A High Performance Channel Sorting Scheduling Algorithm Based On Largest Packet

1
A High Performance Channel Sorting Scheduling
Algorithm Based On Largest Packet
P.G.Sarigiannidis, G.I.Papadimitriou, and
A.S.Pomportsis Department of Informatics,
Aristotle University Thessaloniki, Greece
2
Some keywords for the presentation are

• Optical networks
• WDM technology
• Broadcast and select star topology
• Reservation phase
• Transmission phase
• Scheduling algorithm
• Channel service
• Traffic prediction

3
Optical technology offers

• Huge bandwidth
• Protocol transparency
• Enhanced reliability
• In some cases easy and simple protocol
  functionality

4
Wavelength Division Multiplexing (WDM) offers

• Excellent way of exploiting the huge bandwidth of
  optical fibers
• Concurrency among multiple users transmitting at
  feasible rates
• Gigabit-per-second data rates in independent
  channels, which transmit concurrently data flows
  to a single or multiple users

5
This algorithm focuses on broadcast and select
LAN

• Local area network
• Broadcast and Select topology
• N nodes
• W channels (wavelengths)
• A passive star connects all nodes
• A TT-FR system (a tunable transmitter and a fixed
  receiver per node)

6
(No Transcript)
7
Medium Access Control (MAC) Protocols are

• Common policies for
• Which node will transmit
• On which channel will transmit
• During which time the transmission will be
  executed

8
MAC protocols are classified into

• Pre-transmission coordination based
• They designate at least one wavelength for
  coordination-control reasons
• Pre-allocation based
• The available channels are used only for data
  sending and receiving
• In a special category of pre-transmission
  coordination based the protocols are using
  control packets instead of a control channel

9
The coordination process entails two steps

• The algorithm accepts initially the demands of
  all nodes and organizes them in transmission
  frames
• The demands are stored in a demand matrix
  Ddi,j
• Each transmission frame stores for every node the
  number of timeslots required for transmission to
  a specific channel
• Time is divided in timeslots

10
Two very important prior scheduling algorithms
are

• Online Interval Scheduling Algorithm (OIS)
• K. M. Sivalingam, J. Wang, J. Wu and M. Mishra,
  An interval-based scheduling algorithm for
  optical WDM star networks, Photonic Network
  Communications, vol. 4, no. 1, (January 2002),
  pp. 73-87.
• Predictive online Scheduling Algorithm (POSA)
• E. Johnson, M. Mishra, and K. M. Sivalingam,
  Scheduling in optical WDM networks using hidden
  Markov chain based traffic prediction, Photonic
  Network Communications, vol. 3, no. 3, (July
  2001), pp. 271-286.

11
OIS has the following characteristics

• Begins the construction of the schedule just
  after reading the requests of the first node
  (online algorithm)
• Each node needs to maintain a list of time
  intervals that are available on every data
  channel
• For each node whose request is being processed
  nodes maintain one additional list of intervals
  that have not yet been assigned to the specific
  node for transmission

12
POSA functions as follows

• POSA attempts to reduce the duration of the
  schedule computation process by predicting the
  nodes requests for the next frame
• The predictor uses two different algorithms, the
  learning algorithm and the prediction algorithm
• The learning algorithm is responsible for
  informing and updating the data of the history
  queue
• The prediction algorithm is responsible for
  predicting the demand matrix as accurately as
  possible

13
The new proposed protocol is called First Max
Predictive Online Scheduling Algorithm (FM-POSA)

• It tries to reduce the idle timeslots of the
  final schedule by changing the service order of
  the nodes
• Concurrently it adopts the prediction mechanism
  of POSA in order to reduce the computation time
  of the schedule

14
The algorithm operates in three independent
phases

• The learning phase
• During the first phase the FM-POSA monitors the
  traffic of the network and fills the history
  queues with the actual traffic demands.
• The shifting phase
• FM-POSA stops to construct the schedule based on
  the actual requests and it enters the prediction
  phase.
• The prediction phase
• In the last and most important phase the
  algorithm predicts the nodes requests for the
  next frame based on its learning phase. In
  addition the algorithm performs the forwarding of
  packets to their destinations.

15
The new element that is introduced by FM-POSA is
the order in which the nodes requests are
processed.

• FM-POSA searches for the largest packet and sorts
  nodes according to this.
• The first node that is served is the one with the
  largest packet following by the one with the
  second largest packet and ending with the one
  with the smallest packet.
• If two nodes have packets of equal size the node
  that will be served first is randomly selected.

16
It would be useful to see an example of the
processing of a demand matrix in order to
understand the operation of the proposed
algorithm
Let us consider the demand matrix D.
D
17
Firstly FM-POSA acts as follows

• Action 1 Search and locate the largest packet
  from the requests of the nodes and save it in a
  vector called MAX.

D
MAX
18
Next FM-POSA performs

• Action 2 Reorder the elements in MAX in
  descending order.

MAX
S_MAX
19
The reordering of vector MAX means that the
processing of requests in order to produce the
scheduling matrix is dictated by the final vector
S_MAX.
Firstly, Node 2 will be processed
S_MAX
Secondly, Node 1 will be processed
Thirdly, Node 0 will be processed
20
For the specific demand matrix D OIS/POSA will
construct the following final schedule
21
For the specific demand matrix D FM-POSA will
construct the following final schedule
22
For the specific example

• OIS/POSA wastes
• FM-POSA wastes

23
In the simulation results we consider

• Two network models. The first consists of 4
  channels and a row of nodes (10, 20, 30, 40, and
  50). The second consists of 8 channels and a row
  of nodes (10, 20, 30, 40, and 50).
• N symbols the number of nodes, W symbols the
  number of channels, K is the maximum value for
  incoming packets.

24
Also we consider

• The speed of the line has been defined at 2.4
  Gbps per channel.
• The tuning latency is considered to be equal to
  zero for simplicity reasons.
• A 10 of the simulated frames belongs to the
  learning phase.
• The two algorithms (FM-POSA and POSA) are
  compared under uniform traffic.

25
In the channel utilization for 4 channelsFM-POSA
is better than POSA
26
In the network throughput for 4 channelsFM-POSA
is better than POSA
27
In the relation throughput-delay FM-POSA reduces
a little the mean waiting time of the packets for
4 channels
28
FM-POSA remains better than POSA in channel
utilization for 8 channels
29
In the network throughput for 8 channelsFM-POSA
is better than POSA
30
Again in the relation throughput-delay FM-POSA
remains better than POSA for 8 channels
31
Conclusions

• We introduced a new scheduling algorithm for
  collision free WDM star networks.
• The new scheme offers a better utilization of the
  available channels of the network.
• Also, it brings an improvement in channel
  utilization and network throughput by changing
  the order of the processing of each node based on
  the largest request.

32
Thank you for your attention!!!
33
A New Channel Priority Scheduling Technique for
WDM Star Networks
P.G.Sarigiannidis, G.I.Papadimitriou, and
A.S.Pomportsis Department of Informatics,
Aristotle University Thessaloniki, Greece
34
This algorithm focuses on broadcast and select
LAN

• Local area network
• Broadcast and Select topology
• N nodes
• W channels (wavelengths)
• A passive star connects all nodes
• A TT-FR system (a tunable transmitter and a fixed
  receiver per node)

35
Three prior scheduling algorithms are

• Online Interval Scheduling Algorithm (OIS)
• K. M. Sivalingam, J. Wang, J. Wu and M. Mishra,
  An interval-based scheduling algorithm for
  optical WDM star networks, Photonic Network
  Communications, vol. 4, no. 1, (January 2002),
  pp. 73-87.
• Predictive online Scheduling Algorithm (POSA)
• E. Johnson, M. Mishra, and K. M. Sivalingam,
  Scheduling in optical WDM networks using hidden
  Markov chain based traffic prediction, Photonic
  Network Communications, vol. 3, no. 3, (July
  2001), pp. 271-286.
• Check and sort-predictive online scheduling
  algorithm (CS-POSA)
• P. G. Sarigiannidis, G. I. Papadimitriou, A. S.
  Pomportsis, CS-POSA A high performance scheduling
  algorithm for WDM star networks, Photonic
  Network Communication, vol 11, no 3 (2006), pp.
  209-225.

36
The new proposed protocol is called Load Eclectic
Navigated Algorithm (LENA)

• The algorithm begins the manufacture of
  scheduling matrix with the channel that contains
  the most requests. That means that the algorithm
  does not chooses the channels from the first one
  to the last one, but selects each time the
  channel with the most time requests for the
  specific node.
• Concurrently it adopts the prediction mechanism
  of POSA in order to reduce the computation time
  of the schedule

37
It would be useful to see an example of the
processing of a demand matrix in order to
understand the operation of the proposed
algorithm
Let us consider the demand matrix D.
D
38
Firstly LENA acts as follows

• Step 1 First of all LENA adds each row of the
  traffic matrix D in a new vector S that will
  register the total amount of requests by each
  node

D
S
39
Next LENA performs

• Step 2 Then LENA grades vector S in a declining
  order

S
S
40

• It is examined in which channel have been
  requested most demands.
• In the particular example we observe that in
  channel W2 have been assembled most requests for
  node N2 and the transmission time for them are
  equal to four timeslots.
• Then are examined the other two channels and we
  observe that their demands are equal with 2
  timeslots, so the selection is random.

41

• LENA continues with the set of demands, which
  belongs to node N0. Hence, the channel W0 will be
  selected (three timeslots), channel W1 could be
  follow (two timeslots), and LENA could be finish
  with node N0, by selecting the demands of W0 (two
  timeslot).

42

• Finally, LENA finishes the construction of the
  scheduling matrix, by servicing the requests of
  Node N1. Again, channel W2 (three timeslots) will
  be selected first, channel W1 (two timeslots)
  will follow and the last selection will be the
  channel W0 (one timeslot).

43
For the specific demand matrix D OIS/POSA will
construct the following final schedule
44
For the specific demand matrix D LENA will
construct the following final schedule
45
In the simulation results we consider

• Two network models. The first consists of 8
  channels and a row of nodes (10, 20, 30, 40, and
  50). The second consists of 16 channels and a row
  of nodes (10, 20, 30, 40, and 50).
• N symbols the number of nodes, W symbols the
  number of channels, K is the maximum value for
  incoming packets.

46
Also we consider

• The speed of the line has been defined at 2.4
  Gbps per channel.
• The tuning latency is considered to be equal to
  zero for simplicity reasons.
• A 10 of the simulated frames belongs to the
  learning phase.
• The three algorithms (LENA, CS-POSA, and POSA)
  are compared under uniform traffic

47
In the channel utilization for 8 channelsLENA is
better than POSA and CS-POSA
48
In the network throughput for 8 channelsLENA is
better than POSA and CS-POSA
49
In the relation throughput-delay LENA improves
the network throughput and meanwhile reduces a
little the mean time delay for 8 channels.
50
In the channel utilization for 16 channelsLENA
is better than POSA and CS-POSA
51
In the network throughput for 16 channelsLENA is
better than POSA and CS-POSA
52
In the relation throughput-delay LENA improves
the network throughput and meanwhile reduces a
little the mean time delay for 16 channels.
53
Conclusions

• We presented another novel technique for
  scheduling in WDM star networks.
• This method changes the way of processing of each
  channel.
• It leads to an improvement, in terms of channel
  utilization, network throughput and mean time
  delay, which is proved by simulation results.

54
Thank you AGAIN for your attention!!!