Cross-layer Design of Broadcasting Algorithms

1
Cross-layer Design of Broadcasting Algorithms

• Ilker Onat
• ELG7178F Project Presentation
• SITE, University of Ottawa, Canada
• ionat_at_site.uottawa.ca

2
Layered Design Paradigm

• Pros
• Each layer provides interface to neighbor layers
  and acts alone to perform its tasks
• Modularity Each layer can be modified if
  interfaces preserved
• Standardization
• Provides a framework for new designs, algorithms
• Cons
• Non-optimal design solutions
• restricted clearly defined functions for each
  layer
• no joint optimum for the whole system
• Cross-layer interference
• operation principles may interfere under specific
  operation environments

3
Classification of layer interaction 14

• Variable Interaction
• Example
• Drop rate is dependent on node speed (physical
  layer) and load (network layer).
• Joint dependence is different from single
  dependences may be unpredictable
• Algorithmic Interaction
• Operation principles of layers may interact
• Under specific operation environment and
  algorithms may interact

4
Example MAC and Network layer interaction 14
Two connections 1?2 and 3?4 1-6-2 and 3-5-4
independent 1-x-2 and 3-x-4 not
independent 1-y-2 and 3-z-4 not
independent! Conclusion Even if the paths are
disjoint, many MAC layer drops because of range
overlaps Selection of paths effect MAC Packet
drops at MAC may in return cause network layer to
search new paths
5
What is cross-layer design?

• Coupling adjacent layers to increase performance
• Layer synergy
• Joint optimization Layers manage same resources
  cooperatively
• Exchange of information between layers
• Measurements (e.g., signal strength of the links)
• Statistics (e.g., retransmission count of the MAC
  layer)
• Any information that might help other layers
  operations
• It is not
• A complete integration or getting rid of protocol
  layers

6
Why cross-layer design needed?

• Under layered design paradigm
• Some applications work with low performance
• Diverse applications with different QoS needs may
  not be supported
• Increased difficulties with certain environments
  (e.g., wireless)

7
Why cross-layer design is more important for
wireless networks?

• Wireless has come with its own specific problems
• Capacity-limited broadcast communications medium
• Fast time-varying and fading channels
• High error rate
• Specific challenges
• Network layer requires a graph
• Wireless networks do not come with links P.R.
  Kumar
• Wireless links are not stable ? No stable graph,
  links depend on
• Physical layer, hardware, MAC (interference)
• Mobility
• Heterogeneous applications with constraints, QoS
• Constraints change design principles across all
  layers

8
Examples of cross-layer design 1

• Channel state dependent techniques
• Smart scheduling based on the instantaneous
  channel state of each user (e.g., CDMA High Data
  Rate)
• Exploit fast changes in the channel state
• Priority to users with good channels
• Physical layer info propagated to upper layers
• Transmit power control (physical layer)
• Error rate (physical layer)
• Topology control (network layer)
• Interference reduction (MAC)
• Energy efficiency (node lifetime)
• QoS schemes using MAC and network layer
• Example 802.11e, network layer has to pass the
  packet type to MAC layer

9
Examples of cross-layer design 2

• Routing based on (traditionally topology)
• node parameters (node energy, capabilities)
• channel parameters
• energy efficiency
• Wireless TCP and ECN 6
• TCP cannot differentiate between congestion
  related loss and wireless channel related loss
  any loss causes sharp congestion window
  reduction.
• Wireless channel error rate is high ? lossy
  environment
• use Explicit Congestion Notification (ECN) to
  signal congestion
• cause of the drop is shared with transport layer

10
An example cross-layer design approach for
wireless networks 7

• No deterministic on-off links
• Try to capture the stochastic nature of the
  wireless channel
• Use lognormal shadowing model to decide packet
  success probability over each link
• Packet success probability depends on link
  distance and shadowing parameter
• Unreliable links cause MAC layer retransmissions
• In routing try to choose the path that has
  minimum expected of transmissions including
  retransmissions
• A more complex and intelligent network layer

11
What is a broadcast algorithm?

• A broadcast algorithm
• network layer algorithm
• a message is flooded to reach all nodes
• blind flooding all receivers will retransmit
• smart flooding
• Broadcast
• no specific destination address
• all hearing nodes are inherent destinations
• any wireless transmission is broadcast in nature

12
Applications of Broadcast Algorithms

• Route discovery
• Flooding is a key component for prominent ad hoc
  routing protocols (AODV, DSR)
• Paging, location discovery
• Alarm signals
• Sensor network flooding, node classification
  algorithms

13
Why broadcasting algorithms may need cross-layer
design?

• To avoid broadcast storm problem (as a result of
  blind flooding)
• All receivers respond simultaneously ? severe
  contention ? many collisions
• How to decrease redundancy?
• No RTS/CTS, no ACK for broadcast packets ?
  contention

14
Why broadcasting algorithms may need cross-layer
design?

• To achieve reliability
• How to achieve reliability?
• without ACKs
• with small number of retransmissions
• Drops, delays and re-orderings are important for
  some applications using broadcast
• Drops ? some algorithms stop
• Increased back-offs at MAC ? random delays ?
  delayed arrivals ? non-optimum or incorrect
  operations
• Random delay ? re-ordering ? non-optimum or
  incorrect operations
• Broadcasting should be efficient

15
Solutions for broadcast efficiency

• Decrease number of rebroadcasts (network layer)
• Eliminate certain nodes from transmitting group
  2-5
• select nodes who will retransmit
• assume ideal MAC
• Coordinate transmitting nodes 1
• assume no MAC ? if two packets are scheduled for
  the same time period both dropped
• Cross-layer solutions
• design broadcasting algorithms considering lower
  layer properties
• take into account collisions, re-ordering and
  delays
• low number of rebroadcasts also means more
  immunity to cross-layer interactions

16
Cross-layer solutions for broadcasting
algorithms - 1

• Observation Most packet drops in the multihop
  wireless networks are due to link-layer
  contention (as opposed to buffer overflow) 20
• Network layer help to MAC layer
• MAC layer to share the medium
• Contention based MAC gives up after certain
  number of trials
• Observation Contention based MAC layer drops are
  crippling for broadcast applications
• Network layer may introduce its own scheduling
  scheme to help MAC layer

17
Cross-layer solutions for broadcasting
algorithms 2

• Traditional network layer metrics time and
  message complexity (number of rebroadcasts)
• New cross-layer metrics required
• Reflecting contention based nature (reliability)
• Reflecting energy efficiency
• Combination of metrics rather than a single metric

18
Cross-layer solutions for broadcasting
algorithms 3

• Stochastic channel model (lognormal shadowing) to
  model physical layer 7
• Extend the definition of dominating set for
  probabilistic case
• Broadcast over the dominating set

19
Conclusions

• Performance and standardization conflict
• Application-driven and adaptive layers can
  benefit from sharing information across the
  protocol stacks
• Tailored designs for applications and networks
• Wireless networks get more and more dependent on
  the cross-layer solutions
• Broadcasting algorithms may especially benefit
  from cross-layer design
• New area with many open problems

20
References - 1

• 1 I. Chlamtac and O. Weinstein, The wave
  expansion approach to broadcasting in multihop
  radio networks, in IEEE Transactions on
  Communications. 39(3), pp. 426433, May 1991.
• 2 J. Wu and H. Li, On calculating connected
  dominating set for efficient routing in ad hoc
  wireless networks, in Proceedings of the 3rd
  international workshop on Discrete algorithms and
  methods for mobile computing and communications,
  pp. 714, ACM Press, 1999.
• 3 I. Stojmenovic, M. Seddigh, and J. Zunic,
  Dominating sets and neighbor elimination based
  broadcasting algorithms in wireless networks, in
  Proceedings of IEEE Hawaii International
  Conference on System Sciences, January 2001.
• 4 I. Chlamtac and S. Kutten, On broadcasting
  in radio networks - problem analysis and protocol
  design, in IEEE Transactions on Communications.
  33, pp. 12401246, December 1985.
• 5 R. Bar-Yehuda, O. Goldreich, and A. Itai, On
  the time-complexity of broadcast in multi-hop
  radio networks an exponential gap between
  determinism and randomization, J. Comput. Syst.
  Sci., vol. 45, no. 1, pp. 104126, 1992.
• 6 S. Shakkottai, T. S. Rappaport, and P. C.
  Karlsson, cross-layer design for wireless
  networks, IEEE Communications Magazine, vol.
  41, pp. 7480, October 2003.
• 7 I. Stojmenovic, A. Nayak, J. Kuruvila, F.
  Ovella-Martinez, and E. Villanueva-Pena,
  Physical layer impact on the design and
  performance of routing and broadcasting protocols
  in ad hoc and sensor networks, 2004.
• 8 V. R. Syrotiuk and A. Bikki, Modeling Cross
  Layer Interaction using Inverse Optimization, in
  Ad Hoc Networking, S. Basagni, M. Conti, S.
  Giordano,and I. Stojmenovic, editors. John Wiley
  Sons.

21
References - 2

• 9 V. Kawadia and P. R. Kumar, A cautionary
  perspective on cross layer design, IEEE Wireless
  Communication Magazine, July 2003.
• 10 ICC panel on Defining Cross-layer Design in
  Wireless Networking, http//www.eas.asu.edu/
  junshan/ICC03panel.html, 2003.
• 11 P. Urban, X. Defago, and A. Schiper,
  Contention-aware metrics for distributed
  algorithms Comparison of atomic broadcast
  algorithms, 2000.
• 12 W. Lou and J. Wu, On reducing broadcast
  redundancy in ad hoc wireless networks, IEEE
  Transactions on Mobile Computing, vol. 1, no. 2,
  pp. 111123, 2002.
• 13 S.-Y. Ni, Y.-C. Tseng, Y.-S. Chen, and J.-P.
  Sheu, The broadcast storm problem in a mobile ad
  hoc network, in Proceedings of the 5th annual
  ACM/IEEE international conference on Mobile
  computing and networking, pp. 151162, ACM Press,
  1999.
• 14 C. Barrett, A. Marathe, M. V. Marathe, and
  M. Drozda, Characterizing the interaction
  between routing and mac protocols in ad-hoc
  networks, in Proceedings of the 3rd ACM
  international symposium on Mobile ad hoc
  networking computing, pp. 92103, ACM Press,
  2002.
• 15 J. Tourrilhes, Robust broadcast improving
  the reliability of broadcast transmissions on
  csma/ca, in Personal, Indoor and Mobile Radio
  Communications, 1998.
• 16 K. Tang and M. Gerla, Mac layer broadcast
  support in 802.11 wireless networks, in MILCOM,
  2000.
• 17 S. Guha and S. Khuller, Approximation
  algorithms for connected dominating sets, Tech.
  Rep. 3660, Univ. of Maryland Inst. for Adv.
  Computer Studies-Dept. of Computer Sci., Univ. of
  Maryland, College Park, June 1996.

22
References - 3

• 18 I. Cidon and O. Mokryn, Propagation and
  leader election in a multihop broadcast
  environment, in Proceedings of the 12th
  InternationalSymposium on Distributed Computing,
  pp. 104118, Springer-Verlag, 1998.
• 19 J. Wu and F. Dai, "Broadcasting in Ad Hoc
  Networks Based on Self-Pruning, International
  Journal of Foundations of Computer Science , Vol.
  14, No. 2, April 2003, 201-221.
• 20 Z. Fu, P. Zerfos, H. Luo, S. Lu, L. Zhang,
  and M. Gerla, The impact of multihop wireless
  channel on TCP throughput and loss, in INFOCOM,
  2003.

23
Questions

• Q1) Name one advantage and one disadvantage of
  layered design paradigm
• Answer
• Advantage Modularity Each layer can be
  modified if interfaces preserved
• Disadvantage Non-optimal designs
• Q2) What is blind flooding?
• Answer
• A broadcast pattern in which all receivers
  retransmit
• Q3) Why blind flooding causes broadcast storm
  problem?
• Answer
• Many redundant packets
• Node in the same neighborhood attempt to
  rebroadcast simultaneously, high contention
• No RTS/CTS for broadcast packets