SIP 503 Service Unavailable in VoLTE — IMS Overload and Network Diagnosis

SIP 503 (“Service Unavailable”) is returned when the responding server is temporarily unable to handle the request. In VoLTE and VoNR deployments, SIP 503 can interrupt both IMS registration and call setup. Its significance differs from most SIP error codes because it is server-side: the failure is in the IMS infrastructure, not in the UE’s SDP offer, codec choice, or authorization.

Diagnosing SIP 503 in VoLTE requires distinguishing between the several distinct root causes — overload, DNS failure, bearer establishment failure, and media function unavailability — which each require a different remediation path.

Technical Reference

Root Cause 1: P-CSCF or S-CSCF Overload

The most straightforward cause: the IMS signaling server is processing more requests than its capacity allows. 3GPP TS 24.229 defines IMS overload control mechanisms that complement RFC 7339 (SIP overload control). When an IMS node is overloaded:

• It MAY include a Retry-After header specifying the suggested wait time
• It SHOULD apply selective overload rejection: prioritize registration maintenance over new call setup
• It MAY include a Warning header with a 399 Misc or a proprietary warning code

In drive test logs, IMS overload during a dense event manifests as clusters of INVITE requests receiving 503 within a short time window, with all 503 responses originating from the same P-CSCF address.

Root Cause 2: IMS APN Bearer Not Established

The P-CSCF address is resolved via DNS over the IMS APN (typically a dedicated EPS bearer with QCI 5 for SIP signaling). If the IMS APN bearer is not established — due to a PDN connection failure, a PCRF rejection of the QCI 5 bearer, or a PGW-side issue — the UE cannot reach the P-CSCF at all.

In this case, the 503 is not returned by the IMS server. Instead, the SIP REGISTER or INVITE timer expires (Timer F: 32 seconds by default for non-INVITE, Timer B: 32 seconds for INVITE) and the UE generates a local 503 internally. This is visible in the NAS log as a failed EPS bearer setup event correlated with the IMS registration timeout.

How to distinguish: Check whether the IMS APN EPS bearer (QCI 5) is present in the EPS bearer context before the 503. If the QCI 5 bearer is absent or in a failed state, the 503 is a transport-layer failure, not an IMS server response.

Root Cause 3: DNS Resolution Failure for the P-CSCF

The P-CSCF IP address is typically obtained via PCO (Protocol Configuration Options) in the Attach Accept, or via DNS. A DNS resolution failure (NXDOMAIN, timeout, or unreachable DNS server) prevents the UE from routing the SIP REGISTER request to any P-CSCF. The SIP transaction times out and produces a locally generated 503.

Identifying DNS failure in the log: The DIAG log includes DNS query/response packets. A DNS request for the P-CSCF FQDN with no response or an NXDOMAIN answer immediately preceding the IMS failure confirms DNS as the root cause.

Root Cause 4: MRFP Unavailability

For calls requiring media anchoring (conference calls, calls with media recording, IMS emergency calls), an MRFP (Media Resource Function Processor) is allocated. If the MRFP is unavailable or over capacity, the S-CSCF returns 503 for INVITE requests that require media resources. Regular point-to-point VoLTE calls typically do not require MRFP and are not affected.

Step-by-Step Field Diagnosis

Step 1 — Identify the source of the 503. Decode the Via header stack in the 503 response. A single Via entry from the P-CSCF address confirms the P-CSCF generated the 503. Multiple Via entries indicate the request was forwarded further into the IMS core before the error.

Step 2 — Check for a Retry-After header. Its presence confirms the server is aware of its overload condition and provides a retry interval. Its absence suggests either a transient failure or that the server implementation does not include Retry-After.

Step 3 — Verify the IMS APN bearer status. In the NAS/EPS bearer log, confirm a QCI 5 dedicated EPS bearer is active at the time of the 503. A missing or failed QCI 5 bearer is the primary indicator of a transport-layer failure rather than an IMS server response.

Step 4 — Check DNS resolution for the P-CSCF. In the DIAG log, locate the DNS query for the P-CSCF FQDN (typically pcscf.ims.mnc{MNC}.mcc{MCC}.pub.3gppnetwork.org). A successful DNS response with a valid IP address confirms the transport layer is functional and the 503 comes from the IMS server itself.

Step 5 — Correlate with the time of day and network load. IMS overload 503 responses are correlated with peak usage periods. If 503 appears consistently during specific time windows (morning rush, concert peaks), the cause is IMS signaling capacity, not device or SIM configuration.

SIP Retransmission After 503

RFC 3261 defines timer-based retransmission for SIP transactions. For an INVITE transaction (Timer B = 64 × T1 = 32 seconds by default), the UE retransmits the INVITE at exponentially increasing intervals until Timer B expires. After a 503 with Retry-After, the UE should not retransmit until the Retry-After interval expires.

Excessive 503-triggered retransmissions contribute to IMS signaling load — a feedback loop where overload causes 503, 503 causes retransmissions, retransmissions increase load. 3GPP TS 24.229 §5.1.6A specifies exponential back-off mechanisms to break this loop.

Capturing SIP 503 in the Field

IMS OTA signaling is captured via DIAG log. The decoded SIP REGISTER or INVITE transaction shows the full request and response including Via headers, Retry-After, and Warning. Correlating the 503 timestamp with the QCI 5 bearer status (from the EPS bearer log) and the DNS query result (from the data protocol log) provides a complete root cause picture from a single DIAG trace.

Related SIP Codes

• SIP 480 — Temporarily Unavailable (callee-side unavailability, not server overload)
• SIP 500 — Server Internal Error (server-generated error, not temporary overload)
• SIP 504 — Server Time-out (upstream server did not respond within the timeout)
• SIP 403 — Forbidden (authorization failure, not service availability)
• SIP 488 — Not Acceptable Here (SDP offer rejection, not service availability)