call is set up, the audio or video flows directly between the endpoints or media servers using RTP. SIP traffic commonly uses port 5060 for unencrypted signaling and 5061 for TLS, though providers may use other ports.
Between a company’s network and a carrier, a session border controller (SBC) often sits at the edge. It hides internal addresses, enforces security policy, fixes interoperability differences between SIP implementations and can encrypt traffic. In the US and Canada, carriers also use SIP to pass STIR/SHAKEN caller ID authentication between their networks.
Call quality depends on the network rather than SIP itself: bandwidth, latency, jitter and packet loss. Quality of service (QoS) settings and careful firewall configuration matter more for voice than for most other traffic.
When it matters for buyers
- When replacing PRI or analog lines with SIP trunking.
- When choosing phones and devices, since compatibility with your platform (and support from the provider) varies.
- When integrating systems, such as connecting a PBX to Microsoft Teams Direct Routing, a contact center or a paging system.
- When troubleshooting call quality, where firewall, NAT and SBC settings are frequent causes.
- When reviewing security, since internet-exposed SIP systems are a frequent target for toll fraud and call flooding.
Questions to ask vendors
- Which SIP devices, PBXs and SBCs have you tested with your service?
- Do you support TLS and SRTP encryption, and is it on by default?
- What firewall, NAT and SIP ALG settings do you recommend for our sites?
- Which codecs do you support, and how does that affect bandwidth per call?
- What protections against toll fraud and unauthorized registration do you provide?
- How do you monitor and report call quality, and what data can we see when a call goes wrong?
How it differs from SIP trunking
SIP is the protocol; SIP trunking is a service. A SIP trunk is a carrier connection that uses SIP to link your phone system to the public telephone network. A company can use SIP internally (between its phones and PBX) without buying SIP trunks, and a SIP trunk can connect to a compatible SIP-capable phone system, though the PBX, any SBC and the carrier configuration need to be tested or certified together. SIP is also different from WebRTC, the browser technology for real-time audio and video, although many platforms use both and connect them through gateways.
For buying guidance, see our SIP trunking solution page.
Problems this helps with
- Calls that connect but have no audio or one-way audio
- Phones from one vendor that won't work with another vendor's system
- Connecting our phone system to a new carrier
