Multiple Choice Questions



Similar documents
COMP VoIP 17/2/2013

Broadband Networks. Prof. Dr. Abhay Karandikar. Electrical Engineering Department. Indian Institute of Technology, Bombay. Lecture - 29.

QoS issues in Voice over IP

Requirements of Voice in an IP Internetwork

Transport Layer Protocols

Network Simulation Traffic, Paths and Impairment

Distributed Systems 3. Network Quality of Service (QoS)

Protocols. Packets. What's in an IP packet

Performance Evaluation of VoIP Services using Different CODECs over a UMTS Network

Per-Flow Queuing Allot's Approach to Bandwidth Management

VoIP Bandwidth Calculation

Combining Voice over IP with Policy-Based Quality of Service

SIP Trunking and Voice over IP

STANDPOINT FOR QUALITY-OF-SERVICE MEASUREMENT

QoS Parameters. Quality of Service in the Internet. Traffic Shaping: Congestion Control. Keeping the QoS

TECHNICAL CHALLENGES OF VoIP BYPASS

Clearing the Way for VoIP

Access Control: Firewalls (1)

Encapsulating Voice in IP Packets

Quality of Service in the Internet. QoS Parameters. Keeping the QoS. Traffic Shaping: Leaky Bucket Algorithm

Sources: Chapter 6 from. Computer Networking: A Top-Down Approach Featuring the Internet, by Kurose and Ross

BCS THE CHARTERED INSTITUTE FOR IT. BCS HIGHER EDUCATION QUALIFICATIONS BCS Level 5 Diploma in IT COMPUTER NETWORKS

Local Area Networks transmission system private speedy and secure kilometres shared transmission medium hardware & software

VoIP Bandwidth Considerations - design decisions

TCP in Wireless Mobile Networks

What is Network Latency and Why Does It Matter?

Computer Networks. Chapter 5 Transport Protocols

2.1 Introduction. 2.2 Voice over IP (VoIP)

Advanced Networking Voice over IP: RTP/RTCP The transport layer

WAN Data Link Protocols

technology standards and protocol for ip telephony solutions

Network Management Quality of Service I

Network administrators must be aware that delay exists, and then design their network to bring end-to-end delay within acceptable limits.

Voice over IP: RTP/RTCP The transport layer

Based on Computer Networking, 4 th Edition by Kurose and Ross

Voice over IP. Overview. What is VoIP and how it works. Reduction of voice quality. Quality of Service for VoIP

TDM services over IP networks

Measurement of IP Transport Parameters for IP Telephony

Voice over IP. Demonstration 1: VoIP Protocols. Network Environment

VOICE OVER IP AND NETWORK CONVERGENCE

Curso de Telefonía IP para el MTC. Sesión 2 Requerimientos principales. Mg. Antonio Ocampo Zúñiga

Improving the Performance of TCP Using Window Adjustment Procedure and Bandwidth Estimation

Final for ECE374 05/06/13 Solution!!

Overview of Computer Networks

Unit 23. RTP, VoIP. Shyam Parekh

RTP Performance Enhancing Proxy

Ethernet. Ethernet. Network Devices

How To Understand The Differences Between A Fax And A Fax On A G3 Network

Indian Institute of Technology Kharagpur. TCP/IP Part I. Prof Indranil Sengupta Computer Science and Engineering Indian Institute of Technology

1. The subnet must prevent additional packets from entering the congested region until those already present can be processed.

Indepth Voice over IP and SIP Networking Course

Protocols and Architecture. Protocol Architecture.

VOICE over IP H.323 Advanced Computer Network SS2005 Presenter : Vu Thi Anh Nguyet

1 Introduction to mobile telecommunications

Basic principles of Voice over IP

Application Note How To Determine Bandwidth Requirements

Supporting VoIP in IEEE Distributed WLANs

Voice Over Internet Protocol(VoIP)

Delivering reliable VoIP Services

Communication Systems Internetworking (Bridges & Co)

Understanding Latency in IP Telephony

Evaluating Data Networks for Voice Readiness

TCP over Multi-hop Wireless Networks * Overview of Transmission Control Protocol / Internet Protocol (TCP/IP) Internet Protocol (IP)

Random Access Protocols

VoIP network planning guide

Computer Networks CS321

5. DEPLOYMENT ISSUES Having described the fundamentals of VoIP and underlying IP infrastructure, let s address deployment issues.

Note! The problem set consists of two parts: Part I: The problem specifications pages Part II: The answer pages

Quality of Service (QoS)) in IP networks

Voice over IP. Presentation Outline. Objectives

Quality of Service Analysis of site to site for IPSec VPNs for realtime multimedia traffic.

TCP and Wireless Networks Classical Approaches Optimizations TCP for 2.5G/3G Systems. Lehrstuhl für Informatik 4 Kommunikation und verteilte Systeme

An Introduction to VoIP Protocols

How To Analyze The Security On An Ipa Wireless Sensor Network

EECS 122: Introduction to Computer Networks Multiaccess Protocols. ISO OSI Reference Model for Layers

Attenuation (amplitude of the wave loses strength thereby the signal power) Refraction Reflection Shadowing Scattering Diffraction

ICOM : Computer Networks Chapter 6: The Transport Layer. By Dr Yi Qian Department of Electronic and Computer Engineering Fall 2006 UPRM

The OSI and TCP/IP Models. Lesson 2

10CS64: COMPUTER NETWORKS - II

VoIP QoS. Version 1.0. September 4, AdvancedVoIP.com. Phone:

How To Solve A Network Communication Problem

Voice Over IP Per Call Bandwidth Consumption

Network Performance: Networks must be fast. What are the essential network performance metrics: bandwidth and latency

Latency on a Switched Ethernet Network

IP Network Layer. Datagram ID FLAG Fragment Offset. IP Datagrams. IP Addresses. IP Addresses. CSCE 515: Computer Network Programming TCP/IP

Lecture Objectives. Lecture 07 Mobile Networks: TCP in Wireless Networks. Agenda. TCP Flow Control. Flow Control Can Limit Throughput (1)

IP - The Internet Protocol

The Impact of QoS Changes towards Network Performance

Transport and Network Layer

Communications and Computer Networks

IP SLAs Overview. Finding Feature Information. Information About IP SLAs. IP SLAs Technology Overview

Agilent Technologies Performing Pre-VoIP Network Assessments. Application Note 1402

Data Link Layer Overview

Mobile Computing/ Mobile Networks

Computer Network. Interconnected collection of autonomous computers that are able to exchange information

[Prof. Rupesh G Vaishnav] Page 1

Note! The problem set consists of two parts: Part I: The problem specifications pages Part II: The answer pages

VoIP in Mika Nupponen. S Postgraduate Course in Radio Communications 06/04/2004 1

An enhanced TCP mechanism Fast-TCP in IP networks with wireless links

Multimedia Communications Voice over IP

Transcription:

Comp18112: VoIP Examples/Revision 1 Barry 7/03/11 University of Manchester School of Computer Science COMP18112: Foundations of Distributed Computing 2011 Voice over Internet Protocol (VoIP) Questions on VoIP section Multiple Choice Questions 1. The one-way delay over a 6000 km optical fibre link between Manchester and New York: (a) cannot be estimated (b) exceeds about 300 ms (c) cannot be much less than 30ms (d) should be less than 3 ms 2. The effect of increasing the bandwidth of the connection between two components of a distributed system will be to: (a) decrease the latency (delay) without necessarily affecting the occurrence of lost packets. (b) increase the bit-rates that can be achieved without necessarily affecting latency, jitter or occurrence of lost packets. (c) decrease the number of bit-errors per second and the occurrence of lost packets. (d) increase the bit-rates achievable while also reducing latency, jitter and the occurrence of lost packets 3. If the bandwidth of a signal is 0 to B Hz, to avoid aliasing distortion the sampling rate must be: (a) greater than or equal to 2B Hz (b) greater than or equal to B Hz (c ) greater than B/2 Hz (d) greater than 2B Hz (e) less than 2B Hz 4. The lowest layer providing host-to-host connection is: (a) physical layer (b) data-link layer (c) network layer (d) transport layer 5. Which of the following statements is true? (a) TCP is a fire & forget protocol with provision for port numbers (b) RTP is a reliable protocol with no provision for port numbers (c) UDP is a fire & forget protocol with no provision for port numbers (d) IP is an unreliable protocol with no provision for port numbers. 6.: Which one of the following statements is true?: (a) Appending an even parity check-bit to a sequence of bits will allow the receiver to detect which bit is in error when a single bit-error occurs. (b) Appending an even parity check-bit to a sequence of bits at the transmitter will allow the occurrence of a single bit-error to be detected, but not multiple bit-errors (c ) Appending an odd parity check-bit to a sequence bits will allow the occurrence of any odd number of bit-errors to be detected. (d) Appending an odd parity check-bit will allow an odd number of bit-errors to be detected whereas an even parity check-bit will allow an even number of bit-errors to be detected.

Comp18112: VoIP Examples/Revision 2 Barry 7/03/11 7. The 16 bit checksum included in the IP header will allow: (a) Any bit-error in the header itself to be detected and corrected (b) Any bit-error in the header itself to be detected but not corrected. (c ) The presence of bit-errors in the header itself to be detected in most cases, but not always. (d) The presence of bit-errors in the IP packet to be detected in most cases 8. Which one of the following statements is true: a) A virtual circuit offers reliable connection-oriented service at the network layer with improved quality of service as compared with traditional IP. b) A virtual circuit offers unreliable connection-oriented service at the network layer with improved quality of service as compared with traditional IP. c) A virtual circuit offers unreliable non-connection oriented service at the network layer with similar quality of service as is obtained with traditional IP. d) A virtual circuit offers reliable connection oriented service at the network layer with improved quality of service as compared with traditional IP. Problems & discussion points 1. Since IP was designed primarily for data why is it now being used for VoIP? 2. Why is VoIP telephony more demanding than streaming audio as used for sound clips and Internet radio? 3. What are the advantages & disadvantages of employing speech bit-rate compression in (i) VoIP over wired links and (ii) VoIP over WiFi? 4. Why are mobile VoIP over WiFi devices traditionally power inefficient? 5. Two mobile VoIP devices with different speech sampling rates, 8000 & 8010 Hz, are communicating 20ms (G711) packets over a WLAN. How could you avoid the accumulation of delay and distortion due to buffer under-flows and over-flows. 6. Can VoIP and data co-exist on a WLAN? What problems can occur & what solutions have been proposed? 7. How can NTP try to distribute time with high accuracy? 8. How is connectivity achieved at the network layer? 9. How is connectivity achieved at the transport layer? 10. Why do we need both a transport & a network layer? 11 How will (a) TCP (b) UDP deal with duplicated packets? 12. Why does each IP datagram header have a time to live? 13. Since routers change IP headers, does the checksum have to change? 14. Add an 8-bit check-sum to: 11101010 11010001 10000001 15. Having set up a VoIP link over wire & fibre between Manchester & New York, how could you improve it if RTCP reports zero lost packets at the New York end and 10% lost packets at the Manchester end? 16. With VoIP, why can t we forget about a missing packet and go on with next one i.e. just miss it out? 17. Explain how the VoIP application illustrates the axiom that the latency of many network links is not so low that it can normally be disregarded. 18. Can you relate the two army problem to a scenario different from the releasing of TCP? 19. Why can t the time-stamp within RTP packets be used to estimate one-way delay? 20. Explain why it would be difficult to have a precisely synchronized measure of exact time across all components of a distributed system. Why can t we just listen to the pips on the one-oclock news on the radio and all set our computer clocks at that point in time? 21. While thinking about the pips, if you have a DAB radio why are they always a little late?

Comp18112: VoIP Examples/Revision 3 Barry 7/03/11 Past exam questions: 1 (a) Why is it difficult to measure the end to end or one way delay of links between components of a distributed system by examining time-stamps and sequence numbers included in datagrams conveyed by the links? What is round trip delay and timing jitter and why are these parameters much easier to measure? What are the major causes of round trip delay and timing jitter in IP networks? (b) Why is round trip delay important in interactive voice over IP (VoIP) systems? (c) How does timing jitter affect VoIP voice quality and how can the effects be reduced? (d) In what way is a traditional Internet Protocol (IP) network considered unreliable? Explain how the transmission control protocol (TCP) achieves reliable and connection oriented communication over an IP network? (e) Why is TCP considered unsuitable for voice over IP applications and why is RTP (real time transfer protocol) over UDP (user datagram protocol) a preferred alternative? 2.(a) Explain why buffers are needed in distributed systems by considering the way VoIP is implemented on personal computers. In giving your answer, explain: i) What a buffer is, by referring to the leaky bucket analogy ii) Why input buffers and output buffers are needed when communicating with concurrent A-to-D and D-to-A conversion processes, as implemented on sound cards or equivalent on-board circuitry. (b) Explain why lost packets occur when VoIP is conveyed by wired computer networks. Explain how the occurrence of lost packets can be identified at the receiver. (c) What is meant by jitter as observed at a VoIP receiver and what causes it? How can such jitter be measured for two successive packets conveyed by RTP? (d) Why are jitter-buffers required for interactive telephone conversations using VoIP? Explain how the receiver s jitter-buffer affects the conversation. How would increasing the jitter-buffer size change this effect and what limits the size of jitter-buffer that may be employed with VoIP? Answers and discussion: Multiple choice: 1. (c); 2. (b); 3 (d); 4 (c); 5(d); 6(c); 7(c) ; and 8 (b); Problems & discussion points 1.Initially IP networks were used for VoIP (i) as a means of providing cheap or free telephone calls over the public Internet (ii) to allow private companies, especially large international corporations, to use the dedicated computer network data links owned or leased by them to be use also for telephony. It is convenient to have just one type of network (IP) rather than two (IP and telephone). Nowadays, the public service telephone service (PSTN) and mobile telephony is changing universally to IP, with virtual circuits normally based on NPLS, to eliminate the distinction between telephone networks and computer networks. 2.With real time telephony, the round trip delay (latency) must be less than about 300 ms for natural 2-way conversation and therefore a large jitter-buffer cannot be employed. Streaming can afford much larger jitter-buffer sizes to provide a reservoir of samples in case some packets are excessively delayed. 3.(i) For wired links, Advantages: Shorter packets may reduce congestion, and reduce transmission delays, jitter and lost packets. Disadvantages: Increases processing complexity and buffering delay, introduces some loss of voice

Comp18112: VoIP Examples/Revision 4 Barry 7/03/11 quality and increases distorting effect of any lost packets and bit-errors since the errors are remembered longer. (ii) For wireless (wi-fi) links: Advantages: As for wired links, but the effects are less significant since packets are not significantly shorter because of the long preamble & header needed for wireless synchronisation Disadvantages: As for wired links and the effects are more significant since bit-errors are more frequent and the cost of the processing, in terms of power (energy) consumption, battery life etc. is more significant. 4. Traditional wi-fi uses carrier sensing multiple access CSMA as its medium access control (MAC) mechanism within the data-link layer. Because a VoIP device cannot be sure when it will get a transmit opportunity (i.e. when the radio channel becomes clear for it to transmit something) and when there is VoIP data for it to receive, it cannot be allowed to power down or sleep between transmissions. Mobile phones sleep in this way and save a lot of battery life. VoWiFi traditionally must remain on and consume power all the time. 5. Read Barry sides to address this problem. At either end you need buffers with high water level and low water level markings, referring to the leaky bucket analogy. These must not be too near the bottom or the top of the bucket by the way. If the water (or data) level falls below the low-level mark, add small number of extra zero valued samples, ideally just one or two in quiet parts of the speech where it will not be noticed. If the high level mark is exceeded discard a sample or two again in a quiet part of the speech. If the sampling rate is 8000 Hz at host A and 8010 Hz at host B, host A must discard 10 samples every second in this way and B must introduce 10 extra samples per second. 6. VoIP and data can coexist. VoIP needs regular transmit opportunities for very small packets whereas data may have to send much larger packets from time to time which may block the channel for a while. Newer versions of the IEEE802.11 protocol, e.g. IEEE802.11e and the recent WMM wi-fi multi-media modifications for WLANs have introduced priority options whereby VoIP can be given priority over data. This is similar to the diffserv mechanism mentioned in the VoIP lectures. 7. See textbook by A.Tanenbaum, Distributed Systems, or try wiki. 8. See VoIP slides which refer to the use of ATM, MPLS and virtual circuits. 9. Refer to VoIP slides which refer to the use of TCP for achieving both reliability and connectivity. For connectivity also look at VoIP notes, the Tanenbaum Computer Networks version 4/5 reference and also look at Alvaro s notes on the idea of sockets. 10. The Network layer provides IP addresses, so allows host-to-host communication via routers without port numbers for differentiating packet streams. The transport layer introduces ports to differentiate streams and also introduces degrees of reliability appropriate to different applications (e.g. close to guaranteed reliability, at the expense of delay, for email and virtually no reliability guarantees for VoIP). 11. Duplicate packets can occur when, for example, TCP or lower layer acknowledgements are not received correctly, so retransmissions are sent when they are not necessary. TCP packets carry sequence numbers in the header, so duplications are easily spotted and discarded. UDP does not introduce sequence numbers, so duplicate packets cannot be immediately recognised. This is one of the reasons why VoIP applications use RTP rather than UDP directly. 12. Time-to-live (TTL) is 8-bit number in the header of an IP packet. It is decremented by one each time an IP datagram is read by a router. If the TTL ever reaches zero, the datagram is discarded. This process eliminates the possibility of a datagram being passed endlessly among routers due to some error in the header or in the routers themselves.. 13. The TTL mechanism means that the IP datagram header must be changed by each router as the datagram travels (hopefully) towards its destination, i.e. when it decrements TTL. The time-tolive parameter is a bit awkward since it is in the IP header and therefore the IP header must be modified by routers. Worse still, this means that the checksum must be recalculated or modified by each router when it decrements the TTL parameter. Fortunately modification is pretty easy with the form of checksum used. It would not be so easy if a CRC were used as with DLL layer packets. 14. Summing we get: 1 0 0 0 1 1 1 1 0 0. Adding 10 to 00111100 gives 00111110.

Comp18112: VoIP Examples/Revision 5 Barry 7/03/11 15. Answer discussed in VoIP lecture and will appear in Lecture notes. 16. Some finer points of real time see VoIP lecture notes/slides. 17. Latency cannot be disregarded because of the interactive nature of a real time telephone call. A round trip delay of more than 300 ms makes interaction difficult even if the sound quality is good. People start interrupting each other. The propagation delay (due to the time it takes for light waves to travel along optical fibers at 200x10 6 m/s, not 300x10 6 m/s by the way) for a round trip between Manchester and New York and back, would be about 60 ms if there were one straight optical fiber across the Atlantic. This is much less than 300 ms of course. But when we add buffering delays, switching delays at routers etc, and delays due to congestion, they accumulate and can easily, as we know, exceed what is subjectively acceptable for VoIP. So the delays of computer networks are certainly not negligible for VoIP, even though they may be considered negligible for text and data applications such as email. A 300 ms delay or even a 3 second delay in an email would not trouble us too much. 18. This is all in Tanenbaum Computer Networks ed 4 Chapter 6, section 6.2.3 (page 503). Refer to the two halves of the blue army as B1 and B2. If the commander of B1 wants to attack, he sends a message to B2 and waits for an acknowledgement. If he does not get an ack, he won t attack, but even if he does, how does the commander of B2 who sent it know that B1 has received it? Without the certainty, (this is life or death, not football) B2 will not attack. A new round of messages and ack s do not solve this problem. Such problems occur frequently in communication over unreliable networks, where some irrevocable (non-reversable) action, like breaking a connection or switching off a machine, is to take place. It is assumed that there is no going back once an action has started this perhaps unrealistic assumption for the blue army, but a totally realistic one for breaking connections and powering down. 19. To measure the one way delay for a single packet when it is received we must note the receive time and then subtract the time-stamp introduced by the transmitter from this. But the clocks at the transmitter and receiver are different and cannot easily be synchronised to an accuracy of less than a few milliseconds. We cannot access the transmitter s clock instantaneously at the receiver to record an accurate receive time. 20. Propagation delays across networks or other communication links are not negligible if we wish to measure delay to an accuracy of a few ms. The transmissions carrying pips take time to travel from source to transmitter and then to a radio user in Manchester, and this time will be different for a radio user in say London. 21. This is just buffering delay which is needed by the packetized and compressed digital transmissions of DAB but not needed in old fashioned FM or AM transmissions. With DAB, further buffering delay is also introduced by transmitters to try to synchronise the signals of multiple transmitters around the country to achieve single frequency networks (SFN) and thus save bandwidth. The delay is about 2 seconds.

Comp18112: VoIP Examples/Revision 6 Barry 7/03/11 Solution to VoIP exam questions: 1(a) Measuring one-way delay between a transmitter and receiver would be difficult because the receiver s clock will not correspond to the transmitter s clock. End to end or one way delay is very difficult to measure because we can t have a universal sense of time at two ends of a link. In distributed systems, clocks at different locations will usually show slightly different times. If they did try to become synchronised by accessing one common reference point, the delay involved would be different in each case. This is because of the different distances from common reference point to receiver or transmitter. So the synchronisation would not work. Round trip delay is the time it would take for a packet of information to reach its destination, be instantaneously copied into a returning packet, and be received by the sender. Round trip delay is easily measured because only one clock is needed. Timing jitter means delay variation or the variation in one-way delay. If we can t measure oneway delay, how can we measure variation in one-way delay or jitter? Actually it s easy with timestamped packets. Assume we receive three packets with consecutive sequence numbers (this is important) and their time-stamps (in seconds) are respectively: 1034 1034.006 1034.009 Assume that a timer at the receiver is set to 0 as soon as the first of these 3 packets arrives). So the receiver time for the receipt of the first packet is 0. If the receiver s time for the second & third packets are 0.006 and 0.009 seconds we would conclude that the delay variation (jitter) is zero. But assume that the packets are received at 0, 0.004 and 0.010 seconds. If the one-way delay for the first packet is D (unknown), the delays for the second and third packets are D-0.002 and D + 0.001 respectively. The delay variation or jitter can then be said to be 0.003 seconds. RTCP would do this sort of calculation over a number of packets and take an average. Algorithm: If time-stamps are S1, S2, S3, and received times are R1, R2,R3, Delay variation = [(R3-S3) - (R2-S2)] - [(R2-S2) - (R1-S1)] Major causes of round-trip delay: (1) time to fill each packet (2) Propagation delay along wires, optical fibre or radio. Major causes of timing jitter: Packets taking different routes through the network and sometimes being held up at congested routers. (b) Round trip delay is important in interactive VoIP because if I say hello, the round trip delay determines the time I must wait to hear the other person say hello if he/she responds as soon as he/she hears my hello. If the round trip delay is too long, (i.e. greater than about 300 ms) I will start speaking again thinking the other person may not have heard me. Just then his response will reach me and overlap my speech. This makes interactive conversation becomes difficult (c) Timing jitter affects voice quality by causing some packets to be lost due to excessive delay. The effects can be reduced, at the expense of delay by introducing a jitter-buffer at each receiver. This maintains a reservoir of data. (d) Traditional Internet Protocol (IP) is considered unreliable because there are no guarantees about correct delivery. IP provides connectionless service where datagrams are conveyed independently by routers towards their destination IP addresses. Datagrams may be delayed, damaged, lost or arrive out of order. Different routes may be taken to the same destination. There is no payload CRC or checksum for bit-error checking. TCP provides connection-oriented reliable transmission by introducing sequence numbers and a payload check-sum within its 20 byte header. Reliability is achieved by a mechanism for acknowledging correct receipt and retransmitting packets when necessary. Connectivity is achieved by exchanging and maintaining (remembering) state information, such as sequence numbers, at both ends of a link until the connection is terminated.

Comp18112: VoIP Examples/Revision 7 Barry 7/03/11 (e) TCP is not ideally suited to VoIP because the acknowledging and retransmission mechanisms incur delay and increases congestion. UDP is simpler than TCP, connectionless and unreliable. It is a fire and forget protocol. It simply encapsulates the following UDP datagram within the payload of an IP datagram. The unreliability of UDP is not such a problem for VoIP because voice not quite as sensitive as data to bit-errors and lost packets. An occasional bit-error might hardly be noticed in a voice stream, and also, the loss of a whole packet can be concealed, to a degree, by filling in the gap by a waveform segment that looks and sounds approximately right, perhaps because it resembles the previous segment. However, UDP still has some problems for VoIP since UDP datagrams may be lost, damaged or re-ordered and the receiver must know when this happens. These problems are remedied by the real time protocol (RTP) since each packet is given a sequence number and a time-stamp. It is also useful for the transmitter to know how many of its packets are getting through and with what delay variation. Facilities for doing this are provided by RTP and its sister protocol RTCP (real time transport control protocol). 2(a) (i) A buffer is an array or block of storage. It is like a leaky bucket, with a hole in the bottom, being filled periodically by turning on and off a water tap. An input buffer would be filled at a regular rate an analogue-to-digital converter and emptied periodically by the CPU. This is like a bucket being filled by a constantly dripping water tap and being regularly emptied by the CPU. An output buffer would be filled periodically by the CPU and emptied at a regular rate to feed an external digital to analogue converter. In both cases, the CPU being a little late or early to fill or empty a bucket (i.e. buffer) is not so critical. (ii) Sound cards control their own sampling rates using independent crystal controlled clocks and run continuously, independent of the CPU. The sound processing is therefore distributed between concurrent operations on the sound card and on the CPU. Sound cards must have buffers to store sections of their inputs and outputs to accommodate the fact that the CPU cannot read or write individually every single sample of an input or output stream, because it has many other tasks to perform. The output buffer must never be allowed to become completely empty otherwise the D to A converter would have no samples to convert and would therefore fail catastrophically. Also the input and output buffers must never become so full that they overflow and bits are lost. Imagine a low water mark and a high water mark drawn on the water bucket. (b) With wired networks, packets are rarely undelivered or delivered with bit-errors. Packets can be lost, occasionally, at routers when they are faulty or overloaded by excessive traffic. Most lost packets are just late i.e. they are excessively delayed by the network and arrive too late to be usable. Sequence numbers introduced in the headers of RTC packets allow the occurrence of lost packets to be identified. (c) Jitter is variation of one-way delay. Jitter arises from packets taking different routes towards the destination or being held up in queues at routers along the way. Time-stamps in RTP packets allows intervals between send times of successive packets to be determined. Receive times may be noted at receiver for each packet. Difference between send interval and receive interval is delay variation for successive packets jitter. (d) A jitter-buffer at each receiver allows for variation in delay across the network. The receiver s jitter-buffer introduces delay, and affects the number of packets lost due to excessive delay. Increasing the buffer size decreases the number of lost packets in a wired VoIP link at the expense of increasing delay. A round trip delay of greater than about 300 ms makes conversation difficult, and this fact limits the amount of delay that can be introduced at each receiver.