POTS & PABX

POTS Fundamentals

Plain Old Telephone Service (POTS) is the analog public switched telephone service that has connected homes and businesses since the late 19th century. Despite the digital revolution, POTS established the electrical and signaling conventions that still underpin modern telephony — from battery polarity to voice-bandwidth limits.

The 48V DC Battery Feed

Every POTS subscriber line is powered by the central office (CO) through a 48V DC battery supply. The battery was originally a bank of lead-acid cells; modern COs use rectified mains power with battery backup. The nominal voltage is −48V DC on the tip wire relative to the ring wire — the negative polarity reduces electrolytic corrosion of buried cable conductors. When the subscriber is on-hook (idle), the line presents an open circuit and no current flows. When the handset is lifted (off-hook), the hook switch closes the circuit, and the CO detects the resulting DC current draw (typically 20–120 mA) as a signal to provide dial tone and switch the line into the network.

The battery feed passes through a differential transformer (hybrid coil)that converts the 2-wire subscriber loop to a 4-wire circuit for switching and amplification. The hybrid coil must balance the impedance mismatch between the subscriber line (typically 600–900 Ω) and the 4-wire equipment to minimizetalking current echo. Poorly balanced hybrids cause the speaker to hear their own voice reflected back — a defect known as sidetone.

Dial Tone Generation

Dial tone is the audible signal indicating the CO is ready to accept dialed digits. In North America, dial tone is a simultaneous combination of 350 Hz and 440 Hz(a dissonant two-tone chord). The 350 Hz component is generated by a precision oscillator (historically a tuning fork, later a crystal oscillator), and the 440 Hz component is similarly generated. The combined signal is applied to the subscriber line at approximately −15 to −20 dBm. The frequency pair was chosen to be easily distinguishable from speech and ring signals while remaining comfortable for extended listening.

Internationally, dial tone varies: the UK uses 350+450 Hz (similar to North America), Germany uses 425 Hz single tone, and Japan uses 400+800 Hz. The variation exists because different national telephone administrations (PTTs) independently developed signaling standards before international harmonization through the ITU.

Ring Voltage

To alert a subscriber of an incoming call, the CO applies a ring signalto the line. In North America, this is a 90V AC signal at 20 Hz, applied in a cadence of 2 seconds on, 4 seconds off (2s/4s). The 20 Hz frequency was chosen because it falls below the 300 Hz lower bound of the voice band, ensuring the ring signal does not interfere with speech. The 90V AC amplitude is necessary to drive the electromagnetic ringer — a tuned LC resonant circuit consisting of a coil and a capacitor tuned to approximately 20 Hz (Q ≈ 5–10). The high voltage is needed because the ringer presents a high impedance at 20 Hz (typically 1–4 kΩ), and sufficient current (approximately 25–50 mA) must flow to mechanically strike the bells.

Modern electronic ringers use a piezoelectric buzzer driven by a lower-voltage AC signal (typically 20–50V) at the same 20 Hz frequency. The CO still applies the 90V ring signal for backward compatibility, but the electronic ringer's higher sensitivity reduces the current requirement. When the subscriber lifts the handset during ringing, the hook switch closure is detected as a ring trip— the CO immediately disconnects the ring signal and connects the line to the voice path.

The Subscriber Loop — Tip and Ring

The subscriber loop is a 2-wire metallic circuit running from the central office to the subscriber's premises. The two wires are called tipand ring, named after the tip and ring contacts of the ¼-inch telephone plug used in early switchboards. The tip wire carries the positive (or less negative) potential, and the ring wire carries the negative potential. Together, they form a balanced differential pair that rejects common-mode noise (power-line hum, RF interference).

The maximum subscriber loop length is limited by DC resistance andsignal attenuation. Standard copper cable (24–26 AWG, 0.5–0.4 mm diameter) has a resistance of approximately 130–85 Ω/km per conductor. The CO battery feed must overcome the loop resistance while maintaining sufficient current for the telephone to operate — a minimum of about 20 mA at the far end. This limits the practical loop length to approximately 18 km (11 miles) for 24 AWG cable, beyond which loading coils, repeaters, or carrier systems are required.

Voice Bandwidth — 300 to 3400 Hz

POTS limits the voice channel to 300 Hz to 3400 Hz, a bandwidth of approximately 3100 Hz. This range was determined by the ITU-T (formerly CCITT) based on speech intelligibility studies: the 300 Hz lower bound preserves the fundamental frequencies of male and female speech (85–300 Hz) and the first two formants that carry vowel identity; the 3400 Hz upper bound captures consonant energy (fricatives and sibilants) essential for word recognition. The 300 Hz lower cutoff also eliminates 50/60 Hz power-line hum and low-frequency mechanical vibration.

The 3400 Hz upper limit enables 4 kHz channel spacing in frequency-division multiplexing (FDM) systems, with 100 Hz guard bands above and below each channel. This convention, established in the 1920s for carrier telephony, persists in the 8 kHz sampling rate of PCM voice encoding (G.711): the Nyquist theorem requires sampling at twice the highest frequency, so 8 kHz sampling captures up to 4 kHz of bandwidth. Modern codecs (G.729, Opus) operate within this band but can be configured for wider bandwidth when the network supports it.

Signaling

Telephony signaling encompasses the control signals that establish, maintain, and release calls. Signaling in POTS is divided into supervisory (on-hook, off-hook, ring trip), address (dialing), and informational(busy, reorder, dial tone) categories. Two primary address-signaling methods have dominated: pulse dialing and DTMF.

DTMF — Dual-Tone Multi-Frequency

DTMF is the standard tone-dialing method introduced by AT&T in 1963 (commercial deployment 1968). Each key on the telephone keypad generates a unique pair of tones — one from a low-frequency row group and one from a high-frequency column group. The keypad is arranged in a 4×4 matrix:

• Row frequencies: 697 Hz, 770 Hz, 852 Hz, 941 Hz
• Column frequencies: 1209 Hz, 1336 Hz, 1477 Hz, 1633 Hz

Pressing the "5" key generates 770 Hz + 1336 Hz simultaneously. The 4×4 matrix yields 16 tones — enough for digits 0–9, *, #, and four letters (A, B, C, D). The 1633 Hz column (A/B/C/D) is reserved for special military and network control functions. The tone pairs are deliberately non-harmonically related to minimize false triggering from speech — the frequency ratios are irrational numbers, so no combination of speech formants can accidentally produce a valid DTMF pair.

DTMF tones must meet strict twist requirements: the row tone must be 2–4 dB stronger than the column tone (normal twist) to account for the telephone line's frequency-dependent attenuation (higher frequencies attenuate more). The maximum twist limit is 8 dB. Each tone pair must be present for at least 40 msminimum (with a 40 ms minimum silence between tones) to be reliably decoded by the CO. The detection algorithm uses Goertzel algorithm bandpass filters at each of the 8 DTMF frequencies, checking that both the row and column tones are present above a threshold while ensuring no other tones are present (to reject talk-off from speech).

Pulse Dialing

Pulse dialing (pulse-code dialing) is the older method, originating with Strowger's step-by-step exchange. The subscriber's rotary dial interrupts the loop current at a rate of 10 pulses per second (pps). Each digit is represented by a corresponding number of pulses: digit "1" produces 1 pulse, digit "2" produces 2 pulses, and digit "0" produces 10 pulses. The break/make ratio is approximately 60:40 — the loop is open (break) for 60 ms and closed (make) for 40 ms per pulse.

Between digits, the dial pauses for a pause interval of at least 300 ms (typically 600–800 ms), which the exchange uses to detect the end of the dialed number. The pulse-dialing standard requires the dial mechanism to produce pulses with a rise time of less than 2 ms and a bounce duration of less than 10 ms. The mechanical governor in the rotary dial uses a centrifugal speed regulator to maintain constant pulse rate regardless of how quickly the user releases the dial.

Off-Hook and On-Hook Detection

The CO monitors the DC loop current to determine the subscriber's hook state. On-hook (idle) means the circuit is open — zero current. Off-hook (active) means the handset is lifted, closing the hook switch, and current flows through the loop. The CO detects off-hook by sensing a loop current exceeding a threshold (typically 8–20 mA). When the subscriber hangs up (on-hook), the current drops to zero, and the CO releases the connection. This simple current-detection mechanism is the most fundamental signaling method in telephony.

A flash hook (briefly pressing the hook switch for 100–800 ms) generates an on-hook/off-hook sequence that the CO interprets as a request for special features: call waiting, three-way calling, or call transfer. Flash hook duration must be carefully controlled — shorter than 100 ms is ignored, and longer than 800 ms is interpreted as a hangup. The exact timing varies by country and exchange type.

Central Office Switching

The central office (CO) is the telephone exchange that terminates subscriber lines and provides switching, signaling, and billing functions. The evolution of CO switching technology spans over a century, from mechanical step-by-step systems to modern digital packet switches.

Step-by-Step (Strowger) Switching

Almon Strowger's 1891 patent introduced step-by-step (SxS) switching, in which a series of electromechanical uniselectors directly translate dial pulses into physical switch positions. A 7-digit telephone number drives seven switches in cascade: the first selects a group of 10,000 lines (10 banks × 10 contacts), the second selects 1,000 within that group, and so on down to the individual line. Each wiper arm has 10 (or 20) contact positions, and the pulse train from the subscriber's dial advances the wiper mechanically.

SxS switching was slow (5–8 seconds for a 7-digit call), noisy (audible clicking as wipers advanced), and required frequent maintenance (contact cleaning, oiling). However, it was the first fully automatic switching system and remained in service in some rural US exchanges into the 1990s. The SxS switch also introduced the concept of link switching — selecting a path from an inlet to an outlet through a multi-stage network of switches.

Crossbar Switching

Crossbar switching (patented 1938, deployed 1940s–1960s) replaced the rotary wiper with a matrix of horizontal and vertical bars. A call is established by energizing a horizontal bar (which aligns all contacts in that row) and a vertical bar (which selects the column), with a latching mechanism at the intersection holding the connection. Crossbar was faster (sub-second switching), quieter, and more reliable than SxS. The No. 5 Crossbar (Western Electric, 1948) handled up to 10,000 lines and remained a workhorse of the Bell System for decades.

Electronic Switching System (ESS)

The No. 1 ESS (AT&T/Western Electric, 1965) was the first stored-program electronic telephone exchange. A central computer (initially a 1A processor with 48K words of ferrite core memory) controlled an electromechanical switching matrix. The ESS stored its switching logic in read-only memory (initially magnetic rope read-only memory, later semiconductor ROM), allowing features to be updated without rewiring. Features introduced by ESS included call forwarding, speed dialing, call waiting, and three-way calling.

Later ESS variants included the No. 2 ESS (1970, for smaller offices), the No. 3 ESS (1976, digital switching), and the No. 4 ESS (1976, a long-distance digital toll switch using PCM encoding). The No. 4 ESS could handle over 100,000 simultaneous calls and used time-slot interchange (TSI) to switch 64 kbps PCM channels between T1 and T4 carrier systems. The transition from electromechanical to electronic switching enabled the digital revolution in telephony.

Numbering Plans

The North American Numbering Plan (NANP), introduced in 1947, divides telephone numbers into a 3-digit NPA (Numbering Plan Area, or area code), a 3-digit NXX (central office code), and a 4-digit XXXX(subscriber number). The original 86 area codes were assigned based on call volume: low-volume states received codes with "0" as the middle digit (e.g., 201 New Jersey, 212 New York), while high-volume states received codes with "1" as the middle digit (e.g., 312 Illinois, 213 California). The 0 and 1 middle digits were originally reserved because they could not be confused with the first two digits of a local 7-digit number under pulse dialing.

Class of Service

The CO assigns each subscriber a class of service (COS) that determines permitted features: residential (touch-tone only, no coin), business (touch-tone + pulse, call forwarding), coin-operated (payphone, accept coins), and restricted (no long-distance, no 900 numbers). COS is implemented in the CO's software database and controls which call types the subscriber can originate and receive. Modern ESS and IP-based switches use COS tables with dozens of parameters, including routing restrictions, feature access codes, and billing options.

PABX Evolution

As businesses grew, the cost of running individual telephone lines from every desk to the CO became prohibitive. Private switching systems — key systems, PBXs, and hybrid systems — solved this by sharing a smaller number of CO trunks among many internal extensions, adding call-routing intelligence at the customer premises.

Key Systems — KSU and 1A2

Key systems are the simplest form of customer-premises telephone switching. A Key Service Unit (KSU) is a small cabinet (typically wall-mounted, the size of a circuit breaker panel) that terminates both CO lines and internal extensions. Each telephone on a key system has a line button for each CO line, with a corresponding LED indicating line status (idle, ringing, in use). To make an outside call, the user picks up the handset and presses an idle line button; to answer a ringing line, they press the flashing button.

The 1A2 key system (introduced 1963 by Bell System) was the dominant small-office telephone system for decades. It used a 25-pair cable from the KSU to each telephone, with 2 pairs for the talk path (tip/ring), 1 pair for the ring signal, and 1 pair for the lamp/LED control. A 1A2 system supported up to 9 CO lines and 40+ extensions, with features like intercom, hold, and page. The 1A2 system was purely analog — no digital switching, no stored-program control — and its simplicity made it extremely reliable.

PBX — Private Branch Exchange

A PBX (Private Branch Exchange) is a more sophisticated switching system that provides automatic call routing, operator console support, and advanced features. Early PBXs (1960s–1970s) were electromechanical — essentially miniaturized CO switches (crossbar or SxS) housed in the customer's building. A human operator answered incoming calls and connected them to internal extensions using a cord switchboard.

The transition to digital PBX in the 1980s brought stored-program control, TDM switching matrices, and integration with data networks. Digital PBXs encoded voice as 64 kbps PCM and switched it through time-slot interchange (TSI) circuits. A typical mid-range PBX could serve 500–10,000 extensions with 50–500 CO trunks. Features included automatic call distribution (ACD), call detail recording (CDR), least-cost routing, and integration with modems and fax machines.

Hybrid Systems

Hybrid PBX systems combined digital switching for internal calls with analog trunks for external connectivity. This approach allowed businesses to gradually transition from all-analog to digital without replacing all equipment at once. Hybrid systems could typically accommodate a mix of analog extensions (for standard telephones), digital extensions (for proprietary feature phones), and T1/PRI trunks for external connectivity. The hybrid architecture dominated the 1990s and early 2000s as businesses prepared for VoIP migration.

ACD — Automatic Call Distribution

Automatic Call Distribution (ACD) is a PBX feature that routes incoming calls to groups of agents based on predefined rules. ACD was developed for call centers in the 1970s and became a critical component of customer service operations. The distribution algorithms include:

• Linear hunt: calls always start at the first agent and work down
• Round-robin: calls rotate through agents sequentially
• Most idle: calls go to the agent with the longest idle time
• Skills-based: calls are routed based on the agent's language or expertise
• Predictive dialer: the system dials numbers ahead of time and connects answered calls to available agents

Modern ACD systems integrate with CRM platforms to screen-pop customer data, provide real-time dashboards showing queue depth and agent status, and implementIVR (Interactive Voice Response) front-ends that allow callers to self-select their desired department.

PABX Features

Hunt Groups

A hunt group (or line hunt group) is a set of CO trunks that are searched sequentially when an incoming call arrives. The PBX "hunts" through the group until it finds an idle trunk, then routes the call to the appropriate extension. Common hunt types include sequential (linear), circular (rotary), and most-idle. Hunt groups ensure incoming calls find a path even when some trunks are busy, improving call completion rates.

Call Parking

Call parking places a call on hold in a system-wide parking slot rather than on a specific telephone's hold button. Any extension can retrieve the parked call by dialing the parking slot number. This is useful in warehouses, retail environments, and hospitals where staff move between locations. Parking slots are typically numbered (e.g., *70, *71, *72) and have configurable timeouts — if the call is not retrieved within 30–120 seconds, it rings back to the original extension.

Conference Calling

Conference calling bridges three or more parties into a single conversation. Analog PBXs used a conference bridge — a mixer circuit that summed the audio signals from each participant. The mixer must carefully manageecho and gain: each participant hears the sum of all other voices but not their own, requiring the bridge to subtract each participant's signal from their receive path. Digital PBXs and VoIP systems implement conferencing in software, using digital signal processing (DSP) to manage echo cancellation and voice activity detection for each participant.

IVR — Interactive Voice Response

IVR systems allow callers to interact with the PBX through voice prompts and DTMF input (or speech recognition). An IVR tree typically starts with a greeting ("Thank you for calling. For sales, press 1. For support, press 2...") and routes calls to the appropriate department or queues them. Modern IVR systems useNLU (Natural Language Understanding) to parse speech input ("I need to check my balance") rather than requiring the caller to press digits. IVR is the front-end for most large call centers, handling thousands of simultaneous calls.

Voicemail

Voicemail records messages when a call is not answered. Early systems used cassette tape loops — a continuous magnetic tape recording each message in sequence and overwriting the oldest messages when full. Modern voicemail systems are software-based, storing messages as compressed audio files (typically G.711 at 64 kbps or G.729 at 8 kbps) on disk arrays. Features include message waiting indication (a stutter dial tone or MWI lamp on the phone), remote access (call in from any phone and enter a PIN to retrieve messages), message forwarding, and email-to-voicemail integration.

Call Detail Records (CDR)

CDR captures metadata for every call passing through the PBX: calling number, called number, start time, end time, duration, trunk used, and termination cause. CDR data is essential for billing (allocating costs to departments),traffic engineering (identifying peak hours and trunk utilization), andcompliance (recording call patterns for regulated industries). CDR records are typically stored in a database and can be queried with SQL-like tools for reporting. The CDR format follows ITU-T Recommendation Q.825 for ISDN and is exported as comma-separated values (CSV) or through APIs in modern PBX systems.

Least-Cost Routing (LCR)

Least-cost routing automatically selects the cheapest trunk group for each outbound call based on the destination number, time of day, and carrier rate tables. For example, a call to a local number might route through a local PSTN trunk, a long-distance call might route through a SIP trunk with lower per-minute rates, and an international call might route through a VoIP provider with favorable international rates. LCR continuously updates its routing tables as carrier rates change, ensuring the PBX always selects the lowest-cost path. In multi-site enterprises, LCR can route calls through the nearest site's trunks to minimize long-distance charges.

TDM to VoIP Transition

T1 and E1 Trunking

T1 (North America, Japan) and E1 (Europe, rest of world) are the fundamental digital trunking standards. A T1 line carries 24 channelsof 64 kbps PCM voice over a 1.544 Mbps bit stream, while an E1 carries 30 voice channels (plus 2 signaling/synchronization channels) at 2.048 Mbps. Each voice channel is a DS0 (Digital Signal Level 0), sampling at 8 kHz with 8-bit quantization (G.711 μ-law for T1, G.711 A-law for E1).

T1/E1 framing organizes the bit stream into frames. T1 uses SF (Superframe) or ESF (Extended Superframe) framing: SF contains 12 frames of 193 bits each; ESF contains 24 frames with a 4-kbps data link for maintenance and performance monitoring. E1 uses PCM30 framing: 32 time slots of 8 bits each, with time slot 0 for synchronization (CRC-4) and time slot 16 for common-channel signaling (CCS) or channel-associated signaling (CAS). The T1/E1 physical layer uses AMI (Alternate Mark Inversion) line coding (T1 can also use B8ZS — Bipolar with 8-Zero Substitution — to ensure clock recovery when long strings of zeros occur).

ISDN — BRI and PRI

ISDN (Integrated Services Digital Network) provides end-to-end digital connectivity over copper pairs. The two primary interfaces are:

• BRI (Basic Rate Interface): 2B+D — two 64 kbps B-channels for voice/data and one 16 kbps D-channel for signaling. BRI is intended for small offices and residential use, providing 128 kbps total throughput.
• PRI (Primary Rate Interface): In North America, 23B+D (1.544 Mbps = T1 rate); in Europe, 30B+D (2.048 Mbps = E1 rate). PRI provides a high-capacity digital trunk for PBXs, supporting 23 or 30 simultaneous voice calls over a single physical connection.

ISDN's D-channel carries signaling information (call setup, teardown, and supplementary service requests) using Q.931 protocol. Thisout-of-band signaling is faster and more reliable than the in-band DTMF and pulse signaling of analog trunks. ISDN also supports data calls(using the B-channels for 64 kbps or 128 kbps data), caller ID(sent on the D-channel during call setup), and direct inward dialing (DID)(the PBX passes the dialed extension number to the CO via the D-channel).

SIP Trunking Replacing PRI

SIP trunking replaces PRI trunks with IP-based connectivity over the public internet or a private managed network. Instead of 23 or 30 dedicated voice channels on a T1/E1, a SIP trunk provides a pool of concurrent call paths (channels) that scale dynamically. A business needing 50 simultaneous calls provisions 50 SIP channels on a single internet connection rather than purchasing three separate T1 lines. SIP trunks eliminate the need for PRI hardware (CSU/DSU, PRI cards) and reduce costs by leveraging the existing data network.

The transition from PRI to SIP trunking accelerated in the 2010s as session border controllers (SBCs) solved the security and interoperability challenges of connecting PBX systems to SIP trunk providers. An SBC acts as a firewall and protocol converter, handling NAT traversal, topology hiding, encryption (TLS/SRTP), and codec transcoding between the enterprise and carrier networks.

VoIP PABX Systems

Asterisk — The Open-Source Pioneer

Asterisk is an open-source PBX/telephony platform originally created by Mark Spencer in 1999 and first released as version 1.0 in 2004. Written in C, Asterisk runs on Linux and provides a software-based implementation of PBX features, VoIP gateway functionality, and telephony protocol stacks. Asterisk supports SIP, IAX2 (Inter-Asterisk eXchange), H.323, and MGCP protocols, and can interface with legacy TDM hardware through Digium/Sangoma PCI cards (TE1200 for T1/E1, AEX800 for analog).

Asterisk's architecture is based on a channel abstraction that provides a uniform interface across different physical and virtual interfaces (SIP endpoints, DAHDI T1 channels, conference bridges, voicemail boxes). The dialplan (written in extensions.conf or the more modern ARI/AMI APIs) defines call routing logic using contexts, extensions, and priorities. Asterisk can function as a PBX, an IVR, a voicemail server, a conferencing bridge, a media gateway, or a combination of all these.

FreePBX

FreePBX is a web-based management GUI for Asterisk, providing a graphical interface for provisioning extensions, trunks, ring groups, IVR menus, queues, and call recording. FreePBX eliminates the need to manually edit Asterisk configuration files, making Asterisk accessible to administrators without deep Linux expertise. FreePBX is available as a standalone distribution (built on CentOS/Debian) or as an appliance through Sangoma's commercial offerings (FreePBX UC Edition).

3CX Phone System

3CX is a commercial software-based PBX that runs on Windows, Linux, or as a cloud instance. 3CX provides a complete UC solution including SIP trunking, web conferencing, call center features, Windows/Mac/Android/iOS softphones, and Microsoft Teams integration. 3CX uses a proprietary licensing model based on the number of simultaneous calls rather than extensions, making it cost-effective for businesses with many extensions but moderate concurrent call volumes.

Cisco Unified Communications Manager (CUCM)

Cisco Unified Communications Manager (formerly CallManager) is the enterprise-grade IP-PBX platform for Cisco collaboration solutions. CUCM runs on Cisco UCS hardware or VMware and supports up to 100,000 registered devicesin a single cluster. CUCM integrates with Cisco Jabber (softphone), Cisco IP phones (8800 series), Webex, Unity Connection (voicemail), Expressway (remote access), and UCCX (contact center). CUCM uses SCCP (Skinny Client Control Protocol)for Cisco IP phones and SIP for third-party devices and trunking.

Cloud PBX — RingCentral, Vonage, Zoom Phone

Cloud PBX (hosted PBX / UCaaS) moves the PBX entirely to the provider's data center, eliminating on-premises hardware. The enterprise provides internet connectivity and IP phones; the provider handles call routing, voicemail, IVR, ACD, and all other PBX features. Major cloud PBX providers include:

• RingCentral: Enterprise UCaaS with video meetings, team messaging, and global SIP trunking. Supports up to 10,000 users per account.
• Vonage: SMB-focused cloud PBX with API integrations for CRM and custom workflows.
• Zoom Phone: Integrated with Zoom's video platform, providing a unified voice/video/messaging experience. Zoom Phone uses a proprietary protocol for Zoom-to-Zoom calls and SIP for PSTN connectivity.

Cloud PBX economics are subscription-based: typically $20–$40 per user per monthincluding a set of features (international calling may be additional). The provider handles all maintenance, upgrades, and redundancy, reducing IT burden. The trade-off is dependence on internet quality — voice over IP requires low latency (<150 ms), low jitter (<30 ms), and minimal packet loss (<1%) for acceptable quality.

SIP Architecture

SIP Registrar

The SIP registrar is the server that authenticates SIP endpoints and maintains their current location. When a SIP phone boots up, it sends aREGISTER request to the registrar, providing its IP address and contact URI. The registrar stores this mapping in a location servicedatabase, enabling other SIP entities to route calls to the endpoint regardless of its physical location. The registration expires after a configurable TTL (typically 3600 seconds) and must be periodically refreshed with re-REGISTER requests.

SIP Proxy Server

A SIP proxy forwards SIP requests and responses between endpoints. The proxy does not handle media — it only routes signaling. Proxy types includestateless proxies (which simply forward each message without maintaining dialog state) and stateful proxies (which track transactions and can fork requests to multiple destinations). A SIP proxy consults the location service to determine the current contact URI for the called party, then forwards the INVITE to that address. Proxies can also implement authentication, access control, and load balancing.

Back-to-Back User Agent (B2BUA)

A B2BUA is a SIP entity that terminates one SIP dialog and originates another, acting as both a UAC (User Agent Client) and UAS (User Agent Server). Unlike a proxy, a B2BUA maintains full state for both sides of the call and can modify SIP headers, rewrite SDP, and implement call policies. Most VoIP PBXs (Asterisk, FreePBX, CUCM) function as B2BUAs — they answer the incoming INVITE, create a new internal dialog, and route the call to the destination endpoint. B2BUAs are essential for NAT traversal, codec transcoding, and security (hiding the internal network topology).

RTP Media Streams

Once a SIP session is established, the actual voice media flows asRTP (Real-time Transport Protocol) packets over UDP. RTP provides timestamps (for jitter compensation), sequence numbers (for packet loss detection and reordering), and payload type identifiers (for codec negotiation). RTP packets typically carry 20 ms of audio per packet — at G.711 64 kbps, each packet contains 160 bytes of payload plus 12 bytes of RTP header and 8–12 bytes of UDP/IP headers. The RTP stream flows directly between endpoints (peer-to-peer) when possible, or through a media relay (TURN server) when NAT prevents direct connectivity.

SDP Negotiation

SDP (Session Description Protocol) is embedded in SIP INVITE and 200 OK messages to describe the media parameters of the session. SDP specifies the IP address and port for RTP, the codecs supported (in preference order), the clock rate, and any special parameters (e.g., telephone-event for DTMF, video payload types for video calls). The called party responds with its own SDP offering its preferred codecs, and the caller selects the mutually supported codec. This negotiation ensures interoperability between endpoints that may support different codec sets.

CODECs — G.711, G.729, and Opus

The choice of voice codec determines bandwidth consumption, latency, and voice quality:

• G.711 (ITU-T): 64 kbps PCM (μ-law in North America, A-law elsewhere). Uncompressed, toll-quality voice. No transcoding loss. Requires 80 kbps with RTP/UDP/IP headers (20 ms packets). Most widely supported codec.
• G.729 (ITU-T): 8 kbps CS-ACELP (Conjugate-Structure Algebraic Code-Excited Linear Prediction). 10:1 compression ratio over G.711. Requires licensing fees (patented algorithm). Good quality but introduces ~25 ms algorithmic delay. Total bandwidth: ~31.2 kbps with headers.
• Opus (IETF RFC 6716): Variable bitrate (6–510 kbps), adaptive codec that dynamically switches between SILK (speech) and CELT (music) modes. Royalty-free. Excellent quality at all bitrates. Supports wideband (up to 48 kHz) audio. Increasingly preferred for WebRTC and modern VoIP.

Modern Unified Communications (UCaaS)

Unified Communications

Unified Communications (UC) integrates multiple communication modalities into a single platform: voice (telephony), video (conferencing), messaging (chat/SMS), presence (availability status), and collaboration (screen sharing, file transfer). UCaaS (Unified Communications as a Service) delivers these capabilities from the cloud, eliminating the need for on-premises PBX, video conferencing servers, and messaging gateways. The UCaaS market surpassed $30 billion annually by 2025, driven by remote work, BYOD policies, and the convergence of IT and telecom.

Microsoft Teams Phone

Microsoft Teams Phone integrates cloud PBX functionality into the Microsoft Teams collaboration platform. Teams Phone provides calling plans (domestic and international), direct routing (connecting Teams to existing SIP trunks via SBCs), and operator connect (carrier-managed PSTN connectivity through Teams). Teams Phone supports auto-attendant, call queues, call recording, and compliance recording. With over 300 million monthly active users, Teams is the largest UCaaS platform, and Teams Phone represents a significant displacement of traditional PBX systems.

WebRTC

WebRTC (Web Real-Time Communication) is an open standard (W3C/WHATWG) that enables real-time voice, video, and data directly in web browsers without plugins. WebRTC uses SRTP (Secure RTP) for encrypted media, DTLS(Datagram Transport Layer Security) for key exchange, and ICE (Interactive Connectivity Establishment) for NAT traversal. The WebRTC audio pipeline includes acoustic echo cancellation (AEC), noise suppression (NS), and automatic gain control (AGC) — all implemented in the browser. WebRTC enables browser-based softphones, click-to-call features, and browser-based contact center agent interfaces.

Softphones and BYOD

Softphones are software applications that emulate a physical telephone on a computer, tablet, or smartphone. Popular softphones include Zoiper, Bria, Linphone, and the native Teams/Jabber/Zoom clients. Softphones support all standard PBX features (hold, transfer, conference, voicemail) and leverage the device's built-in microphone and speakers. BYOD (Bring Your Own Device) policies allow employees to use their personal smartphones as their business phone, with a separate work phone number and identity. MDM (Mobile Device Management) solutions enforce security policies (VPN, encryption, remote wipe) on BYOD devices accessing corporate telephony services.

The Enterprise Telephony Migration Path

The evolution of enterprise telephony follows a clear progression, with each technology building on (and eventually replacing) its predecessor:

Stage 1: POTS (1880s–1960s)

Every desk has a dedicated analog line to the central office. No internal switching — all calls go through the CO. The subscriber loop is a simple 2-wire circuit with 48V battery feed, and the telephone is a dumb endpoint. POTS is reliable but expensive at scale: a company with 500 employees needs 500 CO lines.

Stage 2: Key System (1960s–1980s)

A KSU provides basic shared access to CO lines. Each phone has line buttons for incoming/outgoing trunks, and internal calls use a simple intercom circuit. Key systems reduce the number of CO lines needed (typically 1 line per 3–5 users) and add basic features like hold and intercom, but lack call routing intelligence.

Stage 3: PBX (1970s–1990s)

A PBX adds automatic call routing, extension dialing, and operator functions. The PBX shares CO trunks among all extensions using a switching matrix (crossbar, then digital TDM). Features include hunt groups, call forwarding, conference calling, and CDR. The PBX is a significant capital investment (hardware, installation, ongoing maintenance) but dramatically reduces per-line costs.

Stage 4: IP-PBX (2000s–2010s)

An IP-PBX replaces the TDM switching matrix with IP packet switching. Endpoints are SIP phones connected over Ethernet (PoE — Power over Ethernet eliminates separate power supplies). The IP-PBX can run on commodity server hardware (Asterisk on Linux) or as a virtual machine. SIP trunks replace T1/PRI trunks, and QoS (Quality of Service) mechanisms (DiffServ, 802.1p priority tagging) ensure voice packets are prioritized over data traffic.

Stage 5: SIP Trunking (2010s–present)

SIP trunks replace physical T1/E1 circuits with virtual voice channels over the internet. The PBX connects to one or more SIP trunk providers, gaining flexibility (scale channels up/down on demand), cost savings (no per-channel hardware), and geographic flexibility (a local number in any area code). SIP trunking decouples the voice network from the physical infrastructure — the PBX can be on-premises, in a colocation facility, or in the cloud.

Stage 6: Cloud PBX / UCaaS (2010s–present)

Cloud PBX eliminates the on-premises PBX entirely. The service provider hosts the switching platform, manages the PSTN connectivity, and delivers all features as a subscription service. The enterprise provides internet access and IP phones (or softphones). Cloud PBX evolves into UCaaS, integrating voice, video, messaging, presence, and collaboration into a single platform. The endpoint shifts from a desk phone to a laptop, tablet, or smartphone — the "phone" becomes an app rather than a device.

Nikola Tesla and Telephony Infrastructure

Nikola Tesla's contributions to telephony are indirect but foundational. His development of alternating current power systems (1888) provided the electrical infrastructure that telephone exchanges depended on — every CO required reliable AC power converted to the 48V DC battery feed. Tesla's work onresonant circuits and tuned coils (US Patent 645,576, 1900) directly influenced the design of loading coils, filter networks, and impedance-matching transformers in long-distance telephony. His 1893 demonstration of radio frequency transmission at the Franklin Institute laid the groundwork for radio telephony, which enabled the first transatlantic voice link (1927). Tesla's patent for a remote-controlled device (US Patent 613,809, 1898) pioneered the concept of remote signaling and control that became the foundation for SS7 signaling networks, remote PBX management, and modern network operations centers. While Tesla did not invent the telephone, his contributions to AC power, resonant circuits, and radio were essential enabling technologies for the telephone's evolution from a point-to-point device to a global network.

Summary

POTS and PABX represent the complete arc of enterprise telephony: from the analog subscriber loop (48V battery, tip/ring, 300–3400 Hz voice band) through electromechanical and digital switching (Strowger, crossbar, ESS) to packet-switched VoIP (SIP, RTP, cloud PBX). The migration path — POTS → Key System → PBX → IP-PBX → SIP Trunk → Cloud PBX/UCaaS — reflects a consistent trend: from dedicated circuits to shared infrastructure, from on-premises hardware to cloud services, from desk phones to softphones, and from voice-only to unified communications. Each transition has reduced cost, increased flexibility, and expanded the definition of what a "telephone call" can be.