THE BCS PROFESSIONAL EXAMINATIONS BCS Level 5 Diploma in IT October 2009 EXAMINERS' REPORT Computer Networks General Comments The responses to questions were of marginally better quality than April 2009 examinations, although some of them were of varying quality. Again a number of students were on the borderline or near borderline pass category, the latter forming a group not as large as in April 2009 examination. The number of passes was higher than the previous examination and there were more good answers in this session. There were a number of students who were extremely articulate and descriptive in their understanding, unfortunately many of these students fail to apply their knowledge in demonstrated understanding by answering the question in front of them. This has been a significant factor worth noting as it is obvious that student have been cued in the questions to reproduce an answer without actually reviewing the question closely. That said there were some really excellent informed and clearly very engaged networking student prepared to illustrate conceptual answer and backing them up with practical real world network knowledge. In general, students need better preparation based on good understanding of concepts which alone ensures good performance. The students are strongly advised to read examiners reports such as this as part of their preparation for the examination besides preparing answers for questions.. Section A A1 a. A commonly used network access technology for providing Internet access to the home is Asymmetric Digital Subscriber Line (ADSL). Explain what is meant by the term asymmetric and why it is particularly suited to accessing the world wide web? (4 marks) b. The ADSL technology uses the same wires that connect a home to the Plain Old Telephone Service (POTS). Draw a typical ADSL architecture that connects a home to the Internet via the local POTS exchange and describe the function provided by the ADSL router, micro-filter and Digital Subscriber Line Access Multiplexer (DSLAM). c. Briefly describe the Discrete Multi-tone (DMT) modulation scheme used by ADSL and identify those factors which determine how much bandwidth a given customer will receive within their home. (11 marks) 1
Answer pointers a) Asymmetric means that the upstream bandwidth is different to the downstream bandwidth. When accessing the world wide web, the bandwidth required from the user to the web is generally always much lower than the other way around. Typically, a url (low volume data) is issued to the web which will result in the delivery of a large file (the website). 2 marks for asymmetric being unbalanced in up and downstream. 2 marks for recognising that normal web access is also unbalanced, i.e. asymmetric. b) The ADSL technology uses the same wires that connect a home to the Plain Old Telephone Service (POTS). Draw a typical ADSL architecture that connects a home to the Internet via the local POTS exchange and describe the function provided by the ADSL router, micro-filter and Digital Subscriber Line Access Multiplexer (DSLAM). ADSL Router The ADSL router provides an interface for locally attached computers, typically using a WiFi or Ethernet port connection. Micro-filter The micro-filter separates in frequency, the analogue telephone (POTS) from the data stream generated by the ADSL router. DSLAM The DSLAM is within the local exchange and this accepts the data from the telephone lines and provides an interface and access to the Internet service. 4 marks for diagram, 2 marks each for a description of the ADSL Router, Micro-filter and DSLAM Conventional analogue telephony uses frequencies up to 4kHz. ADSL digital signals occupy 26 khz to 1.1MHz. This bandwidth is then divided into 256 channels, each of 4.3kHz. Within these 256 channels, adaptive coding is used (QPSK, QAM) to encode up to 64 kbps per channel. The factors that determine the delivered bandwidth are: The quality of the wire connecting the home to the local exchange The distance from the local exchange (both of the above affect signal attenuation and quality which means that less data can be encoded into each frequency channel.) The contention ratio of the DSLAM will also affect performance in that as more users try and access the Internet so each will get proportionately less bandwidth. DMT: 2 marks for separation of telephony and data, 2 marks for division of frequency into channels, 2 marks for adaptive modulation per channel. Factors: 2 marks for quality and length of wire to exchange, 2 marks for signal degradation and 1 mark for contention ratio of DSLAM. 2
Examiners Comments: This the most troublesome on the paper, attempted by the least number of students and with the lowest performance. Most common mistake was to discuss synchronous and asynchronous networks. Weaker students were less sure of DSLAM function and they tended to have the appearance of experience of end user view of ADSL rather than a taught understanding. Several students gave valiant descriptions of DMT (unfortunately at the expense of the later part of performance) but clearly the majority of the students don t have this knowledge or feel that getting into detail mattered. A2 a. Explain what is meant by the term virtual circuit. (4 marks) b. In networks such as Frame Relay and ATM, virtual circuits are established in accordance with a traffic contract. What is a traffic contract and briefly describe what it defines. (6 marks) c. Explain the process by which a virtual circuit is established using TCP. (11 marks) d. What function is provided by TCP port numbers and what is meant by the term well known port? (4 marks) Answer pointers a) A virtual circuit is the means by which devices establish a communications link between themselves within a packet or cell switched network. Once established a virtual circuit appears as a dedicated and separate link between two points. 2 marks for a VP being a connection, 1 mark for it operating over a packet switched network and 1 mark for it appearing as a dedicated link. b) A traffic contract defines the Quality of Service (QoS) required over a particular virtual circuit connection. The contract therefore needs to define the traffic profile to be supported together with details of its timing requirements. Typical parameters would include average data rate, peak rate, maximum and maximum rates, end to end delay and delay variance requirements. 2 marks for the fact that the Traffic Contract defines the QoS requirements, 1 mark for the fact that the Traffic Contract needs to specify the traffic profile, 3 marks for specifying typical parameters any valid parameter will score 1 mark up to a maximum of 3. c) A TCP connection is uni-directional which means that for two way communications, a connection must be established from each side. Each process does however, follow the same three way handshake procedure. The end-station requesting the connection will issue a TCP segment with the SYN flag set and the sequence number equal to some initial value say x. The receiving end-station if it 3
wants to accept the connection request will return a TCP segment with both the SYN and ACK flags set. It will also choose a sequence number starting value say y. The acknowledgement field of this segment will be set to x+1 to acknowledge receipt of the connection request segment. When the requesting end-station receives this response it will issue one further TCP segment with the ACK field set. The acknowledgement in this segment will be y+1 and the sequence number will be x+1 2 marks for noting that TCP connections are uni-directional. 4 marks for TCP connection request (SYN, sequence number, direction), 4 marks for the TCP response (SYN+ACK, sequence number, acknowledgement number), 4 marks for TCP confirmation (ACK, sequence and acknowledgement numbers. A correctly labelled diagram would also attract full marks as an alternative. d) TCP is able to support protocol multiplexing. This means that more than one protocol can operate simultaneously above a single TCP layer. TCP is able to determine which higher layer protocol to direct traffic to by virtue of the port number used. Port numbers are 16 bits in length and some known as the well known ports are defined to represent certain protocols. For example port 80 = http. 2 marks for port numbers determine the higher layer protocol to use, 2 marks for well known ports being defined to represent standard protocols. Examiners comments: Students, when they were triggered to look at TCP connection establishment they faired well but many having done two previous questions on circuit establishment tried to relate concepts alien to TCP such as fixed route negotiation.. A significant number of students remained in x25 establishment mode and faired badly. The last sub-question is largely well done however some students whether conditioned from answering to low in the stack related port concepts to physical network devise ports and as expected a number see well known to be synonymous with frequently used. A3 4
Answer pointers a. It is often said that the Internet offers a best effort quality of service. What is meant by this term? (6 marks) b. What is meant by the term Virtual Private Network (VPN), and explain how it can be used to provide secure communications over the Internet. (9 marks) c. A salesman has been given a company laptop computer. They are currently attending a conference in another country but need to access their company s network and servers. Explain how; if they have access to the Internet, they could use a VPN to gain secure access their company s network. You may assume that the company network is also connected to the Internet via a Firewall. a) The Internet offers a best effort service which means that it does not guarantee: that data will reach the destination the time it will take for data to reach its destination that the transmission will be error free 2 marks each for delivery, time and errors b) A Virtual Private Network is a means of providing a private network connection over an open or public network. This is achieved through the establishment of virtual circuits or tunnels through the public network. In effect the VPN encapsulates a user s data within a large packet structure; the header of which provides identification. Networks which support VPNs can then use this VPN header to route the packets. The Internet is a shared and open network. A VPN can be established between two points to secure communications through the use a shared security protocol. Say you have two points A and B on the Internet and you wish to secure the communications between them. A and B will agree on a security protocol (encryption technique - IPsec) Information sent from A to B is then secured using this security protocol. Whilst that data is passed over the Internet and could be detected by others, it cannot be decoded without knowledge of the security protocol being used by A and B. The data being sent from B to A can use a different security protocol. VPN 2 marks for providing a private connection through a public network, 1 mark for tunnelling, 2 marks for VPN packet encapsulation. Secure communications 2 marks for the use of a shared security protocol between end points, 2 marks for data encryption. c) 5
The Laptop must be connected to an Internet service via a local ISP. The laptop can then communicate with the remote firewall. The laptop and firewall then form the two ends of a VPN and agree on a security protocol to adopt. However, in order to ensure that the laptop user is valid, the firewall will require some form of authentication before establishing the VPN. Once authenticated, data leaving the laptop will be encrypted using the agreed security protocol and decoded by the firewall. The firewall will then pass the data through to the company s main network and hence, through to the server or whatever the laptop user needs access to. Data sent back to the user is then encrypted by the firewall and decoded by the laptop. This VPN is then said to be tunnelled through the Internet. 1 mark for the laptop and firewall forming two ends of the VPN, 2 marks for firewall authentication, 2 marks for use of an agreed security protocol between the laptop and firewall, 2 marks for data being encrypted whilst sent over the Internet, 2 marks for firewall passing data through to the company s network and 1 mark for the VPN providing a tunnel through the Internet. Examiners comments: Some very good answers. Interesting a number of students opted not to attempt a perhaps indicative of a lack of familiarity with the concept. Some students describing or explicitly stating they were considering WLANs, in terms of some of the key concepts of path negotiation and security. Many students could only focus on secure web pages and their experience of HTTP and public key encryption. That said it scored highest and was the most popular question. Section B B4 IPv4 internetworks operated by a single organisation sometimes use distance vector protocols to manage the transfer of routing information. a) Briefly explain the behaviour of distance vector protocols and illustrate your answer by reference to RIP (Routing Information Protocol). (15 marks) b) What problems can arise with regard to the use of distance vector protocols in networks? Answer Pointers a) Distance vector protocols operate by routers making announcements of networks which they can reach together with the distance (metric) of their route to that destination [3 marks]. The announcements are normally made at regular fixed intervals, typically about once 6
every 30 seconds [3 marks]. The routers do not provide any detail of the nature of the routes being announced and thus some authors refer to DV as being routing by rumour [3 marks]. RIP is a DV protocol. RIPv1 is a fairly old protocol, although still in use. It does not support netmasks and can thus only be used with class based addressing [2 marks]. RIPv2 introduces support for netmasks and thus supports classless internet domain routing (CIDR) [2 marks] and also has a simple authentication mechanism [2 marks]. b) Two problems that occur with such DV protocols are the formation of routing loops [3 marks] and slow convergence [3 marks]. There are also potential security problems as it is fairly easy to set up a computer to introduce spurious messages (with RIPv1) [2 marks] and also as information is often broadcast (RIPv1) then traffic may have to be processed by devices that have no interest [2 marks]. Examiner s Comments: This question was attempted by about 52% of the candidates. A large number of candidates confused the activity of the creation and updating of routing tables with the activity of actual packet forwarding. The creation and updating of routing tables only happens only relatively rarely. As noted above, with the RIP versions of DV this is normally about once every 30 seconds or so. On the other hand, packet forwarding is a VERY regular activity, occurring once for every packet and it will use routing tables which have been created earlier. Part a) VERY clearly asks candidates to illustrate their answer by reference to RIP. A very large number of candidates made no reference to RIP at all. B5 Local Area Networks are often now constructed using wireless technologies, in particular, WiFi (IEEE 802.11). a) Briefly explain the difference between the WiFi operational modes known as infrastructure and ad-hoc. b) Explain the problems created by the presence of what are normally called hidden nodes in a WiFi network. (8 marks) c) Explain why many WiFi networks can have a high error rate caused by cochannel interference. (7 marks) Answer Pointers a) The WiFi operational mode known as infrastructural is one where the devices in the network are considered to be of two types [2 marks]. One devices (possibly more) adopt the role of being access points (APs) [2 marks]. All the other devices are set so as they will only ever attempt to connect to APs [2 marks]. In the operational mode known as ad-hoc, all devices are considered to be similar [2 marks]. Thus, in ad-hoc mode, one device may connect to any other [2 marks]. b) WiFi networks are using wireless radio transmission. WiFi signals only have a limited range [2 marks]. It is thus possible (indeed common) that if an access point is located physically towards the centre of the network, it may be possible for two different devices to be happily able to talk to the access point even though there signals do not reach each other [4 marks]. The result of this is that the two nodes may keep corrupting each others signals (without them being aware) and thus the AP does not receive them correctly [2 marks]. 7
c) In most cases, WiFi networks are using wireless radio transmission in the 2.4Ghz band [2 marks]. This (largely unlicensed) frequency band is used by a large range of other equipment [2 marks] such as bluetooth, microwave ovens and so on [2 marks]. As these are all on similar frequencies, errors caused by this co-channel activity can be common [1 mark]. Examiner s Comments; This question was attempted by about 38% of the candidates. Many candidates did not have a clear understanding of the difference between ad-hoc and infrastructural modes of operation. Many of the answers to part b) incorrectly discussed the issue of unauthorised nodes attempting to intercept traffic rather than the discussing the issues given in the answer pointer above. WiFi networks have grown in practical importance over the last five years or so and candidates are advised to gain a good understanding. Some answers to part c) addressed the issues as expected while other answers were somewhat confused. B6 a) Signals moving over communications links can become modified in various ways. Explain the meaning of the terms attenuation, dispersion and noise in this context. (15 marks) b) Explain the difference between single bit and burst errors and the extent to which parity checking is appropriate as an error detecting technique in both cases. Answer Pointers a) Attenuation is the reduction in amplitude of a signal [3 marks]. This reduction in amplitude may mean signals are interpreted incorrectly at receivers unless amplifiers and/or repeaters are included at appropriate places [2 marks]. Dispersion is the spreading out of a signal in time [3 marks] this is caused as the speed of propagation of signals through medium changes with respect to frequency [2 marks]. Noise is the introduction of unwanted elements to the signal [3 marks]. The noise might come from external sources, but some may be due to molecular vibration of the medium (often called thermal noise) [2 marks]. b) Single bit errors are situations where an isolated bit is corrupt, but those around it are fine [2 marks] whereas burst errors are where a sequence of adjacent bits is all corrupted [2 marks]. Parity checking is quite appropriate for a situation where single bit errors are likely [2 marks] indeed, if both horizontal and vertical parity is used some errors can even be corrected [2 mark]. Parity checking is not really appropriate in situations where burst errors are likely [2 marks]. While parity checking may locate some errors, better techniques such as CRC checks are more appropriate if burst errors are likely to occur. Examiner s Comments: This question was attempted by about 85% of the candidates. Many candidates do not realise that the term thermal noise refers to the molecular vibration of the material but instead believe it is noise due to materials getting very hot. Thermal noise will be present in all materials unless they are at absolute zero. The amount of thermal noise will increase as material get hotter, but in practice it is always present at some level. It will of course be present in optical conductors as well as metallic conductors. Many of the answers offered concerning dispersion do not describe it correctly; candidates are referred to the answer pointers given above. 8