VoIP Terminology

This glossary covers the core terminology used in VoIP testing, SIP signaling, media transport, and quality measurement. Each term includes a concise definition and context for how it applies within CallMeter.

Signaling Protocols

SIP (Session Initiation Protocol)

The signaling protocol used to establish, modify, and terminate VoIP calls. Defined in RFC 3261, SIP handles call setup (INVITE), registration (REGISTER), call termination (BYE), and other signaling operations. SIP does not carry audio or video data. CallMeter uses SIP as the primary signaling protocol for all test and probe operations. See SIP Response Codes for a full reference.

SDP (Session Description Protocol)

A text-based format for describing media session parameters. Defined in RFC 4566, SDP is carried inside SIP messages (typically in the body of INVITE and 200 OK) and specifies codecs, IP addresses, ports, and other media attributes. During call setup, the SDP offer/answer exchange (RFC 3264) determines which codecs and transport parameters will be used for the media session. CallMeter generates SDP offers based on your test configuration and records the negotiated parameters in the SIP message trace.

SRTP (Secure RTP)

An encrypted variant of RTP defined in RFC 3711. SRTP provides confidentiality, message authentication, and replay protection for media streams. SRTP uses symmetric encryption (typically AES-128) with keys exchanged via SDES (SDP Security Descriptions) or DTLS-SRTP handshakes. CallMeter's roadmap includes SRTP support via both SDES and DTLS-SRTP key exchange methods.

DTLS (Datagram Transport Layer Security)

A TLS-based protocol adapted for datagram transport (UDP). Defined in RFC 6347, DTLS provides encryption for UDP-based protocols. In VoIP, DTLS-SRTP (RFC 5764) uses a DTLS handshake to establish encryption keys for SRTP media streams. This is the preferred key exchange method for WebRTC and modern SIP deployments.

SIP Dialog

A peer-to-peer SIP relationship between two user agents that persists for the duration of a call. Defined in RFC 3261, a dialog is identified by a Call-ID, a local tag, and a remote tag. A dialog is established when the initial INVITE receives a response with a To tag. All subsequent in-dialog requests (re-INVITE, BYE, INFO, UPDATE) share the same dialog identifiers. Understanding dialogs is important when analyzing SIP traces in CallMeter -- all messages within a single call share the same dialog.

SIP Transaction

A request-response exchange between a SIP client and server. Each transaction consists of a request, zero or more provisional responses, and exactly one final response. Transactions are identified by the branch parameter in the Via header. SIP defines two transaction types: INVITE transactions (which include the ACK for non-2xx responses) and non-INVITE transactions (REGISTER, BYE, OPTIONS, etc.). In CallMeter SIP traces, each request-response pair forms a distinct transaction within the broader call dialog.

Early Media

Audio or video that flows before a SIP call is formally answered (before the 200 OK). Early media is signaled via 183 Session Progress with an SDP body and is commonly used for ringback tones, IVR prompts, or announcements from the network. When analyzing CallMeter results, early media affects the time-to-first-media metric because RTP begins flowing during the RINGING phase rather than after the call is established. Early media is negotiated per RFC 3960.

SDES (SDP Security Descriptions)

A key exchange method for SRTP defined in RFC 4568 where encryption keys are carried inline in the SDP as a=crypto: attributes. SDES is simpler than DTLS-SRTP but has the disadvantage that keys are visible to any SIP intermediary that can read the SDP body. SDES is commonly used in enterprise SIP deployments where the signaling path is trusted.

Transport Protocols

RTP (Real-time Transport Protocol)

The protocol that carries audio and video media in VoIP calls. Defined in RFC 3550, RTP provides timing information, sequence numbers, and payload type identification for media packets. Each RTP packet contains a timestamp for playout synchronization, a sequence number for ordering and loss detection, an SSRC identifier, and the media payload. CallMeter measures dozens of RTP-level metrics per endpoint per second. See Metrics Reference.

RTCP (RTP Control Protocol)

A companion protocol to RTP that carries quality statistics and control messages. Defined in RFC 3550 alongside RTP, RTCP reports include Sender Reports (SR), Receiver Reports (RR), and various feedback messages. RTCP provides the raw data for calculating metrics like jitter, round-trip time, and packet loss. CallMeter collects RTCP statistics from both directions (send and receive) using dual metric collectors per endpoint.

AVPF (Audio-Visual Profile with Feedback)

An RTP profile defined in RFC 4585 that adds immediate feedback capabilities to the standard Audio-Visual Profile (AVP). AVPF enables feedback messages like NACK (request retransmission), PLI (Picture Loss Indication), and FIR (Full Intra Request) to be sent immediately rather than waiting for the next scheduled RTCP interval. Essential for video calls where timely error recovery is critical. CallMeter supports AVP, AVPF, SAVP, and SAVPF RTP profiles.

CCM (Codec Control Messages)

RTCP feedback messages defined in RFC 5104 for controlling codec behavior. Includes FIR (Full Intra Request) for requesting keyframes and TMMBR/TMMBN (Temporary Maximum Media Stream Bit Rate Request/Notification) for bandwidth control. Used alongside AVPF for video stream management.

NAT Traversal

NAT (Network Address Translation)

A networking technique that maps private IP addresses to public addresses. NAT is the primary obstacle to peer-to-peer media connectivity in VoIP because the private addresses in SDP offers are not reachable from the public internet. SIP-aware NAT traversal mechanisms (STUN, TURN, ICE) solve this problem. When testing through NAT, CallMeter workers must be able to send and receive RTP on the negotiated ports.

ICE (Interactive Connectivity Establishment)

A framework defined in RFC 5245 for finding the best network path between two endpoints. ICE gathers multiple candidate addresses (local, server-reflexive via STUN, and relayed via TURN), tests connectivity between all candidate pairs, and selects the optimal path. ICE is mandatory in WebRTC and increasingly common in enterprise SIP deployments. CallMeter's roadmap includes ICE support for testing NAT traversal scenarios.

STUN (Session Traversal Utilities for NAT)

A protocol defined in RFC 5389 that allows a client behind NAT to discover its public IP address and port. STUN servers are lightweight reflectors that respond with the source address they see. The discovered address is used as a candidate in ICE or inserted into SDP for NAT traversal. STUN works for most NAT types but fails with symmetric NAT.

TURN (Traversal Using Relays around NAT)

A relay protocol defined in RFC 5766 that provides media relay for cases where direct connectivity fails (e.g., symmetric NAT, restrictive firewalls). TURN allocates a relay address on a public server and forwards all media through it. TURN always works but adds latency and consumes server bandwidth. It is the fallback mechanism in ICE when STUN fails.

RTP Identifiers

SSRC (Synchronization Source)

A 32-bit identifier carried in every RTP packet header that uniquely identifies a media stream within an RTP session. Each sender in a session has a unique SSRC. RTCP reports reference SSRCs to associate quality statistics with specific streams. CallMeter uses SSRC ownership to correctly attribute send-side and receive-side quality metrics to the right endpoint.

CSRC (Contributing Source)

A 32-bit identifier in the RTP header that identifies sources contributing to a mixed audio stream. When an audio mixer combines multiple streams, the CSRC list in the output identifies the original senders. Relevant when testing conferencing systems that use server-side audio mixing.

RTP Multiplexing

The technique of sending multiple media streams (e.g., audio and video) over a single transport address (IP:port pair) rather than using separate ports for each stream. Defined in RFC 8858, RTP multiplexing reduces the number of ports needed and simplifies NAT traversal. Multiplexed streams are distinguished by their SSRC identifiers. In SDP, multiplexing is negotiated using the a=rtcp-mux attribute, which indicates that RTP and RTCP share the same port. CallMeter supports rtcp-mux for both audio and video streams.

Payload Type

A 7-bit field in the RTP header that identifies the codec used for the media payload. Static payload types (0-95) are assigned by IANA for common codecs: PCMU is 0, PCMA is 8, G.722 is 9. Dynamic payload types (96-127) are negotiated via SDP for codecs like Opus, H.264, VP8, and VP9. The a=rtpmap: attribute in SDP maps dynamic payload types to codec names and clock rates.

Quality Metrics

MOS (Mean Opinion Score)

A numerical measure of voice call quality on a scale of 1 (bad) to 5 (excellent). Originally defined by ITU-T P.800 as a subjective score from human listeners, MOS is now commonly estimated algorithmically from network parameters using the E-model (ITU-T G.107). CallMeter calculates estimated MOS from jitter, packet loss, RTT, and codec characteristics. A MOS above 4.0 indicates good quality. Below 3.5 typically indicates noticeable degradation. See MOS Score.

R-Factor (Transmission Rating Factor)

A quality metric from the ITU-T G.107 E-model, ranging from 0 to 100. The R-Factor considers codec impairment, packet loss, jitter, delay, and equipment characteristics to produce a single quality number. R-Factor maps directly to MOS: R above 80 corresponds to MOS above 4.0 (good), R between 70-80 is acceptable, and R below 60 indicates poor quality. See R-Factor.

PDD (Post-Dial Delay)

The time between sending the SIP INVITE and receiving the first provisional response (180 Ringing or 183 Session Progress). PDD measures how long it takes the SIP infrastructure to route the call and alert the callee. Industry benchmarks consider PDD under 3 seconds acceptable and under 1 second excellent. CallMeter measures PDD as part of its call timing metrics.

ASR (Answer Seizure Ratio)

The percentage of call attempts that result in a successful answer (200 OK). Calculated as (answered calls / total attempts) * 100. ASR is a key performance indicator for SIP infrastructure health. Enterprise targets typically range from 40% to 60% depending on the use case. CallMeter calculates ASR per test run and per probe execution.

NER (Network Effectiveness Ratio)

The percentage of call attempts where the network successfully delivered the call, regardless of whether the callee answered. NER includes answered calls plus timeouts and user-initiated rejections (which are not network faults). A high NER with low ASR means the network is healthy but callees are unavailable. NER above 98% indicates excellent network performance.

Audio Concepts

Codec (Coder-Decoder)

An algorithm that encodes audio or video into a compressed digital format for transmission and decodes it at the receiving end. Different codecs offer trade-offs between compression efficiency, quality, latency, and computational cost. CallMeter supports PCMA, PCMU, G.722, and Opus for audio, and H.264 and VP8 for video. See Supported Codecs.

PCM (Pulse Code Modulation)

The fundamental digital audio encoding method. PCM samples the audio waveform at regular intervals and quantizes each sample to a discrete value. G.711 (PCMA and PCMU) uses 8-bit PCM at 8000 samples per second, producing 64 kbps uncompressed audio. PCM is the baseline for all other audio codecs.

G.711

The ITU-T standard for narrowband audio coding. Two variants exist: A-law (PCMA, used in Europe and international networks) and u-law (PCMU, used in North America and Japan). Both operate at 8 kHz sampling rate with 64 kbps bitrate. G.711 is universally supported by SIP endpoints and is the safest codec choice for compatibility testing.

G.722

An ITU-T wideband audio codec operating at 16 kHz sampling rate. Despite its 64 kbps bitrate (same as G.711), G.722 provides significantly better audio quality because it captures frequencies up to 7 kHz compared to G.711's 3.4 kHz. Often marketed as "HD Voice" by VoIP providers. Widely supported in enterprise IP phones.

Opus

A modern, royalty-free audio codec defined in RFC 6716. Opus is uniquely versatile: it supports sampling rates from 8 kHz to 48 kHz, bitrates from 6 kbps to 510 kbps, and can switch between narrowband speech and full-bandwidth music encoding. Opus is the mandatory audio codec for WebRTC and offers the best quality-to-bitrate ratio of any audio codec. Adaptive bitrate and forward error correction are built in.

Ptime (Packetization Time)

The duration of audio contained in each RTP packet, expressed in milliseconds. Common values are 20ms (default for most codecs), 10ms (lower latency), and 30ms or 40ms (higher efficiency). Ptime affects latency, packet rate, and loss impact: longer ptime means fewer packets per second but each lost packet removes more audio.

Jitter Buffer

A buffer at the receiving end that absorbs variations in packet arrival timing (jitter) to produce smooth, continuous audio playout. The jitter buffer introduces deliberate delay to reorder packets and fill timing gaps. Adaptive jitter buffers dynamically adjust their size based on observed jitter. CallMeter reports jitter buffer metrics including buffer size, target delay, and late/lost packet counts.

PLC (Packet Loss Concealment)

A technique used by audio decoders to mask the effect of missing packets. PLC algorithms generate replacement audio based on the characteristics of recently received audio, producing a less noticeable gap than silence. Quality varies by codec: Opus has excellent PLC, while G.711 PLC is basic. CallMeter measures PLC events per second as an audio quality indicator.

DTX (Discontinuous Transmission)

A technique that reduces transmission during silence periods. When the speaker is not talking, DTX stops sending media packets (or sends only comfort noise packets), saving bandwidth. DTX affects packet-level metrics because the apparent packet rate drops during silence. Some jitter buffer implementations may interpret silence as packet loss.

VAD (Voice Activity Detection)

An algorithm that detects whether a segment of audio contains speech or silence. VAD is used in conjunction with DTX to determine when to suppress transmission. It is also used for audio level metering and silence detection in conferencing systems.

Comfort Noise

Artificially generated background noise transmitted during silence periods (when DTX suppresses speech packets). Comfort noise prevents the jarring effect of complete silence that would make the listener think the call has dropped. Defined in RFC 3389 as a separate RTP payload type (CN, payload type 13).

Transcoding

The process of converting media from one codec to another in real time. Transcoding occurs when two endpoints in a call use different codecs (e.g., one side uses Opus and the other uses G.711). An SBC or media gateway decodes the incoming codec and re-encodes it in the outgoing codec. Transcoding consumes significant CPU and introduces additional latency and quality loss. CallMeter can test transcoding paths by configuring different codec sets on caller and callee groups.

Offer/Answer Model

The SDP exchange mechanism defined in RFC 3264 that negotiates media session parameters. The offerer (typically the caller) sends an SDP body listing all codecs, IP addresses, and ports it supports. The answerer (callee) responds with the subset it accepts. This single exchange determines the codecs, media addresses, and RTP profiles for the entire call. Re-INVITEs can renegotiate parameters mid-call. CallMeter's SIP traces show the full offer/answer exchange for diagnostic analysis.

Video Concepts

Keyframe (I-Frame)

A complete video frame that can be decoded independently without reference to any other frame. Keyframes are larger than predicted frames but are essential for starting or recovering video playback. When packet loss corrupts predicted frames, the receiver requests a new keyframe via PLI or FIR feedback messages. CallMeter tracks keyframe request rates as a video quality indicator.

GOP (Group of Pictures)

The sequence of video frames between two consecutive keyframes. A GOP typically starts with a keyframe followed by predicted frames (P-frames) and optionally bidirectionally predicted frames (B-frames). Longer GOPs provide better compression but increase recovery time after packet loss because the receiver must wait for the next keyframe.

NACK (Negative Acknowledgement)

An RTCP feedback message defined in RFC 4585 requesting retransmission of a lost RTP packet. NACK enables error recovery without waiting for the next keyframe. It works well on low-latency networks but is less effective when RTT is high because the retransmitted packet may arrive too late for playout. CallMeter measures NACK counts as a feedback metric.

PLI (Picture Loss Indication)

An RTCP feedback message defined in RFC 4585 indicating that the receiver has lost a video picture and cannot decode subsequent frames. PLI requests the sender to generate a new keyframe. Unlike NACK, PLI does not specify which packet was lost. It is a coarser recovery mechanism used when the decoder state is corrupted. CallMeter reports PLI counts per endpoint.

FIR (Full Intra Request)

An RTCP feedback message defined in RFC 5104 requesting the sender to generate a keyframe. FIR is functionally similar to PLI but is defined as a codec control message rather than a transport-layer feedback message. Both serve the same practical purpose in most implementations.

RTCP Feedback

SLI (Slice Loss Indication)

An RTCP feedback message indicating the loss of a specific rectangular region (slice) of a video frame. More granular than PLI, SLI allows the sender to refresh only the affected region. Supported by some video codecs and encoders.

REMB (Receiver Estimated Maximum Bitrate)

An RTCP feedback message where the receiver estimates the maximum bitrate it can accept based on observed packet loss and available bandwidth. The sender should reduce its encoding bitrate to match. REMB is widely used in WebRTC for congestion control.

SIP Infrastructure

Registrar

A SIP server that accepts REGISTER requests and maintains a binding between a SIP URI (Address of Record) and the contact address where the user can currently be reached. Registration is the first step in any SIP call flow. In CallMeter, you configure registrars with their domain, transport protocol, and SIP account credentials. See Adding a Registrar.

SIP Proxy

An intermediary SIP server that routes requests between endpoints. Proxies make routing decisions based on the Request-URI, Route headers, and local policy. They do not typically modify the media path. Stateful proxies maintain transaction state and can fork requests to multiple destinations.

PBX (Private Branch Exchange)

An enterprise telephone system that manages internal extensions and connects them to the public telephone network. Modern IP-PBXs use SIP for internal signaling. CallMeter is commonly used to stress-test PBX systems by simulating hundreds or thousands of concurrent SIP endpoints registering and placing calls.

SBC (Session Border Controller)

A network device that controls SIP signaling and media at network boundaries. SBCs provide security (topology hiding, access control), interoperability (protocol normalization, codec transcoding), and quality of service (bandwidth management, traffic shaping). SBCs are a common target for CallMeter stress tests because they are often the bottleneck in enterprise VoIP deployments.

B2BUA (Back-to-Back User Agent)

A SIP entity that acts as an endpoint on both sides of a call, terminating the incoming SIP dialog and creating a new outgoing dialog. Unlike proxies, B2BUAs have full control over signaling and can modify headers, SDP, and routing. SBCs and many PBX systems operate as B2BUAs.

User Agent

A SIP endpoint that initiates or receives SIP requests. A user agent can be a softphone, an IP phone, a conference server, or a CallMeter worker simulating a SIP endpoint. The SIP specification defines User Agent Client (UAC) for initiating requests and User Agent Server (UAS) for responding.

UCaaS (Unified Communications as a Service)

A cloud-delivered unified communications platform that combines voice, video, messaging, and collaboration. UCaaS providers use SIP infrastructure at scale and are primary users of platforms like CallMeter for quality assurance, capacity testing, and continuous monitoring.

Endpoint

In the context of CallMeter, an endpoint is a single simulated SIP user agent that registers with a registrar, places or receives a call, exchanges media, and reports metrics. Each endpoint in a test run operates independently with its own SIP credentials, registration state, and media session.

Network Quality

QoS (Quality of Service)

Network mechanisms that prioritize certain types of traffic. In VoIP, QoS ensures voice and video packets are handled with lower latency and loss than best-effort traffic. QoS is implemented through packet marking (DSCP/TOS), traffic shaping, and queue management on routers and switches.

TOS (Type of Service)

An 8-bit field in the IP header used for packet priority classification. The TOS field has been redefined as the Differentiated Services (DS) field in modern networks. The first 6 bits form the DSCP value and the remaining 2 bits are used for Explicit Congestion Notification (ECN).

DSCP (Differentiated Services Code Point)

A 6-bit value in the IP header that classifies packets for QoS treatment. VoIP media typically uses DSCP EF (Expedited Forwarding, value 46) for voice and AF41 (Assured Forwarding class 4, value 34) for video. SIP signaling uses CS3 (Class Selector 3, value 24). Proper DSCP marking ensures VoIP traffic receives priority handling across the network.

DiffServ (Differentiated Services)

A network architecture defined in RFC 2474 that provides scalable QoS by classifying and marking packets at network boundaries and applying per-hop behaviors (PHBs) at each router. DiffServ replaced the original TOS field interpretation with the DSCP framework. Three standard PHBs exist: Expedited Forwarding (EF) for low-latency traffic like voice, Assured Forwarding (AF) for important but less delay-sensitive traffic like video, and Best Effort (BE) for everything else. When testing VoIP infrastructure, verifying that DSCP markings are preserved end-to-end is critical -- some network segments may strip or remark DSCP values, degrading voice quality.

Jitter

The variation in packet arrival times. In a perfect network, RTP packets arrive at perfectly regular intervals matching the packetization time. Jitter measures how much actual arrival times deviate from this ideal. High jitter forces the receiver's jitter buffer to grow, increasing end-to-end delay. CallMeter measures jitter per RFC 3550's interarrival jitter algorithm and reports it in milliseconds. Jitter under 30ms is generally acceptable for voice; above 50ms typically causes noticeable quality degradation.

Packet Loss

The percentage of RTP packets sent but never received. Packet loss directly degrades audio and video quality. At 1% loss, voice quality is slightly affected. At 5% loss, voice becomes noticeably degraded. At 10% or more, calls become unintelligible. For video, even 1-2% loss causes visible artifacts and freezes unless error recovery (NACK, FEC) compensates. CallMeter measures packet loss per second from RTCP Receiver Reports and from sequence number gap analysis.

RTT (Round-Trip Time)

The time it takes for a packet to travel from sender to receiver and back. In VoIP, RTT affects perceived call quality because it determines the delay between speaking and being heard. RTT under 150ms is ideal for interactive conversation. Between 150-300ms, users notice delay. Above 300ms, conversation becomes difficult due to talk-over. CallMeter calculates RTT from RTCP Sender Report / Receiver Report exchanges per RFC 3550 Section 6.4.

CallMeter Platform Terms

Test

A reusable SIP testing scenario configuration in CallMeter. A test defines the number of endpoints, call duration, codec selection, media files, group configuration, and worker assignment. Tests can be run multiple times, each producing an independent test run. See Creating a Test.

Probe

A lightweight, scheduled monitoring test that runs continuously at a defined interval (5, 15, 30, or 60 minutes). Probes use the same SIP call infrastructure as tests but are designed for ongoing monitoring rather than one-time testing. Probes evaluate threshold conditions and report health status. See Creating a Probe.

Worker

A compute unit that executes SIP endpoints for CallMeter tests and probes. Workers run as containers and connect to the CallMeter platform via a persistent connection. Cloud workers are managed by CallMeter. User-owned workers are Docker containers you deploy on your own infrastructure for testing internal SIP systems. See Workers.

Group

A subset of endpoints within a test that share the same configuration: registrar, codecs, media files, region, and worker assignment. Multiple groups enable testing scenarios like cross-registrar calling, geographic distribution, and mixed codec environments. See Key Concepts.

Test Run

A single execution of a test configuration. Each time you run a test, CallMeter creates a new test run with its own set of endpoints, metrics, and SIP traces. Multiple runs of the same test allow you to compare results over time or across different conditions. See Running a Test.

Buildup Time

The period during which CallMeter staggers endpoint registration and call initiation. Rather than registering all endpoints simultaneously (which would create an unrealistic SIP burst), buildup spreads registrations over a configurable duration. This simulates gradual ramp-up and tests how infrastructure handles increasing concurrent load.

Caller / Callee

The two roles an endpoint can take in a CallMeter test. Callers initiate SIP INVITE requests. Callees listen for and answer incoming calls. In a typical test, half the endpoints in a group are callers and the other half are callees. Cross-group targeting allows callers in one group to call callees in a different group, testing inter-registrar routing.

Related Pages

• SIP Response Codes -- Full SIP response code reference
• Supported Codecs -- Audio and video codec specifications
• Metrics Reference -- All 150+ metrics measured by CallMeter
• Key Concepts -- Platform building blocks