EnergyEfficient Routing Protocols in Sensor Networks

1
Energy-Efficient Routing Protocols in Sensor
Networks
Presenter Limin Wang Advisor Sandeep
Kulkarni March 13, 2003
2
Motivations

• Traditionally, routing protocols for ad hoc
  networks are evaluated in terms of packet loss
  rates, routing message overhead, and route
  length.
• In sensor networks, energy consumption is the
  important metric.
• Sensor nodes have limited and non-replenishable
  energy resources.
• Energy use maps directly to system lifetime and
  utility.

3
Major Sources of Energy Waste

• Overhearing a node picks up packets that are
  destined to other nodes.
• Idle listening listening to receive possible
  traffic that is not sent.
• Idle energy dissipation cannot be ignored.
  (Idlereceivetransmit ratio are 11.051.4 or
  122.5 or 11.21.7.)
• Collision two nodes may transfer data to each
  other at the same time or several nodes transfer
  data to the same node at the same time.
• Control packet overhead

4
Some Observations

• Power off radio conserves energy both in
  overhearing due to data transfer, and in idle
  state energy dissipation when no traffic exits.
• Application- or routing-layers approach provide
  better information about when the radio is not
  needed.
• When there is significant node redundancy, we can
  power off some intermediate nodes while still
  maintaining connectivity.

5
Outline

• Algorithms
• GAF
• Span
• LEACH
• Conclusion
• Common Techniques
• Potential Problems

6
Geographical Adaptive Fidelity (GAF)
7
Features of GAF

• Each GAF node uses location information to
  associate itself with a virtual grid.
• Virtual grid all nodes in a particular grid
  square are equivalent with respect to forwarding
  packets.
• Nodes in the same grid coordinate with each other
  to determine who will sleep and how long.
• This determination is moderated by application
  and system information.
• Nodes periodically wake up and trade places to
  accomplish load balancing.
• routing fidelity uninterrupted connectivity
  between communicating nodes.

8
Determining node equivalence

• Use location information and
• virtual grid to determine node
• equivalence.
• virtual grid for two adjacent
• grids A and B, all nodes in A can
• communicate with all nodes in B and vice
  versa. Thus all nodes in each grid are equivalent
  for routing.
• Size virtual grid based on the nominal radio
  range R assume virtual grid is a square with r
  units on a side.
• The distance between two possible farthest nodes
  in any two adjacent grids must not be larger than
  R.
• or

9
GAF state transition

• Nodes are in one of three states sleeping,
  discovery, active.
• Nodes start out in discovery state. When in
  discovery, a node turns on its radio and exchange
  discovery message to find other nodes within the
  same grid.
• Discovery message node id, grid id, estimated
  node active time (enat), node state.
• Nodes negotiate which node will handle routing
  through an application-dependent ranking
  procedure.

10
Load balancing

• Load balancing all nodes remain up and running
  together for as long as possible.
• All nodes in the networks are equally important
  and no one node must be penalized more than any
  of the others.
• Alternative completely exhaust the energy of
  each node in turn while other nodes sleep.
• GAF use load balancing.
• After a node remains in the active
• state for time Ta, it changes its state
• to discovery to give a chance to other
• nodes within the same grid to become
• active
• The active node set Ta to enat, and enat
• is set to be less than the time to use up all
• Remaining energy (enlt). E.g. enlt/2.

11
Node Ranking

• Maximize network lifetime by selecting which
  nodes to handle routing (within the same grid)
• A node in the active state has higher rank than a
  node in discovery state.
• For nodes with the same state, GAF gives nodes
  with longer expected lifetime (enat) higher rank.
• Node ids are used to break ties.
• Application can choose different node rank rules
  according to their bias. E.g. application may
  favor the active nodes until they drain out
  energy in a sensor network.

12
Tuning GAF

• Estimated node active time (enat)
• Expected node lifetime (enlt) assuming the node
  will constantly consume energy at a maximum rate
  until it dies.
• Load balancing energy usage. E.g. enlt/2.
• Discovery message interval (Td)
• A uniform random value between 0 and some
  constant. (avoid contention)
• The range of Td can also be influenced by node
  rank.
• Active duration (Ta)
• Can be set to enat.
• Sleeping duration (Ts)
• Can be set to the enat of the active node.
• Considering node mobility, can use alternative a
  uniform random time between 0 and enat. This
  large range of Ts may often have nodes wake up
  quite early. In GAF, Ts is uniformly from the
  range enat/2, enat.

13
Adapting to high mobility

• Problem
• When nodes move, the active node may leave its
  grid. This may leave the prior grid without an
  active node, reducing routing fidelity. In
  scenarios with high mobility, this can greatly
  increase packet drop rates.
• Solution
• Let some nodes wake up earlier.
• each node estimates the time it expects to leave
  its grid (the expected node grid time or engt)
  and includes this information in the discovery
  message. engt r/s (r is the grid size, s is the
  current speed of a node)
• When other nodes enter sleeping state, they sleep
  for the smaller of enat and engt.
• GAF with node mobility adaptation GAF-mobility
  adaptation (GAF-ma). v.s. GAF-basic (GAF-b)

14
Span An Energy-Efficient Coordination Algorithm
for Topology Maintenance in Ad Hoc Wireless
Networks
15
Span basic idea

• Each node makes periodic, local decisions on
  whether to sleep or stay awake as a coordinator
  and participate in the forwarding backbone
  topology.
• A node decides to volunteer to be a coordinator
  if it discovers that two of its neighbors cannot
  communicate with each other directly or through
  an existing coordinator.
• To keep the number of redundant coordinators low
  and rotate this role amongst all nodes, each node
  delays announcing its willingness with a random
  delay that takes two factors into account the
  amount of remaining battery energy, and the
  number of pairs of neighbors it can connect
  together.
• Using only local information, scaling well with
  the number of nodes.

16
Span achieves four goals

• It ensures that enough coordinators are elected
  so that every node is in radio range of at least
  one coordinator.
• It rotates the coordinators in order to ensure
  that all nodes share the task of providing global
  connectivity roughly equally.
• It attempts to minimize the number of nodes
  elected as coordinators, thereby increasing
  network lifetime, but without suffering a
  significant loss of capacity or an increase in
  latency.
• It elects coordinators using only local
  information in a decentralized manner.

17
What is Capacity

• A connected backbone does not necessarily
  preserve capacity.
• Paths that could operate without interference in
  the original network should be represented in the
  backbone.
• Black nodes are coordinators.
• Packets between nodes 3 and 4 may contend for
  bandwidth with packets between nodes 1 and 2.
• If node 5 were a coordinator, no contention would
  occur.

18
Coordinator Announcement

• Periodically, a non-coordinator node determines
  if it should become a coordinator or not.
• Coordinator eligibility rule if two neighbors of
  a non-coordinator node cannot reach each other
  either directly or via one or two coordinators,
  the node should become a coordinator.
• not the minimum number of coordinators required
  to merely maintain connectedness.
• Good capacity.

19
Announcement Contention

• Announcement contention occurs where multiple
  nodes discover the lack of a coordinator at the
  same time, and all decide to become a
  coordinator.
• Solution delaying the coordinator announcements
  with a randomized back-off delay.
• Each node chooses a delay value, and delays the
  HELLO message that announces the nodes
  volunteering as a coordinator for that amount of
  time.
• If at the end of the delay, the node has not
  received any HELLO message from other potential
  coordinators, it sends HELLO message.
• Otherwise, it reevaluates its eligibility based
  on any HELLO messages received, and makes its
  announcement if and only if the eligibility rule
  still holds.

20
Decide the Back-off Delay (1)

• All nodes have roughly equal energy
• Only topology should play a role in deciding
  which nodes become coordinators.
• Ni the number of neighbors for node i.
• Ci the number of additional pairs of nodes among
  these neighbors that would be connected if i were
  to become a coordinator.
• utility
  of node i
• A node with a high utility should volunteer more
  quickly than one with smaller utility.
• 
  T the round-trip delay for a
  small packet over the wireless
  link.
• 
  R is picked uniformly at random
  
  from the interval 0,1.

21
Decide the Back-off delay (2)

• Nodes may have unequal energy
• Er the amount of energy at a node that still
  remains.
• Em the maximum amount of energy available at
  the same node.
• A node with a larger value of is more
  likely to volunteer to become a coordinator more
  quickly than one with a smaller ratio.

22
Fairness

• Ad hoc networks are rarely uniform.
• In a non-uniform topology a node that connects
  network partitions together will always be
  elected a coordinator (to preserve capacity).
  Such critical nodes will unavoidably die before
  other less-critical ones.
• In a mobile Span network, a given node is rarely
  stuck in such a position, and this improves
  fairness.

23
Coordinator Withdrawal

• In order to ensure fairness, after a node has
  been a coordinator for some period of time, it
  withdraws if every pair of neighbor nodes can
  reach each other via some other neighbors, even
  if those neighbors are not currently
  coordinators.
• To prevent temporary loss of connectivity between
  a coordinators withdrawal message and the
  announcement from a new coordinator.
• grace period A node continues to serve as a
  coordinator for a short period of time even after
  announcing its withdrawal.

24
Low-Energy Adaptive Clustering Hierarchy (LEACH)
25
Assumptions and Observations

• The base station is fixed and located far from
  the sensors.
• Communication between the sensor nodes and the
  base station is expensive.
• All nodes in the networks are homogeneous and
  energy constrained.
• There are no high-energy nodes through which
  communication can proceed.
• Sensor networks contain too much data for an
  end-user to process.
• Automated methods of aggregating data into a
  small set of meaningful information are required.

26
What is LEACH

• A clustering-based protocol that minimizes energy
  dissipation in sensor networks
• Key features
• The nodes organize themselves into local
  clusters, with one node acting as the local base
  station or cluster-head.
• Randomized rotation of the high-energy
  cluster-head position such that it rotates among
  the various sensors in order to not drain the
  battery of a single sensor.
• LEACH performs local data fusion to compress
  the amount of data being sent from the clusters
  to the base station, further reducing energy
  dissipation and enhancing system lifetime.

27
How does LEACH Work

• The operation of LEACH is broken into rounds.
• Each round has
• Set-up phase when the clusters are organized.
• Steady-state phase when data transfers to the
  base station occur.

28
Set-up Phase

• Cluster-head advertisement
• Each node decides whether or not to become a
  cluster-head for the current round with a certain
  probability.
• Cluster-head broadcasts an advertisement message
  to the rest of the nodes.
• Non-cluster-head nodes keep their receivers on to
  hear the advertisements.
• Each sensor node determines to which cluster it
  wants to belong by choosing the cluster-head that
  requires the minimum communication energy (e.g.
  the cluster-head advertisement heard with the
  largest signal strength).

29
Set-up Phase (cont.)

• Cluster Set-up
• Non-cluster node inform the cluster its
  willingness to become a member.
• Schedule Creation
• Cluster-head node create a TDMA schedule telling
  each node when it can transmit.
• This schedule is broadcast back to the nodes in
  the cluster.

30
Dynamic clusters
31
Steady-state Phase

• Data Transmission
• Send data during their allocated transmission
  time.
• The radio of non-cluster-head node can be turned
  off until the nodes allocated transmission time.
  
• Cluster-head nodes always keep their receivers
  on.
• Cluster head node performs signal processing
  functions to compress the data into a single
  signal.
• After a certain time (determined as a priori),
  the next round begins.

32
Determine the Number of Cluster-heads

• Normalized total system energy dissipated vs. the
  percent of nodes that are cluster-heads.
• 0 cluster-heads and 100 cluster-heads is the
  same.
• There exists an optimal percent of nodes that
  should be cluster-heads.

33
Radio Interference

• Transmission in one cluster will affect
  communication in a nearby cluster.
• Each cluster communicates using different CDMA
  codes.
• When a node decides to become a cluster-head, it
  chooses randomly from a list of spreading codes.

34
Hierarchical Clustering

• Cluster-head nodes would communicate with
  super-cluster-head nodes and so on.
• Top layer of the hierarchy sent data to the base
  station.
• For large networks, this could save a tremendous
  amount of energy.

35
Common Techniques of the Three

• Power off radio to conserve energy
• GAF nodes in the same virtual grid coordinate
  with each other to determine who will sleep
• Span non-coordinator nodes can sleep
• LEACH non-cluster-head nodes can sleep until the
  nodes allocated transmission time (in
  steady-state phase).
• Load balancing
• GAF nodes wake up periodically and trade places
  to accomplish load balancing.
• Span rotates the role of coordinator among
  nodes.
• LEACH cluster-head position rotates among the
  various sensors.
• Local decision, distributed algorithm
• Use randomized back-off to avoid contention

36
Potential Problems

• GAF
• Require location information
• Span
• Periodically broadcast HELLO messages to discover
  and react to changes in network topology
• LEACH
• Broadcast cluster-head advertisement and
  membership messages
• Synchronization