Assignment 2
(Due Date: 23:55 pm, Oct. 30, 2020)
Note:
(1) Each question has two weights, the first is for CSC 463 students, and the second is for CSC 569 students.
(2) You should submit your solutions in One file in pdf format to the Connex.
[60%, 40%] (I) Assume that a research group tries to test the performance of AODV on a real testbed. A typical testing scenario is the so-called ¡°monkey walk¡± testing: several people walk according to given instructions. The testing is to be done in a building shown below.
In the testing, there are four nodes A, B, C, and D. Initially, nodes A and B are in the same location in Office A, and nodes C and D are in the same location in Office B. Node A and node D are stationary. Node B moves from Office A to Office B, and simultaneously node C moves from Office B to Office A. Suppose that all four nodes have implemented AODV (http://www.rfc-archive.org/getrfc.php?rfc=3561) and will use it as the routing layer protocol. Node A will begin to send node D CBR (Constant Bit Rate) traffic with an interval time of 250 ms when node B begins to move. The distance between nodes A and D is 500 meters. The walking speed of nodes B and C is 1 meter per second.
For simplicity, we assume that one packet transmission (if the transmission is possible) for one radio hop is 5 ms regardless of the size of the packet. Also we assume that the size of each CBR packet is 128 bytes, a value small enough so that no fragmentation is needed. Each node has a Send Buffer of 64 packets in size. A packet will be dropped if the Send Buffer is full. The radio transmission range for all nodes is the same value of 250 meters. We assume perfect radio propagation and a perfect MAC protocol such that a message can be successfully transmitted if two nodes are within each other¡¯s transmission range.
All other configuration parameters, if used, can be found at Chapter 10 in the draft (http://www.rfc-archive.org/getrfc.php?rfc=3561).
For simplicity, we assume that there is no random delay when broadcasting control messages and all broadcast can be successfully received. We also ignore the local repair mechanism.
The testing begins when nodes B and C begin to move, and it ends when nodes B and C arrive at their destinations (the point D and the point A respectively). What is the throughput (in bps) for this testing scenario? To obtain full mark, you need to illustrate the detailed operations of AODV in this test scenario and show your calculation step by step.
[40%, 30%](II) Consider the following simple network of 3 nodes: A —– B —– D
Assume that an early version of AODV (check the early version from the following paper http://www.cs.ucsb.edu/~ebelding/txt/aodv.ps) is used as the routing protocol at all 3 nodes. Assume that D is the only destination, and both A and B need to send data to D. The link B — D is fragile and may be broken at any time.
(1) Please explain why in the following scenarios {S1-S4} a loop may be formed between A and B.
(2) Read the AODV RFC draft (http://www.rfc-archive.org/getrfc.php?rfc=3561) to point out how the problems are fixed in the newest version of the protocol.
Let A and B have active routes to D to begin with: A —-> B —-> D.
S1. When the link B—D goes down, B generates a RERR. If B does not increment its seqno for D and if the RERR gets dropped, there will be a loop.
S2. When the link B—D goes down, B generates a RERR. Suppose B DOES increment its seqno for D. If the RERR gets dropped, and B deletes its route before A’s route expires, there will be a loop.
S3. Suppose B’s route expires while A is still pointing at it. Consider the following possibilities:
(a) B deletes the route on expiry.
(b) B keeps the route around, unchanged, as an expired route.
(c) B keeps the route around as an expired route, it increments the route’s seqno, and never deletes it.
(d) B keeps the route around as an expired route, it increments the route’s seqno for D, and deletes it after some time.
S4. The AODV process at B is restarted suddenly (reboot, crash, etc.). If A does not detect the restart as a link-breakage, and continues to point to B, there will be a loop. Here, we assume that B restarts in a vanilla state, so this case could be treated as one in which all routes at B are suddenly deleted.
[0%, 30%] (III) This question is to let you understand the impact of radio channel condition on wireless transmission and to learn how performance may be improved by re- transmitting small ACK packets.
A satellite is orbiting the earth at the geostationary orbit. It relays packets between two ground stations on the surface of the earth. The distance between the satellite and the ground stations is 72000 km. Ground station A sends a 1500 bytes packet to the satellite and the satellite forwards it (after it has been completely and successfully received by the satellite) to the ground station B. Ground station B replies with a 30 bytes ACK which is subsequently received by the satellite and forwarded to ground station A (again, it is forwarded, after the satellite has completely and successfully received the ACK packet.) Assume that the bit error rate of the channel from ground station to satellite (and vice versa) is 10^(-5). Assume that the ground station A knows exactly the time at which the ACK would have to be received for its transmitted packet. If an error-free ACK has not been received by that time, the ground station A re-sends the data packet. Let the propagation delay be at the speed of light. Also assume that the data transmission speed on all stations and the satellite to be 1Mb/sec.
What is the average time it takes until the packet is confirmed as delivered (time between start of first transmission from ground station A until an error-free ACK has been completely received by the ground station A)? What is the average time it takes if the bit error rate is degraded to 10^(-3)?