Voice over IP (VoIP)

Voice over IP (VoIP): Telephone Calls as Data Packets

Voice over IP transforms traditional telephone calls into streams of data packets transmitted over the internet. Instead of dedicating a physical circuit for the duration of a call (as traditional telephony does), VoIP digitizes voice, compresses it, packetizes it, and sends it alongside all other internet traffic. This convergence of voice and data onto a single network has dramatically reduced communication costs and enabled new services impossible with traditional telephony.

How VoIP Works

Think of it this way: traditional telephony is like a dedicated train track between two cities — reserved exclusively for your conversation, even during silences. VoIP is like sending postcards through the regular mail system — your voice is broken into small packets that travel alongside everyone else's packets, arriving at the destination to be reassembled in order.

VoIP call flow:

1. Microphone captures analog voice
2. ADC samples at 8-16 kHz
3. Voice codec compresses samples (e.g., G.729: 8 kbps)
4. Packetizer groups 20-30 ms of compressed voice into one packet
5. RTP/UDP/IP headers added (40 bytes overhead per packet)
6. Packets transmitted over IP network
7. Receiver buffers packets, reorders if necessary (jitter buffer)
8. Codec decompresses and DAC produces analog audio
9. Speaker outputs reconstructed voice

Key Protocols

SIP (Session Initiation Protocol): Establishes, modifies, and terminates calls

• Locates called party (DNS-like lookup)
• Negotiates codec and media parameters
• Text-based protocol (like HTTP) — easy to debug
• Standard for most VoIP systems (Ooma, Vonage, enterprise)

RTP (Real-time Transport Protocol): Carries actual voice data

• Sequence numbers for reordering
• Timestamps for synchronization
• Payload type identification
• Runs over UDP (not TCP — retransmission would add unacceptable delay)

Codecs (voice compression):

Quality of Service Requirements

Voice is extremely sensitive to network impairments:

Latency (end-to-end): Must be < 150 ms for natural conversation. Above 300 ms, conversation becomes difficult (round-trip delay causes speaker overlap).

Jitter (variation in delay): Must be < 30 ms. Compensated by jitter buffer (adds fixed delay to smooth variations). Larger buffer = less jitter impact but higher latency.

Packet loss: Must be < 1% for acceptable quality. Most codecs can conceal 1-2% loss using interpolation. Above 5% = unintelligible.

Bandwidth per call:

• G.711: 64 kbps payload + headers = ~87 kbps per direction
• G.729: 8 kbps payload + headers = ~31 kbps per direction
• With silence suppression (VAD): Average bandwidth reduced ~60%

VoIP Advantages

1. Cost reduction: Long-distance calls cost nothing extra (just internet bandwidth)
2. Convergence: Single network for voice, video, data, messaging
3. Features: Easy call forwarding, conferencing, voicemail-to-email, IVR
4. Mobility: Same phone number works anywhere with internet access
5. Scalability: Adding users requires no new physical infrastructure
6. Integration: Click-to-call from CRM, automated call logging, AI transcription

VoIP Challenges

1. Reliability: Depends on internet availability (no dial tone during outage)
2. Emergency calls (E911): Location determination difficult for mobile IP addresses
3. Power failure: Traditional phones powered by phone line; VoIP needs separate power
4. Security: Voice packets can be intercepted if not encrypted (SRTP recommended)
5. QoS over internet: No guaranteed quality on public internet (ISPs may not prioritize VoIP)

Key Takeaways

1. VoIP converts voice to data packets sent over IP networks, eliminating the need for dedicated circuit-switched telephone infrastructure.
2. SIP handles call setup/teardown while RTP carries actual voice data over UDP — chosen for speed over reliability (retransmission adds unacceptable delay).
3. End-to-end latency must stay below 150 ms and packet loss below 1% for acceptable conversational quality.
4. Modern codecs (Opus, AMR-WB) achieve excellent voice quality at 8-24 kbps — enabling multiple VoIP calls on minimal bandwidth.
5. VoIP dramatically reduces communication costs and enables advanced features (UC, video, presence) impossible with traditional telephony.
6. Quality depends on network conditions — enterprise deployments use QoS mechanisms while consumer VoIP relies on sufficient internet bandwidth.