Capita selecta programmatuur: Broadband communication
Inhoud
- 1 Samenvattingen
- 2 General information on examination
- 3 List of Questions
- 3.1 Internet support for multimedia flows
- 3.1.1 Explain the principle of and relation between a user plane and a control plane in classical telephony.
- 3.1.2 Describe 3 cases where VoIP is used.
- 3.1.3 What is SDP.
- 3.1.4 Explain the general principle of SIP and give the basic building blocks.
- 3.1.5 An example of a SIP message is given during the exam: explain.
- 3.1.6 Give 2 specific problems when the Internet is used for voice. What are (partial) solutions to these problems.
- 3.1.7 How does RTP resolve timing problems encountered when transporting voice or video over internet.
- 3.1.8 Explain the principle of a QoS aware IP router.
- 3.1.9 Explain classification, shaping, policing, queueing, scheduling, buffer management as used in a QoS aware IP router.
- 3.1.10 Explain the principle of 2 techniques used to support flow differentiation in IP networks.
- 3.1.11 Explain the operation of MPLS.
- 3.1.12 Explain the principle of VOQ (Virtual Output Queue) and why is it used.
- 3.2 Access Networks
- 3.2.1 Explain the evolution of a classical telephone access network towards a triple play enabled access network.
- 3.2.2 Explain the ADSL architecture (figure from ITU-T rec G.992.1 will be given).
- 3.2.3 Explain the principle of DMT and QAM.
- 3.2.4 Explain the basic principles of : CRC, scrambling, FEC.
- 3.2.5 Explain in detail FEC (coding and decoding).
- 3.2.6 Explain the evolution of a classical coax access network (CATV) towards a triple play enabled access network.
- 3.2.7 Explain the basic principle of MPCP in E-PON. Give 3 polling policies.
- 3.2.8 Explain DBA in E-PON.
- 3.3 Optical Networks
- 3.3.1 Explain the difference between logical and physical topology. Give an example.
- 3.3.2 Explain the principle of WDM and OTDM.
- 3.3.3 Explain the principle of dispersion, attenuation, 3R regeneration, space switching, wavelength switching, optical memory.
- 3.3.4 Give the basic structure of an OXC and OADM.
- 3.3.5 Explain the difference between an optical network with or without wavelength conversion.
- 3.3.6 Explain the principles of control planes in optical networks: static versus dynamic, dynamic overlay versus dynamic peer.
- 3.4 Mobile Networks
- 3.4.1 Explain: FDMA, TDMA, SDMA and CDMA.
- 3.4.2 Explain: TDD, FDD.
- 3.4.3 Explain: hidden terminal problem, exposed terminal problem.
- 3.4.4 Explain RTS/CTS principle.
- 3.4.5 Explain FHSS and DSSS.
- 3.4.6 Describe the basic architecture of a GSM network.
- 3.4.7 Explain roaming in GSM.
- 3.4.8 Explain the evolution towards 3G (briefly explain the different technologies/enhancements).
- 3.4.9 Explain the principle of DFWMAC-DCF using CSMA/CA using an example.
- 3.4.10 Explain the principle of Mobile IP.
- 3.4.11 Make a comparison between GSM and Mobile IP.
- 3.4.12 Explain the figure 5 from the paper “IP Micro-Mobility Protocols (Reinbold, Bonaventure)â€Â. The figure will be given.
- 3.4.13 Explain the principle of Hierarchical Mobile IP and Fast Handoff.
- 3.5 Grid Computing
- 3.5.1 Explain the evolution towards grid computing from 2 different perspectives.
- 3.5.2 Give 4 types of grids and explain the principles.
- 3.5.3 Describe the building blocks of a grid architecture.
- 3.5.4 What is grid middleware.
- 3.5.5 Explain grid scheduling.
- 3.5.6 What is application gridification. Give two examples.
- 3.5.7 What is network aware grid scheduling. When is it important.
- 3.6 Reliability of Communication Networks
- 3.6.1 Explain defect, repair, fault, failure.
- 3.6.2 Explain: protection and restoration, dedicated and shared protection, recovery scope.
- 3.6.3 How does the classical TCP/IP protocol stack cope with network element failures?
- 3.6.4 Explain the principle of facility backup in MPLS.
- 3.6.5 Explain 1+1, 1:1, 1:1 with preemption and OMS-SPRing (in optical networks)
- 3.6.6 Explain single layer recovery in the IP/MPLS over OTN case.
- 3.6.7 What are secondary failures and what is the impact on single layer recovery.
- 3.6.8 Explain the problem of uncoordinated multi-layer recovery. Give a solution.
- 3.6.9 Explain Static versus Dynamic multilayer recovery.
- 3.7 Afkortingen
- 3.1 Internet support for multimedia flows
Samenvattingen
Klik hier om de samenvattingen te bekijken
General information on examination
Broadband Communication
Francqui KULeuven 2005-2006
by Piet Demeester
Material for the examination
All the lectures
- Inaugural lecture
- Internet support for multimedia flows
- Access Networks
- Optical Networks
- Mobile Networks
- Grid Computing
- Reliability of Communication Networks
Publications
- Internet support for multimedia flows:
- The Session Initiation Protocol: Internet-Centric Signaling (Schulzrinne, Rosenberg)
- On the building blocks of quality of service in heterogeneous IP networks(Soldatos, Vayias, Karmentzas)
- Access Networks:
- Media Access Control for Ethernet Passive Optical Networks: An Overview(Zheng, Mauftah)
- Mobile Networks:
- IP Micro-Mobility Protocols (Reinbold, Bonaventure)
- Grid Computing:
- A Gentle Introduction to Grid Computing and Technologies (Buya, Venugopal)
- Reliability of Communication Networks
- Benefits of GMPLS for Multilayer Recovery (Puype et.al.)
Examination
- Closed book exam: 70% from list of questions (see below), 30% not from list
- Open book (slides, notes, publications allowed): e.g. questions that link different subjects together or small exercises
- Evaluation: roughly 70% closed book, 30% open book
- Duration: 3 hours
- Illustrate your answers with clear figures (when appropriate)
List of Questions
- The whole question or a part of a question may be asked
- Some questions are related to the publications that are listed above (they are indicated in italic)
Internet support for multimedia flows
Explain the principle of and relation between a user plane and a control plane in classical telephony.
slides 5-10
- User plane: talk, send voice data
- Control plane: administration of the network, eg. Setting up the connection, routing, ... : overlay
Describe 3 cases where VoIP is used.
slides 15-17
- PSTN to IP (eg. Calling a skype user)
- VoIP on the Intranet: Local phone use, external is sent to PSTN
- IP in the core (backbone) of Telephony providers, also carrying voice
What is SDP.
slides 19-22
Explain the general principle of SIP and give the basic building blocks.
slides 24-27
An example of a SIP message is given during the exam: explain.
slides 33-39
Give 2 specific problems when the Internet is used for voice. What are (partial) solutions to these problems.
Problems
- packet loss (43)
- jitter (44-45) [in the extreme: generates packet loss]
Solutions: (Notes on slide 42)
- error recovery: FEC, interleaving, adding sequence numbers, adding timestamps
Solutions on receiver side: (Notes on slide 42)
- Dejitter buffer
- Packet concealment: (fill with noice or interpolation)
Other information...
- Other problems: link delay (46)
- transmission delay --> use faster access lines (higher bandwidth)
- propagation delay (lightspeed) --> use shorter, faster path (transatlantic fiber vs. satellite)
Terminal actions 49
- CODEC 50-60
- Timing: RTP 61-63
- de-jitter buffer 64
- packet concealment 65
Network actions 66-68
- QoS Router 69-82
- Network coordination 84-96
How does RTP resolve timing problems encountered when transporting voice or video over internet.
slides 61-63 +64 +65 ! Lees ook de notes
RealTime mijn poep, dit is niet real-time as you probably know it: Steekt extra info in packet (type, seq. nr, timestamp & conn ident.) Het hele idee is dat UDP packet nrs te weinig zijn om iets te kunnen doen (eg dejitter of packet concealment).
==> in deze context Real-Time te lezen als: het gaat over gelijklopende klokken, niet over logische volgorde.
Met die extra info kan er op de client side 'slim' omgesprongen worden met de UDP packets (vb: als packages om de 160, en 400 lang geen packet meer ontvangen; jitter/loss of niet ? volgend package heeft SQ+1 en TS+400: geen jitter maar de bedoeling).
Dankzij die extra RTP info is Dejitter en Packet Concealment mogelijk.
==> inderdaad. De RTP informatie is nodig om de "tijds-synchronisatie" van de pakketten juist te kunnen krijgen. (De volgnummers van UDP geven enkel informatie over de logische volgorde.) Een extra voordeel is dat pakketten die enkel stilte bevatten expliciet niet uitgezonden hoeven te worden, wat minder netwerkbelasting veroorzaakt. (die pakketten worden eigenlijk niet geconcealed, want via de UDP nummers weet de ontvanger dat er daartussen inderdaad een stilte was)
Explain the principle of a QoS aware IP router.
slides 69-82
Explain classification, shaping, policing, queueing, scheduling, buffer management as used in a QoS aware IP router.
slides 69-82
- classification (70) :
- identification of the flows (voice, file transfer, ...) on entering the buffers
- enable a different treatment for different flows
- identifiers:
- source/destination IP address (+ ports)
- dedicated classifier / label
- policing (71): "the process of discarding packets (by a dropper) within a traffic stream in accordance with the state of a corresponding meter enforcing a traffic profile."
- Why?
- Control incoming flow against certain specification
- How?
- e.g. Token Bucket Algorithm (72-73)
- discard | mark packet | make best-effort packet (= worst effort)
- Why?
- shaping (74): "the process of delaying packets within a traffic stream to cause it to conform to some defined traffic profile."
- Why?
- Obtain a smooth flow of packets( reduce burst size)
- How?
- e.g. Token Bucket Algorithm (75-76)
- Why?
- queuing (77)
- one queue for all flows
- one queue for one class of flows
- one queue for each individual flow
- scheduling
- priority (81)
- easy to implement
- high priority queues have very low delay
- low priority queues can have very bad behavior (starvation)
- weighted round robin (82) (WFQ)
- easy to implement
- different bandwidths possible
- also low priority traffic is served
- can become very complex with many queues
- priority (81)
- buffer management (79-80)
- When should a packet be dropped?
- Which packet should be dropped?
- RED
- W-RED
Explain the principle of 2 techniques used to support flow differentiation in IP networks.
slides 84-88
- IntServ (Integrated Services)
- DiffServ (Differntiated Services)
Explain the operation of MPLS.
slides 92-96
Explain the principle of VOQ (Virtual Output Queue) and why is it used.
Paper: On the building blocks of quality service in heterogeneous IP networks p. 78
3.5.1 Multi-Stage Queuing and Scheduling
Most research efforts on scheduling assume that the network nodes (routers/switches) are based on an output queuing architecture; i.e. when a packet arrives at an input port, it is transferred as fast as possible to the buffer of the corresponding output port. This means that for a node with N ports, output buffer memory should be accessible at a rate N times the maximum line rate, so as to avoid contention. It is evident that such an architecture, though simple and efficient, exhausts the capabilities of packet memory, when the line rates reach the gigabit scale. To avoid such a scalability constraint, a node may queue the packets at the input ports only. The drawback of input queuing (assuming FIFO queues) is that, at a given moment, only one packet from one of the N input ports can be switched to a specific output port. This means that in case the first packets of two input queues are destined for the same output port, one of them will have to wait, thus possibly delaying the next packet in its queue, although this may be going to a different (and available at that moment) output port. This is called Head-of Line blocking (HOL).
To limit the HOL effect of input queuing, the Virtual Output Queuing (VOQ) architecture was proposed. In this technique, each input port maintains a separate queue for each output port. One key factor in achieving high performance using VOQ is the scheduling algorithm, which is responsible for the selection of packets to be transmitted in each time unit from the input queues (or VOQs) to the output lines . This algorithm has to retrieve state of all N2 input queues, compute a (pseudo-)optimum matching, and perform the switching accordingly, all within one cycle. In addition, the scheme must arbitrate fairly among inputs and outputs and not cause starvation of any queue. Several algorithms, such as parallel iterative matching (PIM), iSLIP, Oldest Queue First (OQF), Longest Queue First (LQF), have been proposed in the literature (see [42], [43]). It was shown that with as few as four iterations of the above iterative scheduling algorithms, the throughput of the switch exceeds 99%. In view of the limitations of the two "pure" architectures, most recent proposals strive to combine the performance of output queuing with the scalability of VOQ-based input queuing. Such an architecture distributes the scheduling process to the input-port schedulers, which need only resolve contention at an input port (and not across input ports like in a VOQ-only node) and to the to output port schedulers, which perform classical output queuing schemes. This is depicted in Figure 4.
An important issue in such multi-stage queuing nodes is the application of scheduling schemes for QoS support. The difference between guaranteeing the QoS in a multi-stage queuing node and doing so in an output-queuing node is mainly due to the fact that, in the latter case, scheduling of packets enqueued in different outputs can be isolated from one another. However, in a multi-stage queuing node some packets may not be promptly scheduled across the switching fabric by the VOQ scheduler. Consequently, they may lose their chances of being serviced in time, which will result in violating their QoS.
Therefore, the key point for providing QoS guarantees in a VOQ node is to design a scheduling algorithm which can guarantee that queued packets are transmitted across the switch fabric promptly according to their QoS requirements. If the delays of queuing packets can be guaranteed, the employed scheduling algorithm will not lead to "starvation" for queued packets at any port. A number of algorithms using different methods to solve this problem have been proposed (e.g., [42]). Basically, these schemes are based on dividing each VOQ to different sub-queues per flow or per class of service (CoS). Their implementation difficulty has probably been one of the main reasons to move from IntServ-like guarantees to more qualitative guarantees. At the output queues, classical scheduling schemes can be applied.
In fact, many current high capacity routers have adopted such architectures combining VOQ and output queuing with a non-blocking switch-fabric; e.g. the Cisco GSR router uses a Modified Deficit Round Robin scheme for servicing the per-CoS VOQs as well as the per-CoS output queues.
Access Networks
Media:NotitiesPresentatie2.pdf
Explain the evolution of a classical telephone access network towards a triple play enabled access network.
Triple play: telephone, tv, internet 19
Explain the ADSL architecture (figure from ITU-T rec G.992.1 will be given).
slides 30-35
Explain the principle of DMT and QAM.
- Discrete Multi Tone 39-42
- Way of using the available bandwidth, by splitting it up in descrete frequency (tone) bands. For each discrete band, the SNR is being determined.
- Quadrature Amplitude Modulation 43-46
- Using the bands of DMT by means of complex modulation technique, where amplitude and phase are used to encode the information. Depending on the SNR of the individual bands, a more complex modulation can be used, (more information per frequency band).
Explain the basic principles of : CRC, scrambling, FEC.
- Cyclic Redundancy Check 59-62
- remainder after division is being added to actual data, to be able to detect errors.
- CRC's for each buffer (fast & interleaved) are transmitted in the following superframe (1st frame)
- receiver checks the actual data for corruption, by adding the checksum, and checking whether the remainder after division is zero (= no errors)
- Scrambling 63-64
- long sequences of 0's or 1's is undesirable. (reason? : synchronisation?)
- scrambling is an operation on the input, which makes it less likely long sequences of 0's or 1's will be transmitted.
- "Scrambling . . . eliminates repetitious data patterns such as all 0's or 1's. Random bit patterns reduce both radio frequency interference and signal crosstalk from one pair to another pair in the cable."
- Forward error correction 65-69
- forward error correction adds information to the data, so that after tranmission, on the receiving side (limited) transmission errors can be corrected. e.g. use Reed-Solomon convolutional encoder
- Interleaving (70) only on interleaved buffer (is only difference with fast buffer)
- weaving bits/frames such that when a burst of data is lost, the recovery algoritm can restore the corrupt transmission by means of the FEC-codes. (::Reduce the locality of data, most loss happens in bursts)
Explain in detail FEC (coding and decoding).
- Forward error correction 65-69
Explain the evolution of a classical coax access network (CATV) towards a triple play enabled access network.
slides 80-83
- Analog to Digital (some part of the spectrum)
- Make the channel bidirection
- split spectrum in up and down channels
- Decrease Distance of the Coax
- to avoid attenuation problems
- use fiber to extend
- Improve Bandwidth usage
- implement advanced modulation techniques
- Connect to a backbone network
- add switching/routing (in Head End)
Explain the basic principle of MPCP in E-PON. Give 3 polling policies.
- Ethernet-Passive Optical Network 96-105
- Passive Optical Networks or PONs are point-to-multipoint connections, made up of fibre optic cabling, of passive splitters and couplers that distribute an optical signal through a branched "tree" topology to connectors that terminate each fibre segment. For PON architectures, there is the choice between either Ethernet or ATM.
- EPON or Ethernet PON (IEEE 802.3ah). Recently, the possibility of a 10-Gb/s EPON standard (known as next generation EPON or NGEPON) is being proposed to the IEEE.
- Multi Point Control Protocol 106-108
- The IEEE 802.3ah Task Force is developing the so-called multipoint control protocol (MPCP), which arbitrates channel access among central office and subscribers. MPCP is used to dynamically assign the upstream bandwidth (subscriber to service provider), which is the key challenge in access protocol design for EPONs. Note that MPCP does not specify any particular dynamic bandwidth allocation (DBA) algorithm. Instead, it is intended to facilitate the implementation of DBA algorithms. The MPCP arbitration mechanism is used to dynamically assign nonoverlapping upstream transmission windows (time slots) to each ONU. Besides auto-discovery, registration, and ranging (RTT computation) operations for newly added ONUs, MPCP provides the signaling infrastructure (control plane) for coordinating data transmissions from the ONUs to the OLT. The basic idea is that the upstream bandwidth is divided into bandwidth units via TDM. These units are assigned to the ONUs as determined by the OLT according to the DBA algorithm in use. The OLT has control over the assignment of these units of bandwidth. These units can be assigned on the fly as needed or can be reserved in advance. For efficiency reasons, any reserved units or fraction of units of bandwidth that go unused can in general be re-assigned on the fly by the OLT to other ONUs that could make use of it.
- MPCP uses two types of messages to facilitate arbitration: REPORT and GATE. Each ONU has a set of queues, possibly prioritized, holding Ethernet frames ready for upstream transmission to the OLT. The REPORT message is used by an ONU to report bandwidth requirements (typically in the form of queue occupancies) to the OLT. A REPORT message can support the reporting of up to 13 queue occupancies of the corresponding ONU. Upon receiving a REPORT message, the OLT passes it to the DBA algorithm module. The DBA module calculates the upstream transmission schedule of all ONUs such that channel collisions are avoided. After executing the DBA algorithm, the OLT transmits GATE messages to issue transmission grants. Each GATE message can support up to four transmission grants. Each transmission grant contains the transmission start time and transmission length of the corresponding ONU. Each ONU updates its local clock using the timestamp contained in each received transmission grant. Thus, each ONU is able to acquire and maintain global synchronization. The transmission start time is expressed as an absolute timestamp according to this global synchronization. Each ONU sends backlogged Ethernet frames during its granted transmission window using its local intra-ONU scheduler. The intra-ONU scheduler schedules the packet transmission from the various local queues. The transmission window may comprise multiple Ethernet frames; packet fragmentation is not allowed. As a consequence, if the next frame does not fit into the current transmission window, it has to be deferred to the next granted transmission window.
- Polling policy: Interleaved Polling with Adaptive Cyclic Time 110-111
Paper: Media Access Control for Ethernet Passive Optical Networks: An Overview
- E-PON p. 145-146
- MPCP p. 147
- Polling policies p. 148
- Poll-and-stop polling
- Interleaved polling
- Interleaved polling with stop
Explain DBA in E-PON.
Dynamic Bandwidth Allocation 109
Paper: Media Access Control for Ethernet Passive Optical Networks: An Overview
- DBA p. 149
Optical Networks
- Optical Packet Switching is not part of the examination subject matter
Useful Fiber Optic glossary: http://www.fiber-optics.info/glossary-a.htm
Explain the difference between logical and physical topology. Give an example.
slides 5
- logical: transport laag ziet een logisch volledig verbonden net,
- physical: maar fysiek zijn er minder verbindingen: bijvoorbeeld een ring of (niet-volledig) net.
=> het logische netwerk is een "overlay" over het fysieke, de topologieën kunnen verschillen
A logical topology is how devices appear connected to the user. A physical topology is how they are actually interconnected with wires and cables.
Explain the principle of WDM and OTDM.
slides 9-17
- Wavelength Division Multiplexing (http://en.wikipedia.org/wiki/Wavelength-division_multiplexing)
- Optical Time Division Multiplexing (http://en.wikipedia.org/wiki/Time-division_multiplexing)
http://www.ist-optimist.org/pdf/trends/TERENA_2003/sld005.htm
Wavelength Division Multiplexing (WDM) and Optical Time Division Multiplexing (OTDM) are the two multiplexing techniques to increase the bit rate transmitted over a fibre above the bit rates which can be generated electronically using Electric Time Division Multiplexing. The limit for ETDM is now 40 Gbit/s and might be shifted up to 80 Gbit/s in the future.
- WDM modulates the electric signals on optical carriers with different wavelengths. The spacing of the wavelengths in practical systems is usually 100 GHz and the bit rate per channel 10 Gbit/s, which as mentioned will be increased to 40 or even 80 Gbit/s. The total bit rate per fibre is limited by the useable spectrum, which in turn is limited by the bandwidth of the optical amplifiers used. The bandwidth of the C-band EDFA is about 4 THz. The band width can be increased by using other amplifiers (L-Band, Raman and hybrid solutions).
- The total bit rate can also be increased by increasing the spectral efficiency, the ratio between single channel bitrate and the channel spacing. It might be possible to increase the spectral efficiency well above 1 bit/s/Hz by using special modulation and coding techniques.
- In laboratory systems a total bit rate of more than 10 Tbit/s has already been demonstrated.
--> Verschillende "channels" over een enkele fiber worden gescheiden door gebruik te maken van verschillende golflengtes licht: verschillende kleuren lasers. (principe van een prisma) Elke kleur kan een aparte gegevensstroom transporteren. De kleuren worden bij ontvangst terug gescheiden, en de verschillende gegevensstromen worden er op die manier uitgehaald.
- Optical Time Division Multiplexing instead uses the potential to generate very narrow optical pulses. These are multiplexed by using suitable delay lines. One major challenge is the demultiplexer which needs operate at speeds of the multiplexed bit rate. 1.28 Tbit/s have already been achieved, using this technique [M. Nakazawa et al.: 1.28 Tbit/s – 70 km OTDM transmission using third- and fourth-order simultaneous dispersion compensation with a phase modulator, PD 2.6, ECOC 2000]. It might be hard to overcome this result, due to problems with spectral and polarisation mode dispersion (PMD) at these high bit rates.
--> Verschillende "channels" over een enkele fiber worden gescheiden door de tijd: elk 'tijdsslot' wordt opgedeeld in het aantal channels, en er wordt voor elk kanaal een deel van de tijd gebruikt om bits op het kanaal te zetten. Hiervoor moet een verschillende kleine delay geïntroduceerd worden voor elk kanaal. Dit kan gebeuren door voor elke kanaal een stuk "wacht-fiber" te maken van verschillende lengte. Dit is eigenlijk hetzelfde principe als slide 6 van lecture 1, bij de PSTN, maar hier wordt dus gebruik gemaakt van optische signalen, ipv elektrische.
Explain the principle of dispersion, attenuation, 3R regeneration, space switching, wavelength switching, optical memory.
slides 19-25
- Dispersion occurs when the light traveling down a fiber optic cable “spreads out,†becomes longer in wavelength and eventually dissipates. Two other major mechanisms of attenuation in optical fibers are absorption and scattering.
- Omdat gebruik gemaakt wordt van een "band" golflengtes bestaat een enkele puls over de fiber eigenlijk uit klein continu spectrum aan golflengtes. Een eigenschap van licht is dat elke golflengte een iets andere snelheid heeft. (langer golflengtes verplaatsen zich trager) Hierdoor zal een enkele puls na versturen "verstrooid" zijn. De langere golflengtes van de puls komen later aan dan de korte: de pulsen worden "breder", duren langer in tijd. Wanneer de pulsen kort op elkaar volgen vormt dit een probleem, omdat overlappingen kunnen optreden aan de ontvangerkant, die er bij het versturen niet waren. Een oplossing is het invoeren van een speciaal stuk materiaal dat de dispertie ongedaan maakt. (introduceert op een korte afstand een groter snelheidsverschil, in de omgekeerde zin).
- Merk op dat dispersion alleen een probleem vormt bij zeer hoge bitrates.
- Dispersion, absorption, and scattering are the three properties of optical fibers that cause attenuation, or a marked decrease in transmitted power, and therefore, have limited progress in areas of high-speed transmission and signal efficiency over long distances.
- Het (optische) signaal verzwakt over afstand. Bij typisch gebruikte golflengtes: 0.2dB/km (25THz). Het probleem is dat analoge optische versterkers (werken met een pump-laser) het signaal niet voor elke golflengte gelijk versterken. Wanneer vele opeenvolgende versterkers van dit type in serie (cascade) gebruikt worden, groeit het verschil tussen de zwak en sterk versterkte signalen. Bovendien werken de huidige versterkers slechts op een beperkt deel van het spectrum. (+/- 1520 tot 1560 nm)
- 3R regeneration: http://www.ohub.net/terminology.php?letter=all&id=7
- [copy-paste]3R regeneration is necessary after an optical lightpath transmits for a long distance. This is because optical fiber transmission system is not ideal, affected by many factors such as dispersion, EDFA noise, non-linear effects, and crosstalk, etc, which may eventually degrade an optical signal to be unrecognizable at the receiver if the signal is not relayed.
- OEO regeneration3R regeneration relays or regenerates optical signals in three domains including power, shape and time. Optical-Electronic-Optical (OEO) conversion is so far the most popular and mature technique for this purpose. The fundamental principle of OEO regeneration is to convert an optical signal into electronic format first so that the time and shape are restored, and then use the electronic signal to modulate an optical laser to generate a new optical signal.
- Besides the OEO technique, it is also possible to carry out 3R regeneration in all-optical domain without converting optical signal into electronic signal. The advantage of all-optical 3R regeneration is its bit-rate transparency, without a bottleneck from electronic modulation. However, currently the all-optical 3R regeneration technique is not mature and very expensive.
- 3R regeneration can occur in two fashion, including (1) inline 3R regeneration, and (2) in-node regeneration. Inline 3R regeneration is implemented in an ultra-long-haul (ULH) optical system, in which the physical distance between two end points of the optical system exceeds a maximum transmission reach before 3R regeneration is required. In contrast, in-node regeneration occurs in an Optical Cross-Connect (OXC) node, in which some OEO transponders are deployed for the regeneration purpose. An OXC node with full OEO regeneration capability is called Opaque OXC node, an OXC node with partial OEO regeneration capability is called translucent OXC node, while an OXC node without any OEO regeneration capability is called transparent OXC node. In general, in-node 3R regeneration is more economical since OEO regenerators are shared by all the incident links to the node, while inline 3R regeneration is more expensive for the regenerators are dedicated to a certain point-to-point ULH system.[/copy-paste]
- De 3R staat voor:
- Re-amplification, (bv EFDA: versterkt en introduceert noise)/ Opmerking: volgens mij introduceert dit geen ruis, het versterkt alleen de al aanwezige ruis:: Gelieve hier in de discussion pagina verder op in te gaan.
- Re-shaping en (bv AOWC: All Optical Wavelenght conversion)
- Re-timing (cursus) / Re-transmit (google): in fase brengen / opnieuw optical signaal genereren
- Space Switching
- MEMS (Micro ElectroMechanical Systems) werken met behulp van mechanische schakel-elementen (spiegeltjes)
- Zijn traag: (ms - domein), en worden daardoor vooral gebruikt om aan statische configuratie van paden in de fysische laag te doen. Deze systemen kunnen verschillende golflengtes op verschillende fibers schakelen. Het veranderen van deze "licht-kleur-paden" gebeurt niet zo frequent (omdat het momenteel nog traag is)... Men kan dus bijvoorbeeld bij breuk van een lijn, een kleur via een ander pad sturen (als die kleur op dat pad nog vrij is!)
- Wavelength Switching
- Er wordt gebruik gemaakt van een AOWC (all optical wavelength converter)
- Hierbij wordt de input van een bepaalde kleur laser omgezet naar een output van andere kleur (kleur van output kan gekozen worden). Als hiervan gebruik gemaakt wordt (in combinatie met space switch) kan er veel dynamischer geschakeld worden. Bijvoorbeeld om een lijnbreuk op te vangen, zijn nu alle vrije frequenties van de overige fibers bruikbaar om de data van de gebroken fiber op te vangen.
- Optical Memory
- Om een datastroom op te slaan, wordt gebruik gemaakt van een lange optische fiber waar de pakketjes op blijven rondgaan, tot ze terug uitgestuurd kunnen worden. De lengte van de "geheugen-fiber" bepaalt de grootte van het geheugen. Wegens attenuatie van het signaal kan deze lus niet onbeperkt blijven bestaan zonder versterkers en 3R-signaal correctie toe te passen. Wanneer hiervoor gebruik gemaakt moet worden van OEO apperatuur, kan beter gekozen worden om de data elektr(on)isch op te slaan.
Give the basic structure of an OXC and OADM.
slides 27-31
Er wordt een virtuele topologie opgezet door middel van de indivituele golflengtes te gebruiken op de glasvezel. (on demand high bandwidth: flexibel). Deze schakelelementen worden gebruikt om circuits op te zetten. (Niet switchen op het niveau van "data-link-frames", maar op basis van een gegevensstroom (grotere granulariteit))
- Optical Cross Connect (op basis van WDM)
- heeft "tributary" poorten (verbinding met de netwerklaag (IP/MPLS))
- heeft "aggregate fiber" poorten (verbinding met andere OXC's)
- twee soorten: met of zonder wavelength conversion
- Optical Add-Drop Multiplexer (http://en.wikipedia.org/wiki/Add-drop_multiplexer & http://en.wikipedia.org/wiki/ROADM)
- eigenlijk gelijkaardig aan een OXC. (in ring-opstelling met slechts 2 "aggregate fiber"-poorten
- ADD: toevoegen van een gegevenspad
- TRANSIT: doorsturen van de gegevens naar de andere aggregate fiberpoort
- DROP: einde van het gegevenspad: gegevens worden afgeleverd aan de netwerklaag
- eigenlijk gelijkaardig aan een OXC. (in ring-opstelling met slechts 2 "aggregate fiber"-poorten
Explain the difference between an optical network with or without wavelength conversion.
slides 32-33
- wanneer er geen golflengteomzetting gebruikt wordt, (enkel space switching) zal de fysische laag uit een aantal gescheiden "vlakken" bestaan die elk hun eigen kleur gebruiken. Er kan enkel via elektronische omzetting van vlak gewisseld worden. Dit is dus niet zo dynamisch: het kan zijn dat een kleur-vlak overbelast geraakt, terwijl andere golflengtes op dezelfde glasvezels ongebruikt zijn.
- wanneer er wel golflengteomzetting gebruikt wordt, kan op het niveau van de fysische laag van kleurvlak veranderd worden (zonder hierbij gebruik te maken van OEO-elementen). Dit is een veel dynamischere situatie.
Explain the principles of control planes in optical networks: static versus dynamic, dynamic overlay versus dynamic peer.
slides 35-39
- static vs dynamic
- static: het transport netwerk wordt statisch en centraal beheerd (telco style). De gebruikers hebben geen invloed op de configuratie van de lichtpaden. Alle informatie over congestie van bepaalde lijnen komt samen in het NOC (Network Operating Center) waar een operator statisch bepaalde veranderingen kan doorvoeren.
- dynamic: Optical Control Plane Signaling Model: Gebruikers-machines configureren zichzelf en veranderen dynamisch welk lichtpad bepaalde trafiek moet volgen. (hoe dit gebeurt: next bullet)
- dynamic overlay vs dynamic peer
- overlay: (client-server style) er is geen automatische informatie-uitwisseling tussen netwerk en data-link laag.
- peer: Er is actieve samenwerking tussen netwerk (ip) en data-link (OTN) laag. Op die manier kan er een betere configuratie van lichtpaden gebeuren.
http://www.ist-optimist.org/pdf/workshops/ECOC2000/Dixit_Nokia.pdf (slide 13 en verder)
Mobile Networks
Explain: FDMA, TDMA, SDMA and CDMA.
slides 6-14
- Frequency Devision Multiple Access (http://en.wikipedia.org/wiki/Frequency-division_multiple_access)
- Time Devision Multiple Access (http://en.wikipedia.org/wiki/Time-division_multiple_access)
- Space Devision Multiple Access (http://en.wikipedia.org/wiki/Space_division_multiple_access)
- Code Devision Multiple Access (http://en.wikipedia.org/wiki/Code_division_multiple_access)
Explain: TDD, FDD.
slides 15 (http://en.wikipedia.org/wiki/Time_division_duplex)
- Time Division Duplex
- Time division duplex (TDD) is the application of time-division multiple access to separate outward and return signals. Time division duplex has a strong advantage in the case where the asymmetry of the uplink and downlink data speed is variable. As the amount of uplink data increases, more bandwidth can be allocated to that and as it shrinks it can be taken away. Another advantage is that the uplink and downlink radio paths are likely to be very similar in the case of a slow moving system. This means that techniques such as beamforming work well with TDD systems.
- Frequency Division Duplex
- Frequency division duplex (FDD) is the application of frequency-division multiple access to separate outward and return signals. The uplink and downlink sub-bands are said to be separated by the "frequency offset". Frequency division duplex is much more efficient in the case of symmetric traffic. In this case TDD tends to waste bandwidth during switchover from transmit to receive, has greater inherent latency, and may require more complex, more power-hungry circuitry.
Another advantage of FDD is that it makes radio planning easier and more efficient since base stations do not "hear" each other (as they transmit and receive in different sub-bands) and therefore will normally not interfere each other. With TDD systems, care must be taken to keep guard bands between neighboring base stations (which decreases spectral efficiency) or to synchronize base stations so they will transmit and receive at the same time (which increases network complexity and therefore cost, and reduces bandwidth allocation flexibility as all base stations and sectors will be forced to use the same uplink/downlink ratio)
slides 16-22
- Hidden terminal (node) problem (http://en.wikipedia.org/wiki/Hidden_node_problem)
- Hidden nodes in a wireless network refer to nodes which are out of range of other nodes or a collection for nodes. Take a physical star topology with an Access Point with many nodes surrounding it in a circular fashion; each node is within communication range of the Access Point, however, not each node can communicate with each other. For example, it is likely that the node at the far edge of the circle can see the access point, which is known as r, but it is unlikely that the same node can see a node on the opposite end of the circle, 2r (or simply the diameter). These nodes are known as hidden. The problem is when node r and r2 start to send packets simultaneously to the access point. Since node r and r2 can not sense the carrier, Carrier sense multiple access with collision avoidance (CSMA/CA) does not work.
- Exposed terminal (node) problem (http://en.wikipedia.org/wiki/Exposed_terminal_problem)
- In wireless networks, the exposed node problem occurs when a node is prevented from sending packets to other nodes due to a neighboring transmitter. Consider an example of 4 nodes labeled R1, S1, S2, and R2, where the two receivers are out of range from one another, yet the two transmitters in the middle are in range of each other and one of the receivers. Here, if a transmission between S1 and R1 is taking place, node S2 is prevented from transmitting to R2 as it concludes that it will interfere with the transmission by its neighbor S1. In order to increase the throughput, the exposed node S2 should be allowed to transmit in a controlled fashion without interfering with the on-going transmission between S1 and R1.
- Oplossing: gebruik MACA (Multiple Access/Collision Avoidance)
- The solution is the use of the IEEE 802.11 RTS/CTS mechanism. When a node hears an RTS from a neighboring node, but not the corresponding CTS, that node can deduce that it is an exposed node and is permitted to transmit to other neighboring nodes.
Explain RTS/CTS principle.
slides 20-21 (http://en.wikipedia.org/wiki/IEEE_802.11_RTS/CTS)
- Request To Send
- Clear To Send
Protocol Description
A node wishing to send data initiates the process by sending a Request to Send frame (RTS). The destination node replies with a Clear To Send frame (CTS). Any other node receiving the CTS frame should refrain from sending data for a given time (solving the hidden node problem). The amount of time the node should wait before trying to get access to the medium is included in both the RTS and the CTS frame. Any other node receiving the RTS frame but not the CTS frame is permitted to transmit to other neighboring nodes (solving the exposed node problem). This protocol was designed under the assumption that all nodes have the same transmission range.
RTS/CTS is an additional method to implement virtual carrier sensing in Carrier sense multiple access with collision avoidance (CSMA/CA). By default, 802.11 relies on physical carrier sensing only which is known to suffer from the hidden node problem.
RTS/CTS packet size threshold is 0-2347. Typically, sending RTS frames is turned off by default (threshold >=2347). If the packet size the node wants to transmit is larger than the threshold, the RTS/CTS handshake gets triggered. If the packet size is equal to or less than threshold the data frame gets sent immediately.
Another situation where terminal A and B trying to send C, where C and B are very close while A is far but still in the range, Near/Far Terminal problem will occur. Since power decay propotional to distance square, B drowns out A's signal (at physical layer), so C cannot receive A. Thus power control is required.
Explain FHSS and DSSS.
slides 23-25
- Frequency Hopping Spread Spectrum (http://en.wikipedia.org/wiki/Frequency-hopping_spread_spectrum)
- Fast Hopping
- Slow Hopping
- Direct Sequence Spread Spectrum (http://en.wikipedia.org/wiki/DSSS)
- Type of modulation ~ CDMA
- but: single code + ALOHA
Describe the basic architecture of a GSM network.
slides 28-31 (http://en.wikipedia.org/wiki/Network_Switching_Subsystem : zeer goeie en relevante pagina!)
Cellen met zendmasten (Base Tranceivers): een aantal masten zijn (per kabel) verbonden (en worden gecontrolleerd) door een BSC (Base Station Controller) Die BSC's zijn verbonden met een MSC (Mobile Switching Center, ~Digital TEX) die switchen de tijdssloten in geval van GSM(-protocol). MSC is de gateway naar de rest van de wereld (POTS)
En dan natuurlijk de gebruikers met hun MS (Mobile Station, ook Handy of Mobieltje genaamd ;-), in america noemen ze dat niet voor niks Cell-phone, tov van telefooncellen in belgië).
Op grotere schaal, is er meer dan 1 MSC nodig. Die worden dan verbonden met een GMSC, die de gateway naar het POTS vormt.
Als je naar 't buitenland gaat, wil je ook kunnen bellen: daarvoor dient Roaming. Extra componenten:
- HLR: Home Location Register (Database met subscribers)
- VLR: Visiting Location Register (Database met alle actieve gebruikers (eigen en "vreemde")
- AuC: Authentication Center (Contract tussen de verschillende providers/operators)
- EIR: Equipment Identity Register
- The EIR (Equipment Identity Register) is often integrated to the HLR. The EIR keeps a list of mobile phones (identified by their IMEI) which are to be banned from the network or monitored. This is designed to allow tracking of stolen mobile phones. In theory all data about all stolen mobile phones should be distributed to all EIRs in the world through a Central EIR. It is clear, however, that there are some countries where this is not in operation. The EIR data does not have to change in real time, which means that this function can be less distributed than the function of the HLR.
Explain roaming in GSM.
slides 30
http://en.wikipedia.org/wiki/Roaming#The_roaming_process
(zie vorige)
Enige uitbreiding aan de architectuur is dat er overeenkomsten moeten zijn tussen operatoren, en dat die gekend moeten zijn in het AuC.
Explain the evolution towards 3G (briefly explain the different technologies/enhancements).
slides 42-47
- 2.5G
- HSCSD: High speed Circuit switched data (http://en.wikipedia.org/wiki/HSCSD)
- GPRS: General Packet Radio Service (http://en.wikipedia.org/wiki/GPRS)
- SGSN: Serving GPRS Support Node
- GGSN: Gateway GPRS Support Node
- (voorgaande zijn software upgrades voor de operatoren)
- EDGE: Enhanced Data rates for GSM Evolution (http://en.wikipedia.org/wiki/EDGE)
- 3G
- UMTS: Universal Mobile Telecommunications System (http://en.wikipedia.org/wiki/UMTS)
Explain the principle of DFWMAC-DCF using CSMA/CA using an example.
slides 57-59
- Distributed Foundation Wireless Medium Access Controll-Distributed Coordination Function
- Carrier Sense Multiple Access with Collision Avoidance
Explain the principle of Mobile IP.
slides 75-78 (http://en.wikipedia.org/wiki/Mobile_IP)
[copy-paste]
Introduction
Mobile IP provides an efficient, scalable mechanism for node mobility within the Internet. Using Mobile IP, nodes may change their point-of-attachment to the Internet without changing their IP address. This allows them to maintain transport and higher-layer connections while moving. Node mobility is realized without the need to propagate host-specific routes throughout the Internet routing fabric.
Applications
Mobile IP is most often found in wireless WAN environments where users need to carry their mobile devices across multiple LANs with different IP addresses. It may also be used in 3G networks to provide transparency when internet users migrate between cellular towers.
In many applications (VPN and VoIP, to name a few), sudden changes in network and IP-address can cause problems.
How Mobile IP Works
In brief, Mobile IP works as follows. A mobile node can have two addresses - a permanent home address and a care-of address, which is associated with the network the mobile node is visiting. There are two kinds of entities in Mobile IP:
- A home agent stores information about mobile nodes whose permanent address is in the home agent's network.
- A foreign agent stores information about mobile nodes visiting its network. Foreign agents also advertise care-of addresses, which are used by Mobile IP.
A node wanting to communicate with the mobile node uses the home address of the mobile node to send packets. These packets are intercepted by the home agent, which uses a table and tunnels the packets to the mobile node's care-of address with a new IP header, preserving the original IP header. The packets are decapsulated at the end of the tunnel to remove the added IP header and delivered to the mobile node.
The Mobile IP protocol defines the following:
- an authenticated registration procedure by which a mobile node informs its home agent(s) of its care-of address(es);
- an extension to ICMP Router Discovery, which allows mobile nodes to discover prospective home agents and foreign agents; and
- the rules for routing packets to and from mobile nodes, including the specification of one mandatory tunneling mechanism and several optional tunneling mechanisms.[/copy-paste]
Make a comparison between GSM and Mobile IP.
slides 79
Explain the principle of Hierarchical Mobile IP and Fast Handoff.
Paper IP Micro-Mobility Protocols p. 49-50
Duidelijker uitgelegd (vind ik - Kurt), mét tekening:
http://coscweb2.cosc.canterbury.ac.nz/research/reports/MastTheses/2004/mast_0403.pdf
Hierarchical: 36-37
Fast Handoff: 38
[copy-paste] 2.3 Hierarchical Mobile IP
The Hierarchical Mobile IP [14] is a proposed extension to Mobile IP which employs a hierarchy of foreign agents to locally handle Mobile IP registration as shown in Fig. 2.3.1, in order to support power-constrained operation and to reduce routing state information in the visited domain. In this protocol mobile hosts send Mobile IP registration messages to update their respective location information. Registration messages establish tunnels between neighbouring FAs along the path from the mobile node (MN) to a gateway foreign agent (GFA). Packets addressed to MNs travel in this network of tunnels, which can be viewed as a separate routing network overlay on top of IP. The use of tunnels makes it possible to employ the protocol in an IP network that carries non-mobile traffic as well. Typically one level of hierarchy is considered where all FAs are connected to the GFA. In this case direct tunnels connect the GFA to FAs that are located at access points. Paging extensions for Hierarchical Mobile IP are presented in [15] allowing the idle MNs to operate in a power saving mode while located within a paging area. The location of MNs is known to HAs and is represented by paging areas. After receiving a packet addressed to a MN located in a foreign network, the HA tunnels that packet to the paging FA, which then pages the MN to re-establishes a path toward the current point of attachment. Paging a MN can take place using a specific communication time-slot in the paging area similar to the paging channel in second generation cellular systems. Paging schemes increases the amount of time a MN can remain in a power saving mode. In this case, the MN only needs to wakeup at predefined time intervals to check for incoming packet requests.
2.5 Fast Handoff
The fast handoff proposal [17] reuses the principle of Hierarchical Mobile IP and addresses two remaining problems. These are mainly the need for a fast handoff management for real-time applications and the presence of triangular routing inside the domain. This proposal assumes that the serving FA anticipates the movement of MHs by sending multiple copies of the traffic to potential neighbour FAs. “Bicasting†is used to support data forwarding to the old and new FAs while the MH is moving between the old and new access points. Fast handoff predicts the movement of MHs through coupling with layer two functionality that is dependent on the type of access technology used. Bicasting uses simultaneous bindings, where the MHs set the “S†bit in the registration request. Depending on the networking model (i.e., flat or hierarchical model) the receiving agent (HA, GFA) will add a new binding for the MH. As in the case of proactive handoff, the fast handoff proposal also assumes that it can anticipate the movement of MHs in advance of handoff. Fast handoff completes the Mobile IP handoff prior to establishing Layer two connectivity or forwarding data. The total delay for fast handoffs is limited to the time needed to perform a layer two handoff. Fig. 2.5.1 shows the difference between fast handoff and proactive handoff.
Grid Computing
Explain the evolution towards grid computing from 2 different perspectives.
- Historical 3-16
- Network: 42
Give 4 types of grids and explain the principles.
slides 46 (use a mirror)-56
- Cycle scavenging grids
- idle-time van cpu van vrijwillige gebruikers wordt gebruikt.
- low-cost
- centrale server verdeeld work-load pakketjes (jobs)
- gebruikers laten pc rekenen, en sturen de resultaten terug
- Computational grids
- hoge verhouding
- weinig gebruikers
- dedicated grid infrastructure
- for scientific research
- centrale scheduler
- Data grids
- dedicated grid infrastructure
- dedicated network
- data-hoeveelheid is immens groot, replicatie is belangrijk (leuk voorbeeld: http://www.cleversafe.org/)
- hoge verhouding
- hoge computationele en netwerk vereisten
- aparte information/storage resources
- Service grids
- generic, through web-services
- wetenschappelijke samenwerking
- sharing of information and processing power
- QoS, realtime service support
- werkt met een "grid-service provider"
Describe the building blocks of a grid architecture.
slides 59-64
- Computational Resources
- Provides processing power
- e.g. Personal Computers, clusters, supercomputers, PDAs,...
- Storage Resources
- Provides storage space for output data
- Storage servers with hard disks, tapes, optical media
- Data Resources
- Provides input data for jobs
- e.g. databases, file servers, instruments, sensors...
- Network Resources
- Provide interconnectivity between
- resources
- management components
- users
- e.g. dedicated Grid links, personal ISP links
- Provide interconnectivity between
- Grid Portal
- User application interface (job submission)
- Administrative interface
- access to applications and resources available in virtual space
- Scheduler
- assigns resources to jobs
- different scheduling algorithms (network aware, service aware,...)
- interacts with resource monitoring components
- up-to-date resource status information
- Information / monitoring Service
- Tracks resource status
- online/offline
- property changes (available memory / diskspace / processing)
- Job / Resource Manager / Security
- Security (encryption, authentication, authorization)
- launches jobs on specific resources
- monitors the status of those jobs
- retrieves job results
- I/O data retrieving/sending, replication
- Grid Site
- Collection of local
- Resources
- Management components
- Grid portals
- Collection of local
- Grid
- Collection of interconnected Grid sites
What is grid middleware.
slides 65-66
Explain grid scheduling.
slides 70-71
What is application gridification. Give two examples.
slides 75, 82-87
Application gridification is the term given to the
process of enabling an existing application for its
execution on a grid infrastructure.
Zie bundeltje 'Six strategies for grid application enablement'
What is network aware grid scheduling. When is it important.
slides 94-96
- Zorgen dat data beschikbaar is wanneer de berekeningen moeten gebeuren. (genoeg bandbreedte alloceren om de 'pipeline' vol te houden)
- Belangrijk bij lage bandbreedte naar de grid-sites.
Reliability of Communication Networks
Explain defect, repair, fault, failure.
slides 9-15
- Network element defect
- Decrease of network element ability to perform a required function
- E.g. a link defect -> poor link quality -> error detection -> packet/frame retransmissions
- Decrease of network element ability to perform a required function
- Network element failure
- Termination of network element ability to perform a required function
- Happens at one particular moment in time
- E.g. a cable cut by an excavator
- Termination of network element ability to perform a required function
- Fault or outage
- Inability of a network element to perform a required function
- Lasts until repair of the network element
- Covers a time interval
- Inability of a network element to perform a required function
failure repair defect1 defect2 V<--------- fault ---------->V ----|---------|-----|----------------------------|-------------> time ...--operational --> <--- not operational ------> <--operational--...
- Failure in a network (slide 11)
- Planned vs. unplanned outages
- Planned: intentionally caused by operational or maintenance operations by the operator ï‚® preventive measures possible
- Unplanned outages: difficult to predict ï‚® defensive measures
- Internal vs. external causes
- Internal: caused by network-internal imperfection (e.g. design error, battery breakdown, component defect)
- External: by surrounding event (e.g. electricity breakdown, storm, earthquake, sabotage, vandalism)
- Commonly occurring failures
- Cable cuts: related to link length; between 50 and 200 days per 1000 km of cable
- Equipment failures
- Accounted or expected failure scenario
- Single-link failure
- Single-node failure
- Shared Risk (Link) Group (SRLG): group of resources affected by same failure
- Planned vs. unplanned outages
slides 17-19
(slide 19, slide 20: bovenste 3 rijen zijn protection, onderste 3 restauration)
- protection and restoration
. | protection | restoration |
---|---|---|
Resource allocation | Pre-reserved | Reserved-on-demand |
Path set-up | Pre-established Pre-qualified |
Established-on-demand |
. | very fast! | . |
- dedicated and shared protection
dedicated | shared |
---|---|
som van alle mogelijke falingen | maximum van alle falingen |
- more resources needed + higher availability + simpler to calculate |
+ less resources needed - lower availability - assumption: Single failure at a time |
- recovery scope
- RHE: Recovery Head End
- RTE: Recovery Tail End
. | local recovery | global recovery |
---|---|---|
link failure | de gebroken link op het pad wordt lokaal ge-by-passed | gebroken pad wordt globaal herberekend |
Node failure | de falende knoop wordt lokaal ge-by-passed | het pad waar de falende knoop op voorkomt wordt van begin tot einde opnieuw berekend |
. | extremely fast only single action to take (send traffic back the way it came) |
shortest way for path between end-points may take a while to calculate |
How does the classical TCP/IP protocol stack cope with network element failures?
slides 26-31
- traditioneel: (IP)
- De routingtabellen worden gevuld, nadat van elke knoop LSA-pakketten ontvangen zijn.
- faling: verschillende fases
- Detection of failing link
- Advertisement of new link-state packets
- Recalculation of shortest paths
- Updating the new routing tables
--> this can take up a lot of time (O(10s)) --> Difficult to speed up
- Emerging IP Fast ReRoute (FRR)
- Werkt met Next Hop (NH)
- (niet altijd bruikbaar => afhankelijk van topologie en gewichten)
- Wat met de hogere lagen (transport-laag)
- TCP: protocol zorgt voor betrouwbaar transport... als pakketten verloren gaan, zullen ze opnieuw uitgezonden moeten worden. Ondertussen zal op het IP-niveau het probleem opgelost moeten geraken. Wanneer dit snel genoeg kan gebeuren, merkt de Applicatie die TCP gebruikt weinig van de onderbreking.
Explain the principle of facility backup in MPLS.
slides 35-37
(meer bepaald slide 36 onderste helft, waarvoor slide 37 de details via een voorbeeld bevat)
Explain 1+1, 1:1, 1:1 with preemption and OMS-SPRing (in optical networks)
slides 39-46
- 1+1
- fast protection
- no signaling required between end-points
- high cost (duplication of wavelengths and corresponding transponders)
- 1:1
- fast protection (slower than 1+1)
- signaling required between end-points (2 actions: coördination)
- high cost (duplication of wavelengths & corresponding transponders)
- BUT: possibility for low priority pre-emptable traffic
- 1:1 met preemption:
- low-priority traffic wordt op de redundante lijn gezet... wanneer er een faling optreedt wordt dit verkeer gepreempt (vervangen door het verkeer met hoge prioriteit)
- similar : N:M protection
- N spare channels for M working channels (N<M)
- OMS-SPRing: Optical Multiplex Section-Shared Protection Ring
- different wavelengths in both directions
- possible to reuse wavelength along the ring
- locale techniek: wanneer een faling gedetecteerd wordt, wordt het verkeer gewoon op de andere richting op de glasvezel gezet. (die golflengte is daar nog vrij)
- zeer snel en eenvoudige methode
- niet optimaal, want verkeer wordt heen en weer gestuurd (lokale aanpak)
- aantal golflengtes worden niet gebruikt (zijn redundant: kunnen evt. gebruikt worden voor low-priority verkeer)
- biedt slecht bescherming tegen een single-link-failure
Explain single layer recovery in the IP/MPLS over OTN case.
slides 50-52
- Optical protection
- Large granularity -> few recovery actions
- Close to root failure
- No delay due to failure propagation
- No need to deal with complex secondary failures
- Known to be fast (at least protection)
- BUT: cannot recover from all failures
- IP-MPLS recovery
- For sure, better failure coverage
- MPLS protection (making use of pre-established backup LSPs) can also be fast
- BUT:
- Can be confronted with complex secondary failure scenarios
- Fine granularity -> many recovery actions
- During recovery increased usage of capacity -> decreased QoS
- Conclusion: combine recovery at both layers
What are secondary failures and what is the impact on single layer recovery.
slides 51-53
Explain the problem of uncoordinated multi-layer recovery. Give a solution.
slides 57-62
Explain Static versus Dynamic multilayer recovery.
slides 69
Afkortingen
- PSTN: Public Switched Telephone Network (ook POTS)
- POTS: Plain Old Telephone Service
- IP: Internet Protocol
- VoIP: Voice over IP
- QoS: Quality of Service
- UNI: User Network Interface
- NNI: Network Node Interface of Network Network Interface
- LEX: Local EXchange
- TEX: Transit EXchange
- LD: Loop Disconnected
- DTMF: Dual Tone Multi Frequency
- TS: Time Slots
- SS7: Signal System 7
- SP: Signaling Points
- SL: Signaling Links
- SSP: Service Switching Point
- STP: Signal Transfer Point
- SCP: Service Control Point
- LNP: Local Number Portability
- ITU: International Telecommunication Union
- ITU-T: ITU - Telecommunication standardization sector
- S/D: Source/Destination
- IN: Intelligent Networks
- NGN: Next Generation Networks
- SDP: Session Description Protocol
- SIP: Session Initiation Protocol
- CODEC: COder/DECoder
- GW: GateWay
- MG: Media Gateway
- SG: Signaling Gateway
- MGC: Media Gateway Controller
- GK: Gate Keeper
- BS: Billing Server
- MEGACO: Media Gateway Control protocol
- IETF: Internet Engineering Task Force
- TDM: Time Division Multiplexing
- UDP: User Datagram Protocol
- ITSP: Internet Telephony Service Provider
- RTP: Real-time (Transport) Protocol
- RTCP: RTP Control Protocol
- AVP: Audio Video Profile
- TTL: Time To Live
- PCM:
- GSM: Global System for Mobile Communications
- FR: Full Rate (zie Codec)
- HR: Half Rate (zie Codec)
- NTP: Network Time Protocol
- UA: User Agent
- UAC: User Agent Client
- UAS: User Agent Server
- HLR: Home Location Register
- TLS: Transport Layer Security
- DNS: Domain Name System
- SRV: Service Record (zie DNS)
- URI: Uniform Resource Identifier
- MIME: Multipurpose Internet Mail Extensions
- ISUP: ISDN User Part
- ISDN: Integrated Services Digital Network
- CR/LF: Carriage Return/Line Feed
- 3GPP: Third Generation Partnership Project
- IMS: IP Multimedia core network Subsystem
- FEC: Forward Error Correction
- A/D: Analog/Digital
- OSPF: Open Shortest Path First
- GEO: Geostationary Earth Orbit Satellites, Geosynchronous Earth Orbit
- MOS: Mean Opinion Score
- PESQ: Perceptual Evaluation of Speech Quality
- PSQM: Perceptual Speech Quality Measure
- PEAQ: Perceptual Evaluation for Audio quality
- SSRC: Synchronization SouRCe
- CSRC: Contributing SouRCe
- CAC: Customer Access Control
- ER: Edge Router
- CR: Core Router
- SLA: Service Level Agreement
- WFQ: Weighted Fair Queuing
- RED: Random Early Detection
- W-RED: Weighted-RED
- RIO: RED with In-Out
- DFWMAC-DCF: Distributed Foundation Wireless Medium Access Controll-Distributed Coordination Function
- CSMA/CA: Carrier Sense Multiple Access with Collision Avoidance
- IntServ: Integrated Services
- DiffServ: Differentiated Services
- BE: Best Effort
- TOS: Type Of Service
- DSCP: DiffServ Code Point
- RSVP: Resource Reservation Protocol
- PHB: Per Hop Behaviour
- EF: Expedited Forwarding
- AF: Assured Forwarding
- BB: Bandwidth Broker
- (Q-)OSPF: (QoS-)Open Shortest Path First
- MPLS: Multi-Protocol Label Switching
- LSR: Label Switched Router
- LIB: Label Information Base
- LSP: Label Switched Path
- ATM: Asynchronous Transfer Mode
- LDP: Label Distribution Protocol
- VOQ: Virtual Output Queue
- CATV: Community Antenna TeleVision
- MSC: Mobile Switching Center
- TP: Twisted Pair
- STP: Shielded TP
- UTP: Unshielded TP
- BER: Bit Error Rate
- SNR: Signal to Noise Ratio
- SDM: Space Division Multiplexing
- QAM: Quadrature Amplitude Modulation
- DSL: Digital Subscriber Line
- DSLAM: DSL Access Multiplexer
- ADSL: Asymetric DSL
- HDSL: High speed DSL
- VDSL: Very high speed DSL
- EC: Echo Cancellation
- BRAS: Broadband Remote Access Server
- ATU-R: ADSL Tranceiver Unit - Remote
- ATU-C: ADSL Tranceiver Unit - Central Office
- SM: Service Module
- NT#: Network Termination (number)
- PABX: Private Automatic Branch eXchange
- NTR: Network Timing Reference
- OAM: Operation Administration and Maintenance
- EOC: Embedded Operations Channel
- AOC: ADSL Overhad Control Channel
- Cell TC: Cell Transmission Convergence (ATM-Cells ~ Ethernet-Frame)
- CRC: Cyclic Redundancy Check
- IDFT: Inverse Discrete Fourier Transform
- DAC: Digital Analog Conversion
- PHY: Physical Layer
- ONU: Optical Network Unit
- TE: Traffic Engineering
- FFT: Fast Fourier Transform
- LSB: Least Significant Bit (die van de 1, meestal meest rechtse)
- EOC: Embedded Operations Channel
- FDM: Frequency Division Multiplexing
- HE: Head End
- VHF: Very High Frequency (band)
- UHF: Ultra High Frequency (band)
- VOD: Video On Demand (?)
- HFC: Hybrid Fiber Coax
- iDTV: interactive Digital TeleVision
- DOCSIS: Data Over Cable Service Interface Specification
- O/E: Opto-Electronic
- CMTS: Cable Modem Termination System
- CM: Cable Modem
- USB: Universal Serial Bus
- PCI: Peripheral Component Interface
- WiFi: Wireless Fidelity
- TDMA: Time Division Multiple Access
- QPSK: Quadrature Phase Shift Keyring
- MAC: Medium Access Control
- BPI: Baseline Privacy Interface
- SF: Service Flows
- PDU: Protocol Data Unit?
- UCD: Upstream Channel Descriptor
- MAP: a type of message in the MAC protocol of DOCSIS (bandwidth allocation map)
- IE: Information Elements
- CBR: Constant Bit Rate
- VBR: Variable Bit Rate
- UGS: Unsolicited Grant Service
- (n)rtPS: (non-)real-time Polling Service
- USGAD: Unsolicited Grant Service with Activity Detection
- CO: Central Office
- FFTx : Fiber To The x
- FFTH: FFT Home
- FFTC: FFT Central Office
- PON: Passive Optical Networks (point to multipoint)
- PPP: Point to Point
- (C)WDM: (Coarse) Wavelength Division Multiplexing (= FDM)
- EPON: Ethernet PON
- NGEPON: Next Generation EPON
- BPON: Broadband PON (was APON)
- APON: ATM PON
- GPON: Gigabit PON
- OLT: Optical Line Terminal
- EFMA: Ethernet in the First Mile Alliance
- FSAN: Full Service Access Network
- MPCP: MultiPoint Control Protocol
- DBA: Dynamic Bandwidth Allocation
- RTT: Round Trip Time
- IPACT: Interleaved Polling with Adaptive Cycle Time
- BGP: Bandwidth Guaranteed Polling
- DEB: Deterministic Effective Bandwidth
- OTDM: Optical Time Division Multiplexing
- PMD: (spectral and) Polarisation Mode Dispersion
- EDFA: Erbium Doped Fiber Amplifier
...
- MTBF: Mean Time Between Failures
- MTTR: Mean Time To Repair
- SRLG: Shared Risk (Link) Group
- RHE: Recovery Head End
- RTE: Recovery Tail End
- SDH: Synchronous Digital Hierarchy
- QoR: (slide 24 lecture 6) Quality of Resilience ??
- LSA: Link State Advertisement
- FRR: Fast ReRoute
- NH: Next Hop
- PoP: Points of Presence
- SGSN: Serving GPRS Support Node
- GGSN: Gateway GPRS Support Node
(199 afkortingen, met wschl een 50tal niet opgenomen 14 jun 2006 23:06 (CEST))
--SCIFFY en Cube 5 juni 2006
--Jeroentrappers 7 jun 2006 17:22 (CEST)
--Kurt 8 juni 2006
--Wouter Horré 13 juni 2006
--en vele anderen