Part A:

1. Assume that two communication nodes are transmitting data connected by a communication link. The network and system parameters are given below. Assume that receiving node uses the XON and XOFF flow control signalling.

• Link transmission rate 10 Mbs
• Node separation 80 km
• Receive buffer size 1.2 MB
• Frame size 1200 bytes
• Receiver information processing rate 100 KB/sec

a. Assuming that the receiver sends a XOFF signal when the receive buffer is P% full in order to avoid any data or packet loss. Also, assume that the receiver read the receive buffer when the buffer is full. Calculate the value of P that will support a lossless communication while offering full utilisation of the receiver buffer.

b. Calculate the time interval between the XON and XOFF signals. An XON signal will be transmitted by the receiver when the buffer occupancy level goes down to 70% or below.

2. Consider two nodes in a communication network are separated by a 100 km distance. Assume that the transmission link speed is 10 Mbs. The I frame length used on the link is 400 bytes including a 20 byte header, and the ACK frame length is 20 byte. Assume that on average out of every 2000 bytes 11 bits are corrupted due to the channel error.

a. Calculate an appropriate window size that can be used to support a window based ARQ protocol.

b. Calculate the effective throughput (in bits/sec) that can be achieved for a) Stop & Wait ARQ protocol and b) Go-Back-N protocol for your calculated window size.

c. Calculate the efficiency of the selective ARQ protocol for above parameters. Then increase the frame length by 100% keeping the same size overhead. Compare these values to explain the effect of frame length.

3. Suppose that a single communication link is used to connect a statistical multiplexer to a statistical demultiplexer. The multiplexer operates in the duplex mode and connected to 8 input traffic sources. The output link operates in the TDM mode with 4 time slots/frame. The output link data rate is 2.048 Mbs with a frame rate of 4000 frames/sec.

For the following parameters calculate the expected end-to-end packet delay.

• Data activity factor α = 0.7
• Link distance = 25 km
• Input packet arrival is exponentially distributed with an inter arrival time of 4 ms.
• Fixed size input packet of 128 bits length.

4. Consider a statistical multiplexer receives data from 20 different input links. Packet inter-arrival times are exponentially distributed with an average value of 10 ms (for each input). Assume that the multiplexer transmit packets on an output link operating at 10 Mbs. The input packet sizes are exponentially distributed with an average value of 100 bytes. The multiplexer has a maximum buffer length of 500 bytes.

a. Calculate the packet loss probability at the multiplexer.

b. Calculate the average number of packets in the multiplexer and the average packet delay introduced by the multiplexer.

c. Calculate the packet loss probability if the packet inter-arrival time is changed to 4 ms.

5. Assume that a 100 Mbs Ethernet switch is used to connect 10 terminals. Terminals are generating data bursts and transmitting those data using the Ethernet link. The switch input buffer size is 2000 bytes. Terminals are generating variable size data bursts with an average burst length of 400 bytes and an interarrival time of 1ms. The Ethernet frame carries an 18 byte overhead and uses a 4 byte FCS (Frame Check Sequence) field. Calculate the average frame delay introduced by the switch.

6. Assume that a LAN contains four terminals connected to a hub operating at 10 Mbs. Frame length used is 1500 bytes. Assume that each link length is 150 meters, velocity of propagation is 2.5×108 m/sec and jamming interval is 3.4 µsec. Calculate following values.

a. Maximum channel capacity wasted for every collision assuming that the LAN uses the CSMA/CD (Carrier Sense Multiple Access with Collision Detection) protocol.

b. Maximum channel capacity wasted for every collision assuming that the LAN uses the ALOHA protocol

7. Consider a polling network consists of 12 user terminals and a central controller known as the Base Station (BS). All user terminals are located at a fixed distance from the BS. The BS to user terminal distance is 100 meters. The system uses a 500 byte information frame, equal size polling and acknowledgement frames are used which are 20 bytes long. The transmission data rate is 12 Mbs. Assume that user terminals are allowed to transmit one information frame per poll. Each information frame is acknowledged by the BS.

a. Calculate the cycle time used in this polling network.

b. Plot the polling and message transmission sequences showing the exact timing of each transmission.

8. Assume that a 12 port Ethernet switch is used to connect eight terminals in a network. The transmission data rate of the switch is 100 Mbits/sec and has a 4 MB buffer. Assume that each computer generate data for transmission using an interarrival time of 5 ms (exponentially distributed) and the packet length is exponentially distributed with a mean value of 1500 byte.

Calculate followings

i) Total traffic load generated at the switch.

ii) Packet loss probability at the switch.

iii) Average packet delay experienced at the switch.

iv) Average queue utilisation.

9. A security video camera transmit images over a two hop communication network where the hops are connected by a packet router. The camera generates image after every 100 ms where the image size is uniformly distributed between 12000 to 18000 bytes. The network limits packets to a maximum size of 1500 byte, which includes a 32 byte header. The transmission links are error free and have a speed of 10 Mbits/sec. Each hop is 100 km long. The packet router uses the datagram technique which operates in the store and forward mode. The router is considered to be a lossless device. Calculate following values:

a. Average number of video packets generated/sec.

b. Queuing delay at the router.

c. End to end packet transmission delay.

10. Suppose that a router receives an IP packet containing 1100 data bytes and has to forward the packet to a network with maximum transmission unit of 400 bytes. Assume that the IP header is 20 bytes long. Show the fragments that the router creates and specify relevant values in each frame header (i.e. total length, fragment offset and more bit).

Part B:

In this section you will design and analyse the performance of a packet switched router serving data and VoIP (Voice over Internet Protocol) traffic. The router is an egress router for a University LAN (Local Area Network). An egress router is a router through which packets leave one network for another network. Assume that the router is connected to an external ISP (Internet Service Provider) network using a 100 Mbs broadband link that uses a carrier class Ethernet protocol. The router receives voice and data packets from different terminals within the LAN. The router transmits received packets on the broadband link by repacketising received data. Traffic and router information are listed below.

Traffic Information:

Data Traffic:

Number of data terminals = 1000 (Data traffic is generated by all terminals.)

Average data burst interarrival time (exponentially distributed) = 20 ms

Average data burst length (exponentially distributed) = 1250 bytes

Data terminal activity factor = 0.17

Date packet is generated by each node by encapsulating the data burst with TCP, IP and Ethernet headers. Also, 32 bit CRC is used along with an 8 byte preamble. Data packets are transmitted via the LAN to the switch.

VoIP Traffic:

Number of VoIP terminals = 1800 VoIP traffic is generated by all terminals.)

Voice packet interarrival time = 20 ms

Voice coding rate = 32 kbits/sec

Voice activity factor = 0.4

Average voice call duration = 3.5 minutes

Average no. of voice calls/terminal = 5 calls/hour

VoIP packets are generated by encapsulating voice bits with UDP/RTP, IP and Ethernet headers. Also, 32 bits CRC is used along with an 8 byte preamble. VoIP packets are transmitted via the LAN to the switch.

Header sizes: TCP: 20 bytes, UDP/RTP: 20 bytes, IP: 20 bytes, Ethernet: 18 bytes.

Router Information:

The router receives both voice and data packets the strip all the header, CRC and preamble bits to generate a new Ethernet packet. Data payload from each incoming packet is repacked into a new outgoing Ethernet packet using the same size header, CRC and preamble bits.

Voice packets received from different terminals is aggregated in to a larger packet to transmit over the broadband link. Packet aggregation information is listed below.

Data packet size over the broadband link = data burst + all headers+ CRC bits +preamble

VoIP packet size over the broadband link = average 15 VoIP payload aggregated + all headers +CRC bits + preamble.

Broadband link speed = 100 Mbits/sec

Switch architecture is shown in the following diagram. Data traffic concentrator accepts data packets from packet classifier (not shown in the diagram), repack with headers, CRC & preamble bits.

Following the repacking it files the packet in the transmission queue. The VoIP concentrator accepts VoIP packets from the packet classifier then aggregates VoIP payloads and generates packets to file in the transmission queue. The transmission queue size is finite.

Calculate following traffic and switch parameters.

i. Total load (packets/sec) generated at the input of the queue by the data packets arriving from all terminals. Also, calculate the traffic load in bits/sec.

ii. Total load (packets/sec) generated by VoIP packets arriving at the input of the queue from all terminals. Also, calculate the traffic load in bits/sec. Note that packet aggregation is used.

iii. Which queuing model can be used to analyse the performance of the above switch? Justify your answer.

iv. Find out the buffer length (in packets and in bits) that will keep the average packet loss below 1%.

v. Using the calculated queue length value find out the average queue size of the switch.

vi. What is the average packet delay introduced by the above switch?

vii. The average goodput and average throughput ratio measured at the output of the switch. The ratio of the switch throughput to the total input traffic. (Ratio of less than one means gain due to VoIP packet aggregation.)