Copyright 2004 F W Bowen

• If I have DSL service, should I do Voice communications over the Internet, or Vo. IP (Voice Over IP)? • Should I switch from conventional phone to cellular? • What about Cable? • What are the technical differences between a conventional telephone, cellular, and Internet voice? • Understanding the service differences and technical issues can aid in making service choices.

There’s an abundance of technologies and choices • Cellular CDMA – Code Division Multiple Access Verizon, Sprint TDMA – Time Division Multiple Access (GSM) Cingular, AT&T, T-Mobile • Wireless - Wi-Fi (802. 11 a, b, g), Wi-Max (802. 16) • Wireline • Cable TV Analog, ISDN DSL, ADSL,

Many of the choices are migrating toward Internet Protocol (IP) near the User and Application Layers But can use different underlying transport technologies: • ISDN (Wireline) • Sonet (Wireline) • TDM (Time Division Multiplexing) (wireline) • Frame Relay Packet Protocol • Ethernet – Gigabit, 100 Base. T, 10 Base. T • Sonet, Proprietary (Cable)

• Intro • Internet Basics • Wireline Basics • Cellular Basics • Cable Basics • Service Criteria • Summary

IP – “Internet Protocol” The Internet uses a packet protocol that defines the basic information packet

Basic IP Packet Layout

Typical Packet Network (Internet) packet flows

Key IP/Internet Characteristics • Variable packet length • Each packet includes its own source address and destination address • Variable packet routing for a “session” • Multiple “sessions” share the same resources • Asynchronous – no network clocking of packets • “Connectionless”

• Packets are processed by routers as they arrive and contend with other packets for router processing • Packet traffic is notoriously bursty and hard to predict • No fixed path, guaranteed delivery order and delay • Multiple “sessions” using the same paths/circuits can cause congestion and delay when arriving packets destined for an output exceed capacity

Packet Delay Packet delay - the time it takes for an originating packet to arrive at the destination. Example: packet leaves origination at 12: 30: 00 and arrives at destination at 12: 30: 03. Delay is 3 seconds

Packet delay should typically represent the server and router processing times plus transmission time without congestion delays

Packet Jitter

Jitter • Jitter typically reflects network delays due to congestion, traffic load differences, packet order processing differences and routing differences • Interruptions in packet forwarding can make jitter very significant • Jitter complicates application processing at the destination for streaming sessions (audio or video) and requires “buildout” buffering to prevent gaps in stream presentation or buffer overrun • This buffering and processing can increase total delay significantly • Variable server packet processing times can also generate significant jitter

Digital Voice Networks – Digitizing Voice Pulse Code Modulation: DS-0 • Digitizer samples voice every 125 microseconds converting amplitude measurement into an 8 -bit signal level octet • Generates a “Digital Signal Level 0 (DS-0)” digital output

Digital Voice Network DS-0 Characteristics • DS-0 constitutes basic 64 kbps digital “channel” – 8, 000 8 -bit octets per second • One octet every 125 microseconds • Digital “voice circuit” consists of two DS-0 signals, one in each direction • Voice frequency range: 0 -8, 000 khz

Digital Voice Network – DS-1 Multiplexer 24 DS-0 channels are combined into one DS-1 (Digital Signal Level 1) digital stream

DS-1 Characteristics • 24 DS-0 Channels plus one framing bit constitute 1 DS-1 Frame • Generates 1. 544 megabit per second data rate • Commonly referred to as a “T-1” (Transmission Facility Level 1) • DS-1 constitutes basic network transmission rate • Higher rates (T-3, OC-12, …) consist of multiples of T-1

Digital Voice Network – Typical Call

Key Digital Voice Network Characteristics • Fixed length octet “packets” • At call origination a dedicated “circuit/channel” is established/reserved between the calling and called parties • “Connection-oriented” • Only octets for one call are sent on this circuit, and only on this circuit • Synchronous network – all switches and facilities are synced to a master network clock (“atomic clock” accuracy) and consecutive octets are clocked through the network in lock-step every 125 microseconds

• Maximum end-to-end delay typically much less than 250 milliseconds – the comfort limit • Standard requirement is that thru-switch delay be no more than 1, 250 microseconds • Jitter effectively zero (compared to packet)

DSL Data Network Data rates vary Down - 224 kbps up to about 6 mbps Up - 224 kbps up tp 1. 5 mbps Packet/ATM Switch Office Digital Trunk Internet Digital Trunk Analog Line Telephone DSLAM Local Class 5 Office

Typical Cellular (CDMA) Network CV- Compressed Voice Packets PCM – Pulse Code Modulation Octets

Compressed Voice • Compressed voice packets consist of 24 -byte packets that each represent a digitized 20 millisecond voice sample • Compressed voice packets typically occupy a 9, 600 bps radio channel • During congestion or radio difficulties, available channel may only be 4, 800 or 2, 400 bps, requiring different voice compression and dramatically reducing voice quality • Compressing voice adds delay for the 20 millisecond sample period and compression processing

Cellular Voice Compression Methods (CDMA) • EVRC – Enhanced Variable Rate Voice Coder • SMV – Selectable Mode Voice Coder

Key Cellular Network Characteristics • Cellular systems are extensions of the voice network that use voice compression packets to fully utilize the available radio spectrum • For a cell phone call, compressed voice is converted to/from PCM • When most calls were mobile to wireline cellular systems worked acceptably well • For mobile-to-mobile calls, compressed voice is converted to PCM and then from PCM to compressed voice • For these calls, the delay is frequently unacceptable > 250 ms • This is a strong motivation to convert cellular systems to IP networks

Cellular Broadband Data Plans (CDMA) • Limited data access available now, short message service also • 1 XRTT – 144 kbps – Verizon trialing now in some markets • 1 XEV-DO – 2. 4 mbps peak, average 620 kbps • 1 XEV-DV – 5 mbps peak, better suited for Vo. IP • Similar GSM/TDMA data with GPRS, EDGE (lower peak rates? )

Wi-Fi Wireless • Wi-Fi wireless provides IP service over shorter distances in “unlicensed” frequency spectrum • Exploding popularity for “Hot Spot” Internet access, meetings, inter-PC and peripheral data communication (printer, keyboard, mouse) • Extensions of the original Ethernet 802. x series IEEE LAN standards • 802. 11 a – 54 mbps max, 27 mbps effective, 5 ghz frequency range • 802. 11 b – 11 mbps max 4 -5 mbps effective, 2. 4 ghz frequency range • 802. 11 g – 54 mbps max, 20 -25 mbps effective, both frequency ranges

Cable Basics • Cable is “digital radio in a wire” • Coax headend - fiber backend • Carries MPEG and IP • Data-Over-Cable Service Interface Specification • DOCSIS 2. 0 (latest version) Symmetric Service • Adapts cellular technology - Advanced TDMA (A-TDMA), Synchronous CDMA (S-CDMA) • 30 MBPS shared capacity • Vo. IP in plans

Quality of Service (Qo. S) addresses user needs: • Effective Bandwidth (data rate) • Delay • Jitter For voice it can include: • Echo – signal returned to sender from receiver • Sidetone – human factors aid • Voice Fidelity – effective frequency range • More detailed fidelity assurances require a “Transmission Plan”.

Different services have different Qo. S requirements: Voice – • Delay (250 ms max), • Jitter (0 to minimal), • Sidetone, • Echo (inaudible), • Voice fidelity -comparable to traditional analog Streaming video and audio (MPEG, MP 3) – • Needs bandwidth, adequately low jitter (Less than delay) • Depends on build-out buffering • Delays in the seconds can be tolerated Web Surfing – delay measured in seconds generally considered acceptable

Sidetone • Sidetone is the amplitude reduced “Mouth” signal added to the “Ear” signal to mimic normal talking situations where the speaker “hears” what she is saying. • It’s been a human factors element of voice networks for many years.

Echo Every voice circuit has some signal reflected back from the other end: Echo, Frequently recognized as a singing or “tinny” sound. The longer the circuit, the greater the delay, the worse the echo. • Voice networks manage Echo by using Echo Cancellers: • Echo cancellers are DSPs (Digital Signal Processors) that remove echo. . • Echo cancellers add to the delay.

Other criteria to consider • Is the DSL, Ethernet equipment (phone/modem) powered by the DSL/Ethernet cable? • What provisions are there for E 911? • What are the reliability statistics for the system? • How many minutes a year is the system down? (Some digital voice switches significantly outperform the requirement of 2 hours in 40 years, or 3 minutes a year) • Does the system have to regularly go down for cold reboots?

• Technologies and Services are converging to Broadband IP and wireless IP • Qo. S, especially for voice, is frequently unacceptable today. • Service providers make no Qo. S guarantees: “Can you hear me now? ” Ask anyway. • Eventually network capacities may be such that Vo. IP becomes more acceptable. • Wire, cellular, cable? Choose your poison, but understand the tradeoffs. Copyright 2004 F. W. Bowen

Cellular All-IP Architecture