BSc CSIT (TU) Science BIT Network and Data Communications (BIT254) – 4th Semester (Model) Question Paper 2079 Nepal

Channelization / multiple-access techniques (6 marks)

Channelization is a multiple-access method in which the available bandwidth of a link is shared in time, frequency or through code among different stations.

1. FDMA (Frequency Division Multiple Access) The total available bandwidth is divided into non-overlapping frequency bands, and each station is permanently allocated one band. All stations can transmit simultaneously but on different frequencies. Guard bands separate adjacent channels to prevent interference.

Frequency
  ^  | st4 |
  |  | st3 |
  |  | st2 |   (each station = a fixed frequency band, all at the same time)
  |  | st1 |
  +---------------------> Time

2. TDMA (Time Division Multiple Access) The whole bandwidth is just one channel shared in time. Each station is allocated a repeating time slot during which it has the entire bandwidth. Stations transmit one after another in their slots; synchronization (guard times) is required.

Frequency
  ^  |st1|st2|st3|st4|st1|st2|...
  |  +------------------------> Time
     (each station = a fixed time slot, full bandwidth)

3. WDMA (Wavelength Division Multiple Access) The optical-fibre equivalent of FDMA. The high optical bandwidth of a fibre is divided into several wavelengths (colours of light), and each station transmits on its own wavelength. A multiplexer (prism/grating) combines the wavelengths onto one fibre and a demultiplexer separates them at the receiver, allowing many simultaneous high-speed optical channels.

Comparison

Line coding of 100100111 (4 marks)

Conventions: NRZ-L — bit value sets the level (here 1 = high, 0 = low). NRZ-I — invert the level on a 1, no change on a 0. Manchester — mid-bit transition: 0 = high→low, 1 = low→high. Differential Manchester — always a mid-bit transition; additionally a transition at the start of the bit for 0 and no start transition for 1.

For NRZ-I, starting from a low reference: the first 1 inverts the level, each subsequent 1 inverts again, and 0s hold the previous level — giving the level pattern shown. The Manchester and Differential Manchester rows show the within-bit transition for each bit; both are self-clocking codes.

Leaky-bucket algorithm (4 marks)

The leaky-bucket algorithm is a traffic-shaping technique used for congestion control. The analogy is a bucket with a small hole in the bottom: no matter the rate at which water (data) is poured in, it leaks out at a constant rate, and if the bucket is full any extra water overflows and is lost.

In networking the bucket is a finite FIFO queue (buffer):

1. Incoming packets are placed in a queue (the bucket) of fixed capacity.
2. The network removes/sends packets from the queue at a fixed (constant) output rate, regardless of how bursty the input is.
3. If the queue is full when a packet arrives, the packet is discarded.

Thus a bursty input stream is converted into a smooth, fixed-rate output stream. Implementation uses a counter/queue: for a byte-counting version, a counter is initialized to the number of bytes allowed per tick and packets are sent only while the counter permits.

Example (3 marks)

Suppose a host sends data in bursts but the network can accept only 2 packets per clock tick.

A bursty input of 5,1,0 packets is smoothed into a steady output of 2,2,2 packets per tick. (If the bucket capacity were, say, 4, the 5th packet in tick 1 would overflow and be dropped.)

Mathematically, if the bucket sends data at rate $\rho$ for a maximum burst of $b$ over time $t$, the maximum output is $\rho t$, enforcing a constant average rate.

Advantages (1.5 marks)

• Smooths bursty traffic into a constant-rate output, protecting the network from congestion.
• Simple to implement (a FIFO queue with a fixed drain rate).
• Provides a hard upper bound on the average data rate sent into the network.

Disadvantages (1.5 marks)

• The output rate is rigidly fixed, so it cannot take advantage of idle network capacity (no bursts allowed even when the network is free) — this is inflexible compared with the token-bucket algorithm.
• When the bucket is full, packets are simply discarded, which can cause data loss for legitimate bursty applications.

Procedure for pure ALOHA (5 marks)

Pure ALOHA is the original random-access (MAC) protocol in which any station may transmit whenever it has data, with no coordination.

1. Whenever a station has a frame, it transmits it immediately on the shared channel.
2. Because stations transmit at will, two or more frames may overlap in time, causing a collision, and all colliding frames are destroyed.
3. After sending a frame the station waits for an acknowledgment for a time-out period (about $2 \times T_p$, twice the maximum propagation time).
4. If no acknowledgment arrives, the station assumes a collision occurred. It waits a random back-off time $T_B$ and then retransmits.
5. To avoid repeated collisions, the back-off uses a random multiple of the maximum propagation time (e.g. binary exponential back-off), and after a maximum number of attempts ($K_{max}$) the station aborts.

The maximum throughput of pure ALOHA is

Vulnerable time (3 marks)

The vulnerable time is the period during which no other station may transmit if the current frame is to survive without collision. For pure ALOHA a frame can collide with a frame that started just before it or just after it, so the vulnerable time is two frame transmission times:

where $T_{fr}$ is the time to transmit one frame. (For slotted ALOHA this reduces to one $T_{fr}$.)

Numerical: collision-free requirement (2 marks)

Given: frame size $= 200$ bits, channel rate $R = 200\text{ kbps} = 200000$ bps.

Frame transmission time:

Vulnerable time:

Requirement: To keep this frame collision-free, no other station must transmit during the 2 ms vulnerable window — i.e. the channel must remain idle for $2T_{fr} = 2\text{ ms}$ around the frame's transmission. Equivalently, the frame rate must be low enough that, on average, less than one frame is generated per 2 ms vulnerable period.

The OSI (Open Systems Interconnection) reference model, defined by ISO, divides network communication into seven layers, each with a specific function:

• The upper layers (5–7) deal with application-level concerns and are usually implemented in software.
• The lower layers (1–4) handle data transport and are partly implemented in hardware.
• At the sender, data flows down through the layers (each adding a header — encapsulation); at the receiver it flows up (each removing its header — decapsulation).

Difference between circuit switching and packet switching (3 marks)

Practical implications (2 marks)

• Circuit switching guarantees constant bandwidth and low, predictable delay, so it suits real-time, continuous traffic such as voice calls; but it wastes capacity during silence and scales poorly.
• Packet switching maximizes utilization by statistically multiplexing many flows over shared links, so it suits bursty data traffic (web, email, file transfer); but it introduces variable delay and possible packet loss, requiring higher-layer protocols (TCP) for reliability. This efficiency and fault-tolerance is why the Internet is built on packet switching.

Satellite network (3 marks)

A satellite network uses one or more communication satellites in orbit as relay stations (repeaters in the sky) to provide long-distance and wide-area communication.

• An earth station transmits a signal on an uplink frequency to the satellite.
• The satellite's transponder receives, amplifies and shifts the frequency, then retransmits the signal back to earth on a downlink frequency.
• Satellites are classified by orbit: GEO (geostationary, ~35,786 km, fixed over one point), MEO (medium earth orbit, e.g. GPS), and LEO (low earth orbit, e.g. Iridium, Starlink).
• Uses: television broadcasting, telephony, GPS/navigation, weather monitoring, military communication and Internet access to remote areas.
• Advantages: very wide coverage (a single GEO satellite covers ~1/3 of the earth) and reach to remote/inaccessible regions. Drawbacks: high propagation delay (≈250 ms one-way for GEO) and high cost.

Maximum bit rate — Nyquist (noiseless channel) (2 marks)

For a noiseless channel the Nyquist theorem gives:

where $B$ = bandwidth and $L$ = number of signal levels.

Given $B = 3400\text{ Hz}$ and $L = 2$ levels:

Maximum bit rate = 6800 bps (6.8 kbps).

ARP (Address Resolution Protocol) maps a known logical (IP) address to its corresponding physical (MAC) address on the same local network. A host needs the destination MAC address to build a data-link frame, but it usually knows only the IP address — ARP resolves this.

Operation:

1. A host (sender) wants to send a packet to an IP address on its LAN but does not know the receiver's MAC address.
2. It first checks its ARP cache (a table of recently learned IP→MAC pairs). If found, it uses the cached MAC and stops.
3. If not found, the sender broadcasts an ARP request frame on the LAN. The request contains the sender's IP and MAC and the target IP address, with the target MAC field empty; the destination MAC is the broadcast address FF:FF:FF:FF:FF:FF.
4. Every host on the LAN receives the broadcast and compares the target IP with its own. All hosts except the intended one discard it.
5. The host whose IP matches sends back a unicast ARP reply directly to the requester, containing its MAC address.
6. The sender stores the resolved IP→MAC mapping in its ARP cache (with a timer for aging) and now transmits the data frame to that MAC address.

Cached entries expire after a timeout so that changes (e.g. a replaced NIC) are eventually relearned. ARP works only within a single broadcast domain; to reach a remote host the sender resolves the MAC of its default gateway instead.

User datagram (UDP) format (4 marks)

UDP is a connectionless, unreliable transport protocol. A user datagram has a fixed 8-byte header followed by the data, with four 16-bit fields:

0          15 16          31
+------------+------------+
| Source Port| Dest. Port |
+------------+------------+
|  Length    |  Checksum  |
+------------+------------+
|        Data ...         |
+-------------------------+

Because the header is small and there is no connection setup, flow control or retransmission, UDP has very low overhead and is fast — suitable for real-time and query/response traffic.

Well-known ports used with UDP (1 mark)

(Any four of the above are acceptable, e.g. 53, 69, 161, 520.)

These are the techniques used by the DNS (Domain Name System) to resolve a domain name into an IP address.

1. Recursive resolution The client (resolver) sends the query to its local DNS server and demands a complete answer. The local server then takes full responsibility: it queries other DNS servers on the client's behalf (root → TLD → authoritative) and only returns the final resolved IP address (or an error) to the client. The client makes one request and gets the final result — the burden of chasing the answer falls on the server.

Client ---query---> Local DNS (resolves on client's behalf) ---> returns final IP

2. Iterative resolution The client itself does the work step by step. If a queried server does not know the answer, it does not ask others; instead it returns a referral — the address of the next server to ask (e.g. "ask this TLD server"). The client then queries that next server, and so on, until it reaches the authoritative server that gives the final IP. Each server answers with the best it knows.

Client -> Root (referral to .com) -> TLD (referral to authoritative) -> Authoritative (IP)

3. Caching Resolving a name from scratch is slow, so every DNS server stores (caches) the mappings it learns. When a server receives a query whose answer it has cached, it returns the cached record immediately without contacting other servers. Each record carries a TTL (Time To Live) that limits how long it may be cached; when the TTL expires the entry is removed so stale data is not served. Caching greatly reduces DNS traffic and lookup time.

Difference between open-loop and closed-loop congestion control (1 mark)

(Open-loop policies are applied at the source/destination in advance; closed-loop policies react to detected congestion.)

Acknowledgment policy (open-loop) (2 marks)

The acknowledgment policy imposed by the receiver can affect congestion. If a receiver acknowledges every packet it receives, the extra ACK traffic itself adds load to the network. To reduce congestion, the receiver may acknowledge only N packets at a time (cumulative/selective acknowledgment) or send an ACK only when it has a packet to send or a timer expires. Fewer acknowledgments mean less traffic injected into the network, helping to prevent congestion. Thus the acknowledgment policy is an open-loop (preventive) mechanism.

Discarding policy (2 marks)

A good discarding policy by routers can prevent congestion while limiting damage to data integrity. When buffers begin to fill, a router selectively discards less-sensitive packets rather than important ones. For example, in audio/video transmission a router may drop low-priority frames (e.g. some less critical packets) while preserving the packets needed to keep the stream intelligible. By dropping the right packets early, the discarding policy reduces the load and prevents congestion from worsening without seriously degrading quality — making it an open-loop preventive policy.

MAC address (2 marks)

A MAC (Media Access Control) address is the unique physical/hardware address burned into a network interface card (NIC) by the manufacturer. It is 48 bits (6 bytes) long, usually written as 12 hexadecimal digits (e.g. 00:1A:2B:3C:4D:5E). The first 3 bytes are the OUI (manufacturer ID) and the last 3 bytes are the device serial number. It operates at the data-link layer and is used to deliver frames to the correct device within a local network.

CRC encoded frame (3 marks)

Generator $G(x) = x^3 + x + 1 \Rightarrow$ bit pattern 1011 (degree 3, so $r = 3$).

Message $M = 11001000$. Append $r = 3$ zeros: dividend $= 11001000\,000$.

Perform modulo-2 (XOR) division by 1011:

11001000000 ÷ 1011

11001000000
1011
----
0111100...      step-by-step XOR (binary long division)

Working it out:

The remainder (last 3 bits) is 000 (the message is exactly divisible by the generator here).

CRC (remainder) = 000.

Transmitted encoded frame = message + CRC = 11001000 000 = 11001000000.

At the receiver, dividing the received 11001000000 by 1011 again should give a zero remainder if no error occurred.

a. Attenuation (2.5 marks)

Attenuation is the loss of signal strength (energy) as a signal travels through a transmission medium. As distance increases, the amplitude of the signal weakens because the medium offers resistance and converts part of the signal energy into heat.

• It is measured in decibels (dB): $\text{dB} = 10 \log_{10} \dfrac{P_2}{P_1}$, where $P_1$ is the input power and $P_2$ the output power. A negative dB value indicates loss.
• Attenuation increases with distance and frequency and limits the maximum length of a cable segment.
• To compensate for it, amplifiers (for analog signals) or repeaters/regenerators (for digital signals) are placed along the line to restore signal strength.

b. SNMP (2.5 marks)

SNMP (Simple Network Management Protocol) is an application-layer protocol used to monitor and manage network devices (routers, switches, servers, printers) from a central station.

• It follows a manager–agent model: a manager (network management station) communicates with agents running on managed devices.
• Each agent maintains a database of variables called the MIB (Management Information Base), whose objects are named using SMI (Structure of Management Information) and OIDs.
• Main operations: GET (read a variable), SET (change a variable), GET-NEXT/GET-BULK (read multiple), and TRAP (an agent's unsolicited alert to the manager about an event).
• It runs over UDP (ports 161 for requests and 162 for traps), making management lightweight and efficient.