Final Exam -- ELG 5125

December 11 , 2003  - -  (no documentation allowed)

 

Please answer 3 sub-questions of Question 6 and four out of the five questions 1 through 5.

 

Please limit your answers to Questions 1 through 5 to one page per question, and to about half a page for each sub-question of Question 6. Please start a new page for each question.

 

Question 1: IP telephony (20 points)

(a)    Please explain in a few words how the Session Initiation Protocol (SIP) works.

(b)   What information is described by the Session Description Protocol (SDP) and how is this information exchanged between the parties involved in a IP telephony call ? – Please explain in a few words.

(c)    How do the two sets of protocols for IP telephony, SIP and H.323, compare with one another ? – Please give a short overview of their similarities and differences; consider both aspects: (1) the signaling protocol for call establishment and termination and (2) the transfer of multimedia streams.

 

Question 2: Mobile users (20 points)

(a)    Please explain in a few words the meaning of (1) device mobility, (2) user mobility, and (3) session mobility.

(b)   We consider in the following a mobile user that receives a call request. The following diagram (taken from the paper “Support for Personal and Service Mobility in Ubiquitous Computing Environments” which is part of the course notes) shows several “boxes” that are involved when a personal agent (which may reside in a PDA) determines whether an incoming call should be accepted and which QoS attributes (audio and video qualities) could be supported given the input/output devices that are locally available. Please explain in a few words, for each of the “boxes”, what it does and why it is important in this context.

 

 

 

Question 3: Optimization of QoS alternatives (20 points)

 

(a) Discussion: In the paper by Richards et al. (also copied in the paper by K. El-Khatib, X. He, G. v. Bochmann, which is part of the course notes), the authors propose the formula

for calculating the overall user satisfaction Stot from the satisfaction values si for different QoS parameters, such as frame rate, resolution, colour depth, etc.. Why did the authors not simply use the following formula, which is the simplest way to calculate an average:

 

(b)   Example application of above formula:

We consider that an MPEG video stream is transmitted from a source to the destination where it is displayed. Please use the above formula to determine which of the following two streams provide a higher user satisfaction. – Explain your reasoning or your calculations.

 

We make the following assumptions:

(1)   The user satisfaction concerning the frame rate is a linear function of the frame rate between the minimum and maximum values. It is zero for 10 frames per second (minimum acceptable value) and is 1 for 30 frames per second (ideal, maximum value).

(2)   Similarly, the user satisfaction concerning the resolution is a linear function of the number of pixels per frame between the minimum and maximum values. It is zero for 10 000 pixels (minimum acceptable value) and is 1 for 90 000 pixels (ideal value).

(3)   Frame rate and number of pixels per frame have equal weight for the determination of the overall user satisfaction.

 

Question (b i) : We assume that the following two streams are available:

            Stream (a): 11 frames per second and 90 000 pixels per frame

            Stream (b): 12 frames per second and 70 000 pixels per frame

 

Question (b ii) : We assume that the following two streams are available:

            Stream (a): 20 frames per second and 50 000 pixels per frame

            Stream (b): 15 frames per second and 70 000 pixels per frame

 

 

Question 4 : Multimedia synchronization (20 points)

We consider real-time video transmission from a video database to a receiver. The timing of the transmission of the sequence of frames by the video server is determined by its local clock. When these fames are received at the receiver, they are first stored in a play-out buffer for a short time and then presented to the user. The timing of the presentation of the frames is determined by the local clock at the receiver.

(a)    The sender and receiver clocks may not be synchronized exactly (e.g. one clock may indicate 5:30 PM while the other indicates 5:32 PM), and may not run at exactly the same speed. Explain in a few words what impact bad synchronization and different clock speeds may have on the video presentation and how the system could still present a high-quality video presentation to the user.

(b)   Let us now assume that voice and video for the same film come from different servers (we may have chosen a different language for the voice part). Again, explain in a few words what impact the bad synchronization and different speeds of the three clocks may have on the presentation and how the system could cope with this situation. – Can the method explained for point (a) be used twice, once for voice transmission and (independently) once for video transmission ? -  or are there additional difficulties?  

 

Question 5: Directories (20 points)

(a)    Please explain in a few words how the DNS and the LDAP directories are organized as a distributed database. That is, how is it possible that I can get information stored in a different country by requesting this information from a local DNS or LDAP server. Explain the principle of operation and the message exchanges between the different servers involved.

(b)   What is the difference between a Java RMI diretory and a Java Jini directory ? – Explain in a few words the information stored and the search facilities available.

 

 

 


 

Question 6: Sub-questions related to the student projects (20 points)

Please answer 3 out of the following 6 sub-questions. You should not select the sub-question that relates to your own project.

 

Sub-question 6-1 on Programmable Networks : Figure A-1 shows the reference model for the P1520 standard for programmable networks. It shows 4 interfaces and 4 system layers plus the “End user applications” layer.

(a)    To which layer or interface belongs the OSPF routing protocol ? – Please explain in a few words.

(b)   To which layer or interface belongs the inquiring of the average buffer utilization of a router, which is required for congestion management ? - Please explain in a few words.

(c)    To which layer or interface belongs the function of congestion management ? – Please explain in a few words.

 

 

Sub-question 6-2 on Differentiated QoS for multicast applications: Figure A-2 shows a comparison of the feature of different QoS-aware routing algorithms for ad-hoc wireless networks.

(a)    Please explain what “BW or delay” means in the second row.

(b)   Please explain what “Re-route” and “link repair” mean in the third row.

(c)    Please explain what “flat” and “clustered” mean in the fourth row.

 

 

Sub-question 6-3 on Server load sharing with differentiated service classes: Figure A-3 shows a diagram which indicates the relationships between different system components in the case of a Web server pool for which the load balancing is handled by a so-called broker. Please explain in a few words the role of the broker and its relationships with the other system components.

 

Sub-question 6-4 on Fairness in DiffServ: Figure A-4 shows components of a DiffServ architecture.

(a)    How is “COS” defined ? and how is a “flow” defined ? – Please explain in a few words.

(b)   What is the role of the “Policer/Shaper”? – Please explain in a few words.

(c)    What does “intra-class fairness” and “inter-class fairness” mean for DiffServ ? – Please explain in a few words.

 

Sub-question 6-5 on Differentiated QoS for multicast applications

: Figure A-5 shows four diagrams corresponding to different types of multicasting.

(a)    Please explain each type of multicasting in a few words.

(b)   What is the advantages of “end-system multicast” ?

 

Sub-question 6-6 on Active Networks : Figure A-6 shows a time-sequence diagram indicating the exchange of certain messages (called “capsules”) in the context of the so-called ANTS version of Active Networks. The example shows a capsule send to Loading Node from the Previous Node which implies the execution of some code at the Loading node; however, the code is locally not available and must be loaded from the Previous Node in this example.

(a)    For what purpose did people propose the concept of “Active Networks” and what does it mean ? – Please explain in a few words.

(b)   Looking at Figure A-6, when is the loaded code executed ?

(c)    According to your imagination, what is the purpose of the last message sent by the Loading Node (which is labeled “Capsule forwarded”) ?

 

 

 

 

I hope you enjoyed the course.  I wish you nice season’s holidays!