VoIP Networks and Voice Digitization Study Guide

Benefits of a VoIP Network

• More efficient use of bandwidth and equipment:
  • Traditional telephony networks allocate a dedicated $64\,kbps$ channel for every individual voice call regardless of activity.
  • VoIP sharing: VoIP allows bandwidth to be shared among multiple logical connections.
• Lower transmission costs:
  • Legacy systems require substantial equipment to combine individual $64\,kbps$ channels into high-speed transport links.
  • VoIP utilizes statistical multiplexing to combine voice traffic alongside data traffic. This consolidation reduces both capital equipment expenditures and operational costs.
• Consolidated network expenses:
  • Instead of maintaining separate infrastructures for voice and data, organizations convert voice networks to use packet-switched architecture. This creates a single integrated network with a common switching and transmission system.
• Improved employee productivity via IP telephony features:
  • IP phones function as complete business communication devices, not just telephones.
  • Directory lookups and database access: Enabled through Extensible Markup Language (XML) applications.
  • Integration: Employees can use the telephone to retrieve customer info, search inventory, and enter orders directly.
  • Notifications: Employees can receive alerts for service issues, such as changes in delivery dates.
  • Mobility: Software-based (softphones) or wireless telephones offer user mobility.
• Access to new communications devices:
  • Packet technology reaches devices inaccessible to modern Time-Division Multiplexing (TDM) infrastructures, including computers, wireless devices, household appliances, Personal Digital Assistants (PDAs), and cable set-top boxes.
  • Enables the marketing of video-telephones, multimedia terminals, and advanced IP phones.

Components of a VoIP Network

• Telephones: Provide telephony features to users. Formats include IP phones, software-based telephones (PC-operated), or traditional analog or ISDN telephones.
• Gateways: Interconnect the VoIP network with traditional telephony devices, including analog/ISDN phones, faxes, circuit-switched PBX systems, or Public Switched Telephone Network (PSTN) switches.
• Multipoint Control Units (MCU): Required for conference calls involving more than two parties. Media from all members is sent to the MCU, which mixes the media and redistributes it to all participants.
• Application Servers: Provide XML-based services, allowing users access to directories and databases.
• Gatekeepers: Used in H.323 networks to provide Call Admission Control (CAC) to prevent over-subscription and translate telephone numbers or names into IP addresses for routing.
• Call Agents: Provide call control, CAC, bandwidth control, and address translation for IP phones or Media Gateway Control Protocol (MGCP) gateways.
• Video Endpoints: Provide video telephony features. For videoconferences, an MCU must be capable of mixing both video and audio streams.

Legacy Analog and Digital Interfaces

• Legacy Analog Interfaces:
  • Foreign Exchange Station (FXS): The interface that provides battery power, sends dial tone, and generates ringing voltage. Routers with FXS ports behave like a central office or PBX to serve analog phones, faxes, or answering machines.
  • Foreign Exchange Office (FXO): The interface that connects to the PSTN or a PBX. It generates on-hook and off-hook indicators to signal loop closure. Routers with FXO ports behave like a telephone, receiving line power and dial tones from the other side.
  • Ear and Mouth (E&M): Derived from "earth and magneto." "Earth" represents electrical ground and "magneto" represents the electromagnet used for tones. Used for trunks to interconnect two PBX-style devices, gateways (acting as PBXs), and PSTN switches.
• Digital Interfaces Overview:
  • BRI: $2$ voice channels, $1$ signaling channel of $16\,kbps$, overhead of $48\,kbps$, total bandwidth of $192\,kbps$.
  • T1 CAS: $24$ voice channels (not clean $64\,kbps$ due to robbed-bit signaling), in-band signaling, $8\,kbps$ overhead, total bandwidth of $1544\,kbps$.
  • T1 CCS: $23$ voice channels, $1$ signaling channel of $64\,kbps$, $8\,kbps$ overhead, total bandwidth of $1544\,kbps$.
  • E1 CAS: $30$ voice channels, $64\,kbps$ signaling, $64\,kbps$ overhead, total bandwidth of $2048\,kbps$.
  • E1 CCS: $30$ voice channels, $1$ signaling channel of $64\,kbps$, $64\,kbps$ overhead, total bandwidth of $2048\,kbps$.

Call Control and Network Architecture

• The 3 Phases of Call Control:
  • Call Setup: Checks call-routing configuration and determines if bandwidth is available. If unavailable, a busy signal is generated.
  • Call Maintenance: Tracks quality parameters including packet loss, jitter, and delay. The call is maintained or dropped based on these metrics.
  • Call Teardown: Notifies devices to free resources so they are available for subsequent calls.
• Distributed Call Control:
  • Example: Router R1 detects an off-hook state and provides dial tone; it collects digits and checks its local routing table. It sends a setup message to R2; R2 checks its table and sends a ringing signal to the local analog port.
• Centralized Call Control:
  • Example: R1 alerts a Call Agent of the off-hook state. The Call Agent signals R1 to provide dial tone and one-by-one passes digits back to the Call Agent. The Call Agent then looks up the destination and signals R2 to ring the specific analog port ($1/0/0$).

Digitizing Voice: Analog-to-Digital Conversion (ADC)

• Three Steps of ADC:
  1. Sampling: The Digital Signal Processor (DSP) samples the analog signal at regular intervals. The resulting output is a Pulse Amplitude Modulation (PAM) signal measured in volts.
  2. Quantization: The PAM signal is matched to a segmented scale to measure the amplitude (height/voltage). Binary numbers are assigned to represent the voltage level of each sample.
  3. Compression: (Optional) Reduces bandwidth requirements. Examples include mp3, wav, and wma.
• Digital-to-Analog (DAC) Steps:
  1. Decompression: Compressed samples are expanded (optional).
  2. Decoding: DSPs decode digital voice to amplitude values to rebuild the original PAM signal.
  3. Reconstruction: The PAM signal is passed through a filter to reconstruct the original analog waveform.

Theoretical Foundations and Frequencies

• Nyquist Theorem:
  • Reconstruction of a signal from samples is possible if the sampling frequency is greater than twice the signal bandwidth:
  • $f_s > 2 \times BW$
  • Higher-than-needed sampling leads to oversampling, which consumes unnecessary bandwidth. Lower rates lead to precise reconstruction becoming impossible.
• Voice Specifics:
  • Human ear range: $20\,Hz$ to $20,000\,Hz$.
  • Human speech range: $200\,Hz$ to $9000\,Hz$.
  • Traditional telephony range: $300\,Hz$ to $3400\,Hz$.
  • Telephony implementation: The highest frequency is set to $4000\,Hz$. Applying Nyquist, the sampling rate is $8000$ samples per second, which is one sample every $125\,ms$.

Quantization Noise and Techniques

• Quantization Scale: Telephony uses $8\text{-bit}$ quantization, resulting in a scale with $2^{8} = 256$ steps. CD audio, for comparison, uses $16\text{-bit}$ ($65,536$ steps).
• Quantization Noise: Rounding the PAM signals up or down to the nearest available step introduces inexactness. This error is a source of distortion.
• Linear Quantization: Results in a low Signal-to-Noise Ratio (SNR) for small signals (poor quality) and a high SNR for large signals.
• Logarithmic Quantization: Provides higher granularity for lower signals, creating a uniform SNR across the range. This corresponds to the logarithmic behavior of the human ear.
• Bit Encoding Structure:
  • $1$ Polarity bit: Indicates positive or negative signals.
  • $3$ Segment bits: Identify the logarithmically sized segment ($0$ to $7$).
  • $4$ Step bits: Identify the linear step within a segment.

Companding and Coding Standards

• Companding: Portmanteau of "compressing" and "expanding." Analog signals are compressed at the source and expanded at the destination.
• Standards (ITU-T G.711):
  • Mu-law: Used in Canada, U.S., and Japan (Bell systems).
  • A-law: Used in other countries. International connections must use A-law (mu-to-A conversion is the responsibility of the mu-law country).
• Pulse Code Modulation (PCM): Basis of the telephone hierarchy. Rate: $8000\text{ samples/sec} \times 8\text{ bits/sample} = 64,000\text{ bps}$.
• Adaptive Differential PCM (ADPCM): Transmits only the difference between consecutive samples using an adaptive predictor algorithm. This allows for more channels (e.g., $48$ channels on a T1).

Voice Codec Characteristics

Mean Opinion Score (MOS)

• MOS Definition: A subjective numerical rating (judged by humans) of perceived quality (range $1$ to $5$).
• Ratings:
  • $5$: Excellent (Imperceptible distortion; direct conversation).
  • $4$: Good (Just perceptible but not annoying). MOS of 4.0 = Toll Quality.
  • $3$: Fair (Perceptible and slightly annoying).
  • $2$: Poor (Annoying but not objectionable).
  • $1$: Unsatisfactory (Very annoying and objectionable).
• Codec MOS Values: G.711 ($4.1$), G.729 ($3.92$), G.729A ($3.9$), G.726 with $32\,kbps$ ($3.85$), G.728 ($3.61$).

Digital Signal Processors (DSP) Roles

• Voice Termination: DSPs convert analog to digital and back. They provide echo cancellation, Voice Activity Detection (VAD), Comfort Noise Generation (CNG), and jitter removal.
• Conferencing:
  • Single-mode: All participants must use the same codec.
  • Mixed-mode: Supports different codecs and provides transcoding; more DSP intensive, supporting fewer conferences per DSP.
• Transcoding: Real-time digital-to-digital conversion between different codecs.
  • Example scenario: A branch office uses G.729 to conserve bandwidth over an IP WAN to connect to a headquarters voice-mail system that only supports G.711. The headquarters router's DSP transcodes the G.729 stream into G.711 for the voice-mail system.

Audience Interaction and Practical Considerations

• Interactive Questions:
  • Which tools in this room can perform Audio-Digital Conversion (ADC) and Digital-Audio Conversion (DAC)?
  • Can you send each other WhatsApp voice notes?
  • What are the practical advantages and disadvantages of ADC and DAC?

Questions & Discussion (Self-Check)

• Q: Which type of call control uses a call agent to route the call?
  • A: Centralized Call Control.
• Q: What is a DSP?
  • A: A Digital Signal Processor, a specialized processor for voice termination, conferencing, and transcoding.
• Q: Name 3 types of analog interfaces used at gateways.
  • A: FXS, FXO, and E&M.
• Q: What are the 3 components of basic call control?
  • A: Call setup, maintenance, and teardown.
• Q: What phase of call control involves determining if bandwidth is available to place the call?
  • A: Call Setup.
• Q: What sampling frequency is recommended by the Nyquist Theorem?
  • A: Greater than twice the signal bandwidth.
• Q: What is the Hz range for traditional telephone systems?
  • A: $300$ to $3400\,Hz$.
• Q: What is the implication of using 8 bits for quantization?
  • A: It results in a scale with $256$ steps, which is adequate for human speech despite introducing quantization noise.
• Q: What is the purpose of logarithmic quantization?
  • A: To provide uniform SNR for both small and large signals, mimicking the human ear's behavior.