How the call works
Calls go one way: we start the call. The platform sends your PBX a SIP INVITE where the called number is the contact's number. The PBX handles it like a number dialed from an internal phone and sends it out through your carrier. If the scenario transfers the contact to an employee, the call to the extension arrives over the same trunk.
The PBX has to meet exactly two requirements: accept a trunk without registration, by IP, and send a call from that trunk to an outbound route. The second one is where PBXs trip up when their routing is hard-wired to a specific inbound number.
Connection settings
| Signaling address | pbx.autocall.kz |
| IP for the firewall | 109.233.108.218 |
| Port | 5060 for UDP (PBXs in Kazakhstan), 5061 for TLS (Kazakhstan and CIS)
|
| Voice (RTP) | subnet 109.233.108.208/28, ports 10000–20000 |
| Codec | G.711A |
| Number format | +7XXXXXXXXXX; in the dashboard you can choose 7…, 8… and local numbers without the area code |
| Registration | not needed for a PBX with a static IP; without one, the PBX registers with us |
| Authentication | optional: if your PBX requires it, enter the login and password in the dashboard |
| Call direction | both ways: the platform calls through your PBX, and the PBX sends inbound calls to us |
Firewall
Signaling comes from one address and voice from a subnet: on our side these are different servers. Open both, or you get the classic case of “the call connected, but the line is silent”.
# signaling udp dport 5060 ip saddr 109.233.108.218 accept tcp dport 5061 ip saddr 109.233.108.218 accept # voice udp dport 10000-20000 ip saddr 109.233.108.208/28 accept
What is different about carrier platforms
SI3000 is a carrier platform rather than an office PBX, and it is usually configured by an integrator or the carrier's operations team, not the company admin. So this article is a list of requirements you can hand to whoever runs the node instead of “click here” steps: menu names differ between versions, the meaning stays the same.
On our side, it is a regular SIP trunk without registration. We need no special headers, no special codecs and no REFER support. The call just has to be accepted and passed on.
What to configure
- A SIP interface toward our platform. Accept calls from
109.233.108.218(pbx.autocall.kz) without registration, identified by address. UDP or TLS transport. - Outbound route. A call from this interface must be routed by the dialed number to your outbound direction, the same one your subscribers' regular calls go through.
- Outgoing number override. Our “From” field carries a service label, not a phone number, and it has to be replaced with your number.
- G.711A codec on this direction, without trying to negotiate anything else.
- Voice addresses. On the media gateways and the firewall, open our subnet
109.233.108.208/28on ports 10000–20000: signaling and voice come from different addresses.
CallerID override is required. In our call, the “from” field holds our internal identifier, the call's number in our system, for example 59336790. It makes it easy to match your call log against our export. It consists of digits, but it is not a phone number: your outbound route must replace it with your number, or the carrier will reject the call. The same identifier is duplicated in the X-AutoCall-Call-Id header.
If there is an SBC at the edge
Carrier nodes almost always have a session border controller. If you have one, it makes more sense to terminate the trunk on it: the SBC accepts our call, converts it to the internal format and hands it to SI3000 as a regular internal peer. The platform itself then gets no external addresses, and the list of allowed IPs stays in one place.
You don't need to buy an SBC just for this connection. Without one, the trunk is set up directly.
Encryption (optional)
We support TLS for signaling and SRTP for voice. If your node supports them, select the TLS transport and the matching voice encryption mode in the dashboard.
SRTP works only together with TLS. The voice encryption key travels right inside the SIP message: over unencrypted signaling it would go out in plain text next to the “encrypted” stream. That only looks like protection, so we don't allow this combination. At the same time, TLS without SRTP is a normal option: in Kazakhstan many PBXs and phones have SRTP stripped from the firmware while TLS works.
Add the PBX in the dashboard
Go to “PBX” → “Connect a PBX”. You will need:
- the PBX's IP address and port;
- transport, UDP or TLS, and voice encryption if you enabled it;
- login and password, only if your PBX requires authentication;
- the limit of concurrent channels and the dialing rate: set them no higher than your carrier allows, and we won't exceed them.
Once saved, the PBX status appears in the list: “Connected” means it answers our connection checks. Then, when you create a call campaign, choose your PBX in the “Call through” field: the estimate is calculated right away at the discounted per-second rate, and retry attempts through your own PBX are free.
If calls don't go through
- No reply to our connectivity checks. The signaling port is closed, or our address is not on the allow list.
- The call arrives and gets rejected. There is no route for a number coming from this direction, or the direction has no outbound calling rights.
- The carrier rejects calls. Your number is not set.
- Connects, but silence. The voice subnet is not open on the media gateways.
If a contractor runs your node, hand them the whole list above, that is enough. For our part, we are ready to join a test call and send a SIP trace from our side.
Can a carrier platform like SI3000 be connected to a call campaign service?
Yes. For the call campaign platform, it is a regular SIP trunk without registration: we send the call to your node's address, and the node routes it on through your own channels. There are no special vendor requirements.
What exactly should I ask the integrator for?
Three things: accept SIP calls from our address without registration, route them to the outbound direction by the dialed number, and set your number as the outgoing one. Plus open the voice subnet on the media gateways.
Do I need an SBC in front of SI3000?
If you already have one at the network edge, it makes sense to terminate the trunk on it: SI3000 then sees a familiar internal peer, and all work with the external address stays on the SBC. You don't need to buy an SBC just for this connection.
Which codec should I use?
G.711A. It is the same as alaw, the default in Kazakhstan networks, so there is no transcoding.