Date:
December 27, 2011
Duration: 90 Minutes
Support: Open Book
________________________________________________________________________
Task 1: Wireless LAN acc. to Standard IEEE 802.11
WLANs based on the IEEE 802.11 Standard operate according to the principle
CSMA/CA. Various modes are defined as DCF, PCF, and HCF.
a) In DCF mode, several stations may have a frame ready to be sent.
a1) Which station is is allowed to send ?
a2) When and under which conditions can that station start sending ?
a3) What happens in case of a collision due to the fact that 2 stations
start sending quasi simultaneous and how is this problem resolved ?
a4) What can be done to avoid interference by "hidden stations" during
the transmission of a frame ? Explain the method shortly.
b) PCF mode
b1) For which applications has the PCF mode been defined ?
b2) How long is the contention-free (CF) period ?
b3) By which mechanism are collisions avoided during the CF-period ?
Name the method and explain it shortly.
c) HCF mode
c1) For which applications and purpose has HCF been defined ?
c2) Explain shortly how the HCF-EDCA mode is realised using the
Arbitration Interframe Spacing (AIFS) method.
Task 2: Throughput Bounds for Token-Ring LANs
Consider a Ring-Structure with the following parameters:
Number of stations: N
Total length of the ring: l = 2 km
Propagation speed
c = 200000 km/s
Transmission rate
r = 1 Gbit/s
Constant frame length L = 1000 bit
Station latency can be neglected.
a) Calculate
a1)
a2)
a3)
a4)
the transmission time h for a frame
the propagation delay tau on the medium
the equivalent ring length l* in bits
the "a"-factor.
b) The parameters are chosen such that all three principal token passing
modes are applicable.
b1) Calculate the maximum medium utilization rho(max) for the three token
passing modes in case of "greedy sources" (there are always frames to send).
b2) Illustrate the case "Single Frame Mode" in a time diagram and verify the
result of question b1) for this mode.
Task 3: ATM Networks
Consider a WAN which operates acc. to the Asynchronous Transfer Mode (ATM).
a) General Properties of ATM
a1)
a2)
a3)
a4)
Which connection principle is used in ATM ?
How is the property of "Sequence Integrity" achieved in ATM ?
Which functions does the Header Error Control (HEC) perform ?
Consider an ATM Switching Node acc. to the Figure on page 219 of the
Course Notes. Construct the Translation Table for the established
Virtual Channel Connections (VCC). Which parts in the cell header of
transpassing cells have to be modified ?
b) Cell Synchronization
On each incoming physical link the function of "Cell Synchronization" has to be
performed.
b1) Why is this function of Cell Synchronization necessary ?
b2) At which instant is the cell synchronization lost ?
b3) Define the maximum time necessary to re-synchronize the link after
the instant when the cell synchrinization has been lost and no further
errors occur.
c) Service Classes in ATM
Which Service Class has to be chosen when
c1) a constant bitrate (CBR) is required ?
Name the service class and explain how jitters, caused by the asynchronous
operation of AT`M networks, are compensated ?
c2) Internet packets have to be transported through an ATM network.
Name the service class and describe shortly the function of the AAL.
Task 4: Internet Protocol (IP):
Consider the IP protocol acc. to IPv4 operated over an Ethernet LAN.
a) Host Addressing and Routing
a1) Identify the properties of addressing on the IP layer and on the MAC layer
with respect to Host Identification and Host Location.
a2) Which protocol provides the translation between both addresses ?
a3) On which addresses operates a router ?
a4) Under which conditions uses a router the "Fast Path" and the "Slow Path"
for packet forwarding ?
a5) Which information has to be exchanged for the Link-State Algorithm and for
the Distance Vector Algorithm and how is this information used to update
the Routing Tables ?
b) Data Unit Fragmentation
b1) Why is data unit fragmentation necessary ?
b2) Which two methods are known to deal with this problem ?
Discuss advantages and disadvantages of them shortly.
Task 5: Transport Protocols
Transport protocols are a vital part of the "TCP/IP Internet Protocol Suite".
a) Function of Transport Protocols
a1) Which basic functions are performed by the internet tarnsport protocols
UDP and TCP ? It is suffient to name them together with reasons why these
function are needed.
a2) How can, generally, a lost information block (e.g., a packet loss) be
recovered and by which mechanisms can deadlocks be avoided ?
a3) How is a lost information block recovered in UPD and in TCP ?
a4) Which connection principles are applied within the protocols UDP and TCP
and why ?
b) Flow and Congestion Control inTCP
b1) Why is the method of "Slow Start" in TCP required ?
b2) Characterize the functions of Flow Control and Congestion Control in TCP
according to the following criteria:
Purpose, Indicators (Metrics), Parameters, and Procedures.
Good Luck !
Best regards,
Paul J. Kuehn
Solution
Task 1: Wireless LAN acc. to Standard IEEE 802.11
a) DCF Mode
a1) Station X with the smallest number of slots in the Contention Window
s(min) = min(s1,s2,...,sN); si = number of slots of station X = i.
a2) Sending after reception of the ACK + after interframe spacing DIFS
+ after s(min) slot times.
Condition: CCA (Clear Channel Assessment) = idle.
a3) No ACK will be received; collision resolution acc. to an exponential
backoff window mechanism.
a4) Application of Virtual Channel Sensing and RTS/CTS Option using the
Network Allocation Vector (NAV) in the header of the RTS Control Frame.
The sending station sets the NAV such that all other stations receiving
this frame refrain from sending during the time NAV.
b) PCF Mode
b1) PCF mode is defined for a polling cycle initiated by the Access Point AP
for typically real-time applications which have to meet hard timing constraints.
b2) AP keeps a list of PCF stations from which the maximum duration of a
Contention-Free period is derived; the CF period starts with a beacon frame
and the corresponding maximum duration of the CF period within the
Duration ID field, upon which all non-PCF stations set their NAV value.
b3) Collisions are avoided by scheduled Request (Poll)/Response (Data+CF-ACK)
cycles cyclically applied to all PCFstations by the AP and by refraining from
sending of all nion-PCFstations during the CF-period (NAV).
c) HCF Mode
c1) The Hybrid Coordination Function (HCF) has been defined to support access
priorities between 4 classes of service.
c2) In HCF-EDCA the classes are defined by staggered Interframe Spacing (AIFS).
This can be achieved by setting the minimum and maximum values of the Collision
Window size CW of each class accordingly.
Task 2: Throughput Bounds for Token-Ring LAN's
a) Parameter Calculation
a1) Transmission Time h = L/r = 1 microsecond
a2) Propagation Delay tau = l/c = 1 microsecond
a3) Equivalent Ring Length l* = r x tau = 1000 bit
a4) a-Factor = tau/h = 1.
b) Principal Token Passing Modes
b1) Single Frame Method:
Single Token Method:
Multiple Token Method:
rho(max) = 1/(2 + 1/N)
rho(max) = 1/(1 + 1/N)
rho(max) = 1/(1 + 1/N)
b2) Time Diagram
h
Station 1: ------I=========I------------------------------------------------------------------------------------------h
Staion 2: -----------------------------------------------------I=========I-------------------------------------------.
.
.
.
.
.
.
.
.
.
----> time
Station N-1: .
.
.
.
.
tau
.
h
tau/N .
.
Station 1: -----I------------------I=========I____ I-------------------I---------------------------------------------tau
h
tau/N
Station 2: ---------------------------------------------------I-------------------I=========I_____I-----------------
Verification: rho(max) = h/(tau + h+ tau/N) = 1/(2+1/N).
Task 3: ATM Networks
a) General Properties of ATM Networks
a1) Virtual Connection, Connection-Oriented
a2) Sequence Integrity is achieved by guiding all cells of a connection along the same
connection path and by strict FIFO queuing disciplines in the cell buffers.
If re-ordering takes place in a switch, then the original order has to be established
again at the output of the switch.
ATM avoids overhead and does not require cell synchronization flags.
a3) HEC performs
1. Single and multiple error detection
2. Single error forward correction
3. Cell Delineation, i.e., the detection of cell borders by error syndrome calculation
a4) Translation Table for VPI/VCI indicators for example of Figure on page 219:
Incoming Cell
VPI/VCI
Outgoing Cell
VPI/VCI
VPI1/VCI1
VPI1/VCI2
VPI4/VCI1
VPI4/VCI2
VPI3/VCI3
VPI2/VCI4
VPI5/VCI1
VPI5/VCI2
Modification of VPI and VCI
Modification of VPI and VCI
Modification of VPI only
Modification of VPI only
b) Cell Synchronization
b1) Cell synchronization is necessary as cells arrive back-to-back without
spacing or a cell synchronization flag.
b2) Cell synchronization is lost at the transition from the state SYNCH into
state HUNT in the figure on page 224.
b3) Maximum time necessary to re-synchronize in case of no further errors:
t(resynch,max) = 424 x dt + delta x 424 x dt = 424 x (delta +1) x dt,
where dt indicates the time for one bit (clock period).
c) Service Classes in ATM
c1) CBR is required for streaming data with periodic transmission of constant
sized information blocks for real-time applications as, e.g., continuous voice or
music, also called "circuit emulation" or AAL Type 1
Jitters, caused by the asynchronous transfer mode, have to be compensated at
the destination by buffering and synchronous play-out.
c2) Connectionless Data Communication (CL) or AAL Type 5.
The ATM Adaptation Layer (AAL) performs the following functions:
Convergence Sublayer: Assembly of several IP packets into the payload part
of an AAL CPCS and encapsulation of the corresponding PDU (sending side),
and, espectively, decapsulation of the PDU to get the IP packets back (receiving side).
Segmentation and Reassembly Sublayer: Fragmentation of the CPCS PDU and
putting the fragments into the payload part of ATM cells, completion of the the ATM cells
(sending side) and, respectively, decapsulation of ATM cells and assmbling the CPCS PDU
from the payload parts of the cells.
Task 4: Internet Protocol (IP)
a) Host Addressing and Routing
a1) IP Layer: IP address serves as Host Identifier a n d Host Locator; thus,
from the IP address the location of the Host or the Router and their physical interface
have to be derived.
Note: As the IP address is not changed in case of a Mobile Host, a "care-off" address
has to be assigned to the host when leaving its original (home) location.
MAC Layer: The Mac addresses are "flat" addresses, i.e., they do not define the location
of the host and do not change in case of a Mobile Host. The location of a host has to be
found by corresponding entries in routing tables (e.g., within a Bridge or Switch).
a2) The translation between IP and MAC addresses is performed by the Address Resolution
Protocol (ARP) or simply by a broadcast request on the MAC layer and response by the
host station with the corresponding addresses.
a3) Routers operate exclusively on IP addresses.
a4) Packet forwarding from input Line Card to output Line Card follows on
- the Fast Path interconnecting both interfaces directly by the switch fabric in case the
routing information is cached in the input Line Card
- the Slow Path interconnecting at first the incoming Line Card with the Routing Engine
holding the complete Routing Table and, secondly, the Routing Engine with the outgoing Line
through the switch fabric in case when the Line Card does not hold the corresponding
destination IP address in its cache.
a5) Exchanged information for routing tables:
Link -State Algorithm:
Periodic exchange of the latest Routing Table on the links betwee the routers by the
Flooding Algorithm (i.e., propagate the adjused table information to all outputs but the input
a message has been received; discovery of the currently existing neighbor nodes by Hellomessages.
Distance Vector Algorithm:
Periodic exchange of distance information to destination nodes calculated by a decentralized
Shortest Path Algorithm through updating is own distance vectors upon distance vectors
received from neighbor nodes.
Updating is done by re-calculation of the distances through each neighbor node and selecting
the shotest path and corresponding output interface.
b) Data Unit Fragmentation
b1) Data unit fragmentation is necessary as different network types in the lower layers use different
PDU sizes which are optimized according to their physical properties
(examples: ATM cells, MAC frames of different LAN or MAN types).
b2) Method 1: Fragmentation of PDU's (packets) by transpassed Routers when necessary and
delivery
of the fragments to the destination host for re-assembly.
Method 2: Discovery of the minimum PDU size along the path to the destination and fragmentation
by
the origination host.
Advantage of Method 1, hosts are not involved, but routers become more complex
Advantage of Method 2: routers remain simpler, but overhead is required by additional signalling
using
the ICMP message protocol.
Task 5: Transport Protocols
a) Function of Transport Protocols
a1) UDP:Connectionless exchange of UDP PDU's between host ports (endsystems) independently of
each other checksum control of header and payload.
TCP: Bidirectional Connection-Oriented exchange of byte streams with flow and congestion control
functions. Error recovery by exchange of acknowledgement information onbyte sequence
numbers. Flow control by adapting the flow rate to the currently available path bandwidth and
congestion control by manupulating the window size.
a2) Lost information blocks can be discovered by missing sequence numberrs of blocks. Error
correction
either by retransmission or by forward error correction (erasure coding).
Deadlocks have to be avoided that time for the acknowledgement is controlled by a timer and in
case
of a missing acknowledgement by automatic re-transmission.
a3) Recovery of lost information in
UDP:UDP discovers only errors and discards errored PDU's to avoid misdelivery. In cases of realtime
applications retransmissionis notpossible to avoid excessive delays.
TCP: TCP controls the information by the exchange of byte sequence numbers. In case of lost
information the sender repeats the transmission.
a4) Connection Principles:
UDP: Connection-Less (CL); datagram transport; as no recovery is necessary connections are not
required.
TCP: Connection-Oriented (CO); Sequence numbers are exchanged for acknowwledgements, so
a
connection is required to initialize both protocol engines at the host sides and to carry out a
logically synchronized operation.
b) Flow and Congestion Control
b1) "Slow Start" is required as the Protocol Engines have no or only little knowledge about the available
bandwidth ofthe channel between the hosts or ont he capability of the partner host. So TCP probes
by the "Slow Start" the available bandwidth and increases it if no losses occur until a reasonable
state iof "Congestion Avoidance" is reached where the throughput rate varies between an upper
and a
lower value.
b2) Flow Control:
Purpose: Adapt sender to the capabilities ofthe receiver by signalling a "receiver
advertized window"
Indicators: Receiver Advertized Window size
Parameter: Number of Maximum Segmengt Sizes (MSS)
Procedure: TCP Protocol, i.e. by sending bidirectionally blocks of information, including
piggybacked
acknowledgements and Receiver Advertized Window sizes.
CongestionControl:
Purpose: Reaction to sudden bottlenecks in the network along the path between sender and
receiver
by throttling down the sending rate (negative feedback principle)
Indicators: Single and multiple packet losses
Parameter: Congestion Window added to the window size to limit the flow in case of congestion
Procedure: TCP Protocol, i.e., Fast retransmit, fast recovery and slow start mechanisms
Many new proposals as Explicit Congestion Notification, Retransmission Timing,
Active Queue Management.