Chapter 4

Routing Protocols

1
 Overview
 Routing in WSNs is challenging due to distinguish from
other wireless networks like mobile ad hoc networks or
cellular networks.
 First, it is not possible to build a global addressing scheme for a
large number of sensor nodes. Thus, traditional IP-based protoc
ols may not be applied to WSNs. In WSNs, sometimes getting t
he data is more important than knowing the IDs of which nodes
sent the data.
 Second, in contrast to typical communication networks, almost
all applications of sensor networks require the flow of sensed d
ata from multiple sources to a particular BS.

2
Overview (cont.)
 Third, sensor nodes are tightly constrained in terms of energ
y, processing, and storage capacities. Thus, they require care
fully resource management.
 Fourth, in most application scenarios, nodes in WSNs are ge
nerally stationary after deployment except for, may be, a few
mobile nodes.
 Fifth, sensor networks are application specific, i.e., design re
quirements of a sensor network change with application.
 Sixth, position awareness of sensor nodes is important since
data collection is normally based on the location.
 Finally, data collected by many sensors in WSNs is typically
based on common phenomena, hence there is a high probabil
ity that this data has some redundancy.
3
Overview (cont.)
 The task of finding and maintaining routes in WSNs is
nontrivial since energy restrictions and sudden change
s in node status (e.g., failure) cause frequent and unpr
edictable topological changes.
 To minimize energy consumption, routing techniques
proposed for WSNs employ some well-known routing
strategies, e.g., data aggregation and in-network proce
ssing, clustering, different node role assignment, and
data-centric methods were employed.

4
Outline

 4.1 Routing Challenges and Design Issues in WSNs

 4.2 Flat Routing
 4.3 Hierarchical Routing
 4.4 Location Based Routing
 4.5 QoS Based Routing
 4.6 Data Aggregation and Convergecast
 4.7 Data Centric Networking
 4.8 ZigBee
 4.9 Conclusions

5
 Chapter 4.1
Routing Challenges and Design Issues in
WSNs

6
Overview
 The design of routing protocols in WSNs is influenced by man
y challenging factors. These factors must be overcome before e
fficient communication can be achieved in WSNs.
 Node deployment
 Energy considerations
 Data delivery model
 Node/link heterogeneity
 Fault tolerance
 Scalability
 Network dynamics
 Transmission media
 Connectivity
 Coverage
 Data aggregation/convergecast
 Quality of service
7
Node Deployment
 Node deployment in WSNs is application dependent a
nd affects the performance of the routing protocol.
 The deployment can be either deterministic or random
ized.
 In deterministic deployment, the sensors are manually
placed and data is routed through pre-determined path
s.
 In random node deployment, the sensor nodes are scat
tered randomly creating an infrastructure in an ad hoc
manner.

8
Energy Considerations
 Sensor nodes can use up their limited supply of energ
y performing computations and transmitting informati
on in a wireless environment. Energy conserving form
s of communication and computation are essential.
 In a multi-hop WSN, each node plays a dual role as da
ta sender and data router. The malfunctioning of some
sensor nodes due to power failure can cause significan
t topological changes and might require rerouting of p
ackets and reorganization of the network.

9
 Data Delivery Model
 Time-driven (continuous)
 Suitable for applications that require periodic data monitorin
g
 Event-driven
 React immediately to sudden and drastic changes
 Query-driven
 Respond to a query generated by the BS or another node in t
he network
 Hybrid
 The routing protocol is highly influenced by the data r
eporting method

10
Node/Link Heterogeneity
 Depending on the application, a sensor node can have
a different role or capability.
 The existence of a heterogeneous set of sensors raises
many technical issues related to data routing.
 Even data reading and reporting can be generated fro
m these sensors at different rates, subject to diverse Q
oS constraints, and can follow multiple data reporting
models.

11
Fault Tolerance

 Some sensor nodes may fail or be blocked due to lack

of power, physical damage, or environmental interfere
nces
 It may require actively adjusting transmission powers
and signaling rates on the existing links to reduce ener
gy consumption, or rerouting packets through regions
of the network where more energy is available

12
Scalability
 The number of sensor nodes deployed in the sensing a
rea may be on the order of hundreds or thousands, or
more.
 Any routing scheme must be able to work with this hu
ge number of sensor nodes.
 In addition, sensor network routing protocols should b
e scalable enough to respond to events in the environ
ment.

13
Network Dynamics
 Routing messages from or to moving nodes is more ch
allenging since route and topology stability become i
mportant issues
 Moreover, the phenomenon can be mobile (e.g., a targ
et detection/ tracking application).

14
Transmission Media
 In general, the required bandwidth of sensor data will
be low, on the order of 1-100 kb/s. Related to the trans
mission media is the design of MAC.
 TDMA (time-division multiple access)
 CSMA (carrier sense multiple access)

15
Connectivity
 High node density in sensor networks precludes them
from being completely isolated from each other.
 However, may not prevent the network topology from
being variable and the network size from shrinking du
e to sensor node failures.
 In addition, connectivity depends on the possibly rand
om distribution of nodes.

16
Coverage
 In WSNs, each sensor node obtains a certain view of t
he environment.
 A given sensor’s view of the environment is limited in
both range and accuracy.
 It can only cover a limited physical area of the enviro
nment.

17
Data Aggregation/Convergecast

 Since sensor nodes may generate significant redundan

t data, similar packets from multiple nodes can be agg
regated to reduce the number of transmissions.
 Data aggregation is the combination of data from diffe
rent sources according to a certain aggregation functio
n.
 Convergecasting is collecting information “upwards”
from the spanning tree after a broadcast.

18
Quality of Service
 In many applications, conservation of energy, which i
s directly related to network lifetime.
 As energy is depleted, the network may be required to
reduce the quality of results in order to reduce energy
dissipation in the nodes and hence lengthen the total n
etwork lifetime.

19
Routing Protocols in WSNs: A taxonomy

Routing protocols in WSNs

Network Structure Protocol Operation

Network Structure Protocol Operation
Flat routing Negotiation based routing
• SPIN • SPIN
• Directed Diffusion (DD) Multi-path network routing
Hierarchical routing • DD
• LEACH Query based routing
• PEGASIS • DD, Data centric routing
• TTDD QoS based routing
Location based routing • TBP, SPEED
• GEAR Coherent based routing
• GPSR • DD
Aggregation
• Data Mules, CTCCAP

20
Reference
 J. N. Al-Karaki and A. E. Kamal, “Routing techniques in wirele
ss sensor networks: a survey,” IEEE Wireless Communications,
vol. 11, no. 6, pp. 6-28, Dec. 2004.

21
 Chapter 4.2
Flat Routing

22
Overview
 In flat network, each node typically plays the same role and sen
sor nodes collaborate together to perform the sensing task.
 Due to the large number of such nodes, it is not feasible to assi
gn a global identifier to each node. This consideration has led t
o data centric routing, where the BS sends queries to certain re
gions and waits for data from the sensors located in the selecte
d regions. Since data is being requested through queries, attribu
te-based naming is necessary to specify the properties of data.
 Prior works on data centric routing, e.g., SPIN and Directed Di
ffusion, were shown to save energy through data negotiation an
d elimination of redundant.

23
 4.2.1
SPIN
Sensor Protocols for Information via Negotiation

24
SPIN -Motivation
 Sensor Protocols for Information via Negotiation, SPI
N
 A Negotiation-Based Protocols for Disseminating Informatio
n in Wireless Sensor Networks.
 Dissemination is the process of distributing individual
sensor observations to the whole network, treating all
sensors as sink nodes
 Replicate complete view of the environment
 Enhance fault tolerance
 Broadcast critical piece of information

25
SPIN (cont.)- Motivation
 Flooding is the classic approach for dissemination
 Source node sends data to all neighbors
 Receiving node stores and sends data to all its neighbo
rs
 Disseminate data quickly
 Deficiencies
 Implosion
 Overlap
 Resource blindness

26
SPIN (cont.)-Implosion

Node
x x
The direction
of data sending
The connect
between nodes
B C

x x
D

27
SPIN (cont.)- Overlap

r
q s
Node
The direction
of data sending
The connect
between nodes
The searching A B
range of the
node
(q, r) (s, r)

C
28
SPIN (cont.)- Resource blindness
 In flooding, nodes do not modify their activities based
on the amount of energy available to them.

 A network of embedded sensors can be resource-awar

e and adapt its communication and computation to the
state of its energy resource.

29
SPIN (cont.)
 Negotiation
 Before transmitting data, nodes negotiate with each other to
overcome implosion and overlap
 Only useful information will be transferred
 Observed data must be described by meta-data

 Resource adaptation
 Each sensor node has resource manager
 Applications probe manager before transmitting or processin
g data
 Sensors may reduce certain activities when energy is low

30
SPIN (cont.)- Meta-Data
 Completely describe the data
 Must be smaller than the actual data for SPIN to be beneficia
l
 If you need to distinguish pieces of data, their meta-data sho
uld differ

 Meta-Data is application specific

 Sensors may use their geographic location or unique node ID
 Camera sensor may use coordinate and orientation

31
SPIN (cont.)- SPIN family
 Protocols of the SPIN family
 SPIN-PP
 It is designed for a point to point communication, i.e., hop-
by-hop routing
 SPIN-EC
 It works similar to SPIN-PP, but, with an energy heuristic
added to it
 SPIN-BC
 It is designed for broadcast channels
 SPIN-RL
 When a channel is lossy, a protocol called SPIN-RL is use
d where adjustments are added to the SPIN-PP protocol to
32
account for the lossy channel.
SPIN (cont.)- Three-stage handshake protocol
 SPIN-PP: A three-stage handshake protocol for point-t
o-point media
 ADV – data advertisement
 Node that has data to share can advertise this by transmitti
ng an ADV with meta-data attached
 REQ – request for data
 Node sends a request when it wishes to receive some actua
l data
 DATA – data message
 Contain actual sensor data with a meta-data header
 Usually much bigger than ADV or REQ messages

33
SPIN (3-Step Protocol)

DATA
A

REQ
AQ

ADV
DAA D AT
R
V E
Q
DTA
V AD
DAD
A VEA
RT

RE
Q V B
A
ATD
D A
AD
DTVA

V
DAA

REQ
REQ

34
SPIN (3-Step Protocol)

DATA
A
DA
TA A
DAT

B
ATA
D
A
DAT

Notice the color of the data packets sent by node B

35
SPIN (3-Step Protocol)

DATA
A
DA
TA A
DAT

B
ATA
D
A
DAT

SPIN effective when DATA sizes are large :

REQ, ADV overhead gets amortized
36
SPIN (cont.)- SPIN-EC (Energy-Conserve)
 Add simple energy-conservation heuristic to SPIN-PP
 SPIN-EC: SPIN-PP with a low-energy threshold
 Incorporate low-energy-threshold
 Works as SPIN-PP when energy level is high
 Reduce participation of nodes when approaching low-energy-t
hreshold
 When node receives data, it only initiates protocol if it can p
articipate in all three stages with all neighbor nodes
 When node receives advertisement, it does not request the da
ta
 Node still exhausts energy below threshold by receiving ADV
or REQ messages
37
SPIN (cont.)- Conclusion
 SPIN protocols hold the promise of achieving high performanc
e at a low cost in terms of complexity, energy, computation, an
d communication
 Pros
 Each node only needs to know its one-hop neighbors
 Significantly reduce energy consumption compared to floodi
ng
 Cons
 Data advertisement cannot guarantee the delivery of data
 If the node interested in the data are far from the source, da
ta will not be delivered
 Not good for applications requiring reliable data delivery,
e.g., intrusion detection
38
SPIN (cont.)- Reference
 J. Kulik, W.R. Heinzelman, and H. Balakrishnan, “Negotiation-
based protocols for disseminating information in wireless sens
or networks,” Wireless Networks, Vol. 8, pp. 169-185, 2002.

39
 4.2.2
Directed Diffusion
A Scalable and Robust Communication Parad
igm for Sensor Networks

40
 Overview

 Data-centric communication A sensor field

Sources
 Data is named by attribute-value pairs Event
Event
 Different form IP-style communication
 End-to-end delivery service
 e.g.
 How many pedestrians do you obser
Directed
ve in the geographical region X? Diffusion

Sink Node

41
Overview (cont.)
 Data-centric communication (cont.)
 Human operator’s query (task) is diffused
 Sensors begin collecting information about query
 Information returns along the reverse path
 Intermediate nodes aggregate the data
 Combing reports from sensors

 Directed Diffusion is an important milestone in the dat

a centric routing research of sensor networks

42
Directed Diffusion
 Typical IP based networks
 Requires unique host ID addressing
 Application is end-to-end

 Directed diffusion – use publish/subscribe

 Inquirer expresses an interest, I, using attribute values
 Sensor sources that can service I, reply with data

43
Directed Diffusion (cont.)
 Directed diffusion consists of
 Interest - Query which specifies what a user wants
 Data - Collected information
 Gradient
 Direction and data-rate
 Events start flowing towards the originators of interests
 Reinforcement
 After the sink starts receiving events, it reinforces at least
one neighbor to draw down higher quality events

44
Data Naming
 Expressing an Interest
 Using attribute-value pairs
 e.g.,

Type = Wheeled vehicle // detect vehicle location

Interval = 20 ms // send events every 20ms
Duration = 10 s // Send for next 10 s
Rect = [-100,100, 200,400] // from sensors in this area

45
Interests and Gradients
 Interest propagation
 The sink broadcasts an interest
 Exploratory interest with low data-rate
 Neighbors update interest-cache and forwards it
 Flooding
 Geographic routing
 Use cached data to direct interests

 Gradient establishment
 Gradient set up to upstream neighbor
 Low data-rate gradient
 Few packets per unit time needed

46
Gradient Set Up
 Inquirer (sink) broadcasts exploratory interest, i1
 Intended to discover routes between source and sink

 Neighbors update interest-cache and forwards i1

 Gradient for i1 set up to upstream neighbor

 No source routes
 Gradient – a weighted reverse link
 Low gradient  Few packets per unit time needed

47
Exploratory Gradient
Exploratory Request
Gradient

Event
Event

Low Data-rate Interest

Low Data-rate Interest

Low Data-rate Interest

48
Data Propagation
 A sensor node that detects a target
 Search its interest cache
 Compute the highest requested data-rate among all i
ts outgoing gradients
 Data message is unicast individually
 A node that receives a data message
 Find a matching interest entry in its cache
 Check the data cache for loop prevention
 Re-send the data to neighbors

49
Reinforcement (1/4)
 Positive reinforcement
 Sink selects the neighboring node
 Original interest message but with high data-rate
 Neighboring node must also reinforce at least one neighbor
 Low-delay path is selected
 Exploratory gradients still exist: useful for faults

Reinforced gradient
Event
Event Reinforced gradient
Source
A sensor field
Sink

50
Reinforcement (2/4)
 Path failure and recovery
 Link failure detected by reduced rate, data loss
 Choose next best link (i.e., compare links based on infrequ
ent exploratory downloads)
 Negatively reinforce lossy link
 Either send interest with base (exploratory) data rate or all
ow neighbor’s cache to expire over time
Link
Link A-M
A-M lossy
lossy
Event
Event D A
A reinforces
reinforces BB
M B
B reinforces
reinforces CC
Source A
C
C reinforces
reinforces DD
C
B Sink
or
or
A
A negative
negative reinforces
reinforces M
M
M
M negative
negative reinforces
reinforces D
D
51
Reinforcement (3/4)
 Multipath routing B
Source
 Consider each gradient’s link quality
Event
Event
 Using negative reinforcement A
Sink
 Path Truncation
 Loop removal Multiple paths
 For resource saving
 Ex:
 B gets same data from both A and D, but
D always delivers late due to looping D E
 B negative reinforces D, D negative rein
forces E, E negative reinforces B A B C
 Loop B→E →D →B eliminated A removable loop
 Conservative negative reinforces useful for
fault resilience
52
 Design Considerations
 Design Space for Diffusion
Diffusion element Design Choices
Interest Propagation •Flooding
•Constrained or directional flooding based on location
•Directional propagation based on previously cached data
Data Propagation •Reinforcement to single path delivery
•Multipath delivery with selective quality along different
paths
• Multipath delivery with probabilistic forwarding
Data caching and •For robust data delivery in the face of node failure
aggregation •For coordinated sensing and data reduction
• For directing interests
Reinforcement •Rules for deciding when to reinforce
•Rules for how many neighbors to reinforce
•Negative reinforcement mechanisms and rules
53
Directed Diffusion: Pros & Cons
 Different from SPIN in terms of on-demand data querying mec
hanism
 Sink floods interests only if necessary (lots of energy savings)
 In SPIN, sensors advertise the availability of data
 Pros
 Data centric: All communications are neighbor to neighbor w
ith no need for a node addressing mechanism
 Each node can do aggregation & caching
 Cons
 On-demand, query-driven: Inappropriate for applications req
uiring continuous data delivery, e.g., environmental monitori
ng
 Attribute-based naming scheme is application dependent
 For each application it should be defined a priori
 Extra processing overhead at sensor nodes
54
Conclusions
 Directed diffusion, a paradigm proposed for event monitoring s
ensor networks
 Directed Diffusion has some novel features - data-centric disse
mination, reinforcement-based adaptation to the empirically be
st path, and in-network data aggregation and caching.
 Notion of gradient (exploratory and reinforced)
 Energy efficiency achievable
 Diffusion mechanism resilient to fault tolerance
 Conservative negative reinforcements proves useful

55
References
 C. Intanagonwiwat, R. Govindan, and D. Estrin, “Directed Diff
usion: A Scalable and Robust Communication Paradigm for Se
nsor Networks,” in the Proceedings of the Sixth Annual Interna
tional Conference on Mobile Computing and Networks (MobiC
om’00), August 2000.
 C. Intanagonwiwat, R. Govindan, D. Estrin, J. Heidemann, and
F. Silva, “Directed Diffusion for Wireless Sensor Networking,”
IEEE/ACM Transactions on Networking, Vol. 11, No. 1, Feb. 2
003.

56
 Chapter 4.3
Hierarchical Routing

57
Overview
 In a hierarchical architecture, higher energy nodes can be used
to process and send the information while low energy nodes ca
n be used to perform the sensing of the target.
 Hierarchical routing is mainly two-layer routing where one lay
er is used to select cluster heads and the other layer is used for
routing.
 Hierarchical routing (or cluster-based routing), e.g., LEACH, P
EGASIS, TTDD, is an efficient way to lower energy consumpti
on within a cluster and by performing data aggregation and fusi
on in order to decrease the number of transmitted messages to t
he base stations.

58
 4.3.1
LEACH
Low-Energy Adaptive Clustering Hierarchy

59
LEACH
 LEACH (Low-Energy Adaptive Clustering Hierarchy), a cluste
ring-based protocol that minimizes energy dissipation in sensor
networks.
 LEACH outperforms classical clustering algorithms by using a
daptive clusters and rotating cluster-heads, allowing the energy
requirements of the system to be distributed among all the sens
ors.
 LEACH is able to perform local computation in each cluster to
reduce the amount of data that must be transmitted to the base s
tation.
 LEACH uses a CDMA/TDMA MAC to reduce inter-cluster an
d intra-cluster collisions.

60
LEACH (cont.)
 Sensors elect themselves to be local cluster-heads at a
ny given time with a certain probability.
 Each sensor node joins a cluster-head that requires the
minimum communication energy.
 Once all the nodes are organized into clusters, each cl
uster-head creates a transmission schedule for the nod
es in its cluster.
 In order to balance the energy consumption, the cluste
r-head nodes are not fixed; rather, this position is self-
elected at different time intervals.

61
LEACH
100 m
叢集區

觀測區域

~100m

Sensor (Non Cluster Head)

Sensor (Cluster Head)
Initial Data
Aggregated Data

62 Base Station
LEACH: Adaptive Clustering
 Periodic independent self-election
 Probabilistic
 CSMA MAC used to advertise
 Nodes select advertisement with strongest signal strength
 Dynamic TDMA cycles

All nodes marked with a given symbol belong to the same cluster, and
the cluster head nodes are marked with a ●.
63
Algorithm
 Periodic process
 Two phases per round:
 Setup phase
 Advertisement: Execute election algorithm
 Members join to cluster
 Cluster-head broadcasts schedule

 Steady-State phase
 Data transmission to cluster-head using TDMA
 Cluster-head transfers data to BS (Base Station)

64
Algorithm (cont.)
Fixed-length cycle

Setup phase Steady-state phase

Time slot Time slot Time slot

1 2 3

Advertisement phase Cluster setup phase Broadcast schedule

Self-election of cluster
heads Members Cluster head Broadcast
Cluster heads compete compete with CDMA code to members
with CSMA CSMA
65
Algorithm Summary
 Set-up phase
 Node n choosing a random number m between 0 and 1
 If m < T(n) for node n, the node becomes a cluster-head where

 P
 if n  G
T ( n )   1  P [r * mod(1/ P )]
0 otherwise ,


 where P = the desired percentage of cluster heads (e.g., P= 0.05), r=the c

urrent round, and G is the set of nodes that have not been cluster-heads in
the last 1/P rounds. Using this threshold, each node will be a cluster-head
at some point within 1/P rounds. During round 0 (r=0), each node has a p
robability P of becoming a cluster-head.
66
Algorithm Summary (cont.)
 Set-up phase
 Cluster heads assign a TDMA schedule for their members w
here each node is assigned a time slot when it can transmit.
 Each cluster communications using different CDMA codes t
o reduce interference from nodes belonging to other clusters.
 TDMA intra-cluster
 CDMA inter-cluster
 Spreading codes determined randomly
 Broadcast during advertisement phase

67
Algorithm Summary (cont.)
 Steady-state phase
 All source nodes send their data to their cluster heads
 Cluster heads perform data aggregation/fusion through local
transmission
 Cluster heads send aggregated data back to the BS using a si
ngle direct transmission

68
An Example of a LEACH Network
 While neither of these diagrams is the optimum scenario, the se
cond is better because the cluster-heads are spaced out and the
network is more properly sectioned

Good case scenario Bad case scenario

Node
Cluster-Head Node
X Node that has been cluster-head in the last 1/P rounds
69 Cluster Border
Conclusions
 Advantages
 Increases the lifetime of the network
 Even drain of energy
 Distributed, no global knowledge required
 Energy saving due to aggregation by CHs

 Disadvantages
 LEACH assumes all nodes can transmit with enough power t
o reach BS if necessary (e.g., elected as CHs)
 Each node should support both TDMA & CDMA
 Need to do time synchronization
 Nodes use single-hop communication

70
Reference
 W. Heinzelman, A. Chandrakasan, and H. Balakrishnan, “Ener
gy-efficient communication protocol for wireless sensor netwo
rks,” Proceedings of the 33rd Hawaii International Conference
on System Sciences, January 2000.

71
 Chapter 4.4
Location Based Routing

72
Overview
 Sensor nodes are addressed by means of their locations.
 The distance between neighboring nodes can be estimated on the basis of
incoming signal strengths.
 Relative coordinates of neighboring nodes can be obtained by exchangin
g such information between neighbors.
 To save energy, some location based schemes demand that nod
es should go to sleep if there is no activity.
 More energy savings can be obtained by having as many sleepi
ng nodes in the network as possible.
 Hereby, two important location based routing protocols, GEAR
and GPSR, are introduced.
 Geographical and Energy Aware Routing (GEAR)
 Greedy Perimeter Stateless Routing (GPSR)
73
 4.4.1
GEAR
Geographical and Energy Aware Routing

74
Geographical and Energy Aware Routing
(GEAR)
 The protocol, called Geographic and Energy Aware Routing (G
EAR), uses energy aware and geographically-informed neighb
or selection heuristics to route a packet towards the destination
region.
 The key idea is to restrict the number of interests in directed dif
fusion by only considering a certain region rather than sending
the interests to the whole network. By doing this, GEAR can co
nserve more energy than directed diffusion.
 The basic concept comprises of two main parts
 Route packets towards a target region through geographical a
nd energy aware neighbor selection
 Disseminate the packet within the region

75
Energy Aware Neighbor Computation
 Each node N maintains state h(N, R) which is called learned co
st to region R, where R is the target region
 Each node infrequently updates neighbor of its cost
 When a node wants to send a packet, it checks the learned cost
to that region of all its neighbors
 If a node does not have the learned cost of a neighbor to a regio
n, the estimated cost is computed as follows:
c(Ni, R) = αd(Ni, R) + (1-α)e(Ni)
where
α = tunable weight, from 0 to 1.
d(Ni, R) = normalized the largest distance among neighbors of N
e(Ni) = normalized the largest consumed energy among neighbors
of N
76
Energy Aware Neighbor Computation (con
t.)
 When a node wants to forward a packet to a destinatio
n, it checks to see if it has any neighbor closer to desti
nation than itself
 In case of multiple choices, it aims to minimize the lea
rned cost h(Nmin, R)
 It then sets its own cost to:
h(N, R) = h(Nmin, R) + c(N, Nmin)
c(N, Nmin) = the transmission cost from N and Nmin

77
Forwarding Around Holes

K L T

B–T= 5 C–T=2
x x
F G xH I J

h(C,T) = h(B,T)+c(C,B)
5
A B C D E

α is set to 1. Initially, at time 0, at node S, among all neighbors of S, B, C, D

are closer to T than S. h(B,T)=c(B,T)= 5 , h(C,T)=c(C,T)=2, h(D,T)=c(D,T)= 5 .
After that , h(C, T) = h(B, T) + 1

78
Recursive Geographic Forwarding
 Once the target region is reached, the packets are disseminated
within the region by recursive geographic forwarding
 Forwarding stops when a node is the only one in a sub-region

Ni

79
Recursive Geographic Forwarding (cont.)
 When network density is low recursive geographic for
warding is subject to two pathologies: inefficient trans
missions and non-termination

80
Recursive Geographic Forwarding (cont.)
 Inefficient Transmission
 Recursive geographic forwarding vs. Restricted flooding

Restricted flooding 1 time for A

Recursive Geographic
sending and 4 times for receiving
Forwarding 3 times for sending
= consuming
and 3 times for receiving =
5 units of energy
consuming 6 units of energy C

E B

F
81
Recursive Geographic Forwarding (cont.) P
athologies

 Non-Termination
 In the recursive geographic forwarding protocol, packet forwa
rding terminates when the target subregion is empty.

E H

A C
L
F
82
Recursive Geographic Forwarding (cont.) P
roposed solution for pathologies
 Solution:
 Node degree is used as a criteria to differentiate low d
ensity networks from high density ones
 Choice of restricted flooding over recursive geographi
c forwarding if the receiver’s node degree is below a t
hreshold value

83
Conclusion
 GEAR strategy attempts to balance energy consumpti
on and thereby increase network lifetime
 GEAR performs better in terms of connectivity after i
nitial partition

84
References
 Y. Yu, D. Estrin, and R. Govindan, “Geographical and Energy-Aware
Routing: A Recursive Data Dissemination Protocol for Wireless Sens
or Networks”, UCLA Computer Science Department Technical Repor
t, UCLA-CSD TR-01-0023, May 2001.
 Nirupama Bulusu, John Heidemann, and Deborah Estrin. “Gps-less l
ow cost outdoor localization for very small devices”. IEEE Personal
Communications Magazine, 7(5):28-34, October 2000.
 L. Girod and D. Estrin. “Robust range estimation using acoustic and
multimodal sensing”. In IEEE/RSJ International Conference on Intel
ligent Robots and Systems (IROS 2001), Maui, Hawaii, October 2001
.
 Nissanka B. Priyantha, Anit Chakraborty, and Hari Balakrishnan. “T
he cricket location-support system”. In Proc. ACM Mobicom, Boston
, MA, 2000.
 Andreas Savvides, Chih-Chieh Han, and Mani B. Strivastava. “Dyna
mic fine-grained localization in adhoc networks of sensors”. In Proc.
ACM Mobicom, 2001.
85
 4.4.2
GPSR
Greedy Perimeter Stateless Routing

86
Greedy Perimeter Stateless Routing (GPS
R)
 Greedy Perimeter Stateless Routing (GPSR) proposes the aggr
essive use of geography to achieve scalability
 GEAR was compared to a similar non-energy-aware routing pr
otocol GPSR, which is one of the earlier works in geographic r
outing that uses planar graphs to solve the problem of holes
 In case of GPSR, the packets follow the perimeter of the planar
graph to find their routes
 Although the GPSR approach reduces the number of states a n
ode should keep, it has been designed for general mobile ad ho
c networks and requires a location service to map locations and
node identifiers.

87
Algorithm & Example
 The algorithm consists of two methods:
greedy forwarding + perimeter forwarding

 Greedy forwarding, which is used wherever possible,

and perimeter forwarding, which is used in the regions
greedy forwarding cannot be done.

88
Greedy Forwarding (cont.)
 Under GPSR, packets are marked by their originator
with their destinations’ locations
 As a result, a forwarding node can make a locally opti
mal, greedy choice in choosing a packet’s next hop
 Specifically, if a node knows its radio neighbors’ posit
ions, the locally optimal choice of next hop is the neig
hbor geographically closest to the packet’s destination
 Forwarding in this scheme follows successively closer
geographic hops, until the destination is reached

89
Greedy Forwarding (cont.)

90
Greedy Forwarding (cont.)
 A simple beaconing algorithm provides all nodes with
their neighbors’ positions: periodically, each node tran
smits a beacon to broadcast MAC address, containing
its own identifier (e.g., IP address) and position
 Position is encoded as two four-byte floating point qu
antities, for x and y coordinate values
 Upon not receiving a beacon from a neighbor for long
er than timeout interval T, a GPSR router assumes that
the neighbor has failed or gone out-of-range, and delet
es the neighbor from its neighbor table

91
Greedy Forwarding (cont.)
The Problem of Greedy Forwarding

D |xD|<|wD|and|yD|
x will not choose to
forward to w or y
v z using greedy
forwarding
void

w y
x x
x

92
The Right-Hand Rule: Perimeters
 In previous works, use the right-hand rule to map perimeters b
y sending packets on tours of them. The state accumulated in th
ese packets is cached by nodes, which recover from local maxi
ma in greedy forwarding by routing to a node on a cached peri
meter closer to the destination.
 This approach requires a heuristic, the no-crossing heuristic, to
force the right-hand rule to find perimeters that enclose voids i
n regions where edges of the graph cross

x z

y
93
Right-Hand Rule Does Not Work with
Cross Edges

u
D

 x originates a packet to u
w
 Right-hand rule results in the
tour x-u-z-w-u-x
x

94
Remove Crossing Edge

u
v

w
Make the graph planar
Remove (w,z) from the
x graph
 Right-hand rule results in
the tour x-u-z-v-x
95
Make a Graph Planar
 A graph in which no two edges cross is known as plan
ar. A set of nodes with radios, where all radios have id
entical, circular radio range r, can be seen as a graph:
each node is a vertex, and edge (n, m) exists between
nodes n and m if the distance between n and m, d(n,
m)≦r.
 Convert a connectivity graph to planar non-crossing g
raph by removing “bad” edges
 Ensure the original graph will not be disconnected
 Two types of planar graphs:
 Relative Neighborhood Graph (RNG)
96
 Gabriel Graph (GG)
Planarized Graphs (cont.)
Relative Neighborhood Graph (RNG)

w
u v

97
Planarized Graphs (cont.)
Gabriel Graph (GG)

w
u v

98
Planarized Graphs (cont.)
 An algorithm for removing edges from the graph that
are not part of the RNG or GG would yield a network
with no crossing links
 The RNG is a subset of the GG
 It is because RNG removes more edges
 Hereby, the RNG is used
 If the original graph is connected, RNG is also connec
ted

99
Connectedness of RNG Graph
 Key observation
 Any edge on the minimum spanning tr
ee of the original graph is not removed
 Proof by contradiction: Assume
(u,v) is such an edge but removed in R
NG

u
v

100
 Planarized Graphs (cont.)
Relative
Original Gabriel Graph (GG) Neighborhood Graph
(RNG)

The full graph of a radio

The GG subset of The RNG subset of the
network, 200 nodes, uniformly
the full graph full and GG graphs.
randomly placed on a 2000 x
2000 meter region, with a radio
range of 250 m.

101
Combining Greedy and Planar Perimeters
 All data packets are marked initially at their originators as gree
dy mode
 GPSR packet headers include a flag field indicating whether th
e packet is in greedy mode or perimeter mode
 Packet sources also include the geographic location of the desti
nation in packets
 Only a packet’s source sets the location destination field, it is le
ft unchanged as the packet is forwarded through the network
 Upon receiving a greedy-mode packet for forwarding, a node s
earches its neighbor table for the neighbor geographically close
st to the packet’s destination
 When no neighbor is closer, the node marks the packet into per
imeter mode
102
GPSR

greedy fails

Greedy Forwarding Perimeter Forwarding

have left local maxima

greedy works greedy fails

103
Combining Greedy and Planar Perimeters
(cont.)

Lf
e0

If forwarding node to D < Lp to D,

x returns a packet to greedy mode
Lp

104
Conclusion

 GPSR’s benefits all stem from geographic routing’s us

e of only immediate-neighbor information in forwardi
ng decisions.
 GPSR keeps state proportional to the number of its nei
ghbors, while both traffic sources and intermediate DS
R routers cache state proportional to the product of the
number of routes learned and route length in hops.

105
References
 B. Karp and H. T. Kung, “Greedy Perimeter Stateless Routing f
or Wireless Networks”, Proc. 6th Annual ACM/IEEE Int'l. Con
f. Mobile Comp. Net., Boston, MA, pp. 243-54, August 2000.
 G. G. Finn, “Routing and addressing problems in large metrop
olitan-scale internetworks”, Tech. Rep. ISI/RR-87-180, Inform
ation Sciences Institute, March 1987.
 S. Floyd and V. Jacoboson, “The synchronization of periodic ro
uting messages”, IEEE/ACM Transactions on Networking, Vol.
2, pp. 122-136, April 1994.
 B. Karp “Greedy perimeter state routing”, Invited Seminar at t
he USC/Information Sciences Institute, July 1998.
 J. Saltzer, D. P. Reed, and D. Clark, “End-to-end arguments in
system design”, ACM Transactions on Computer Systems, Vol.
2, No. 4, Pages: 277-288, November 1984.

106
 Chapter 4.5
QoS Based Routing

107
Overview
 QoS is the performance level of service offered by a n
etwork to the user.
 The goal of QoS is to achieve a more deterministic net
work behavior so that the information carried by the n
etwork can be better delivered and the resources can b
e better utilized.
 In QoS-based routing protocols, the network has to ba
lance between energy consumption and data quality.
 In particular, the network has to satisfy certain QoS m
etrics, e.g., delay, energy, bandwidth, etc. when delive
ring data to the BS.
108
Parameters of QoS Networks
 Different services require different QoS parameters
 Multimedia
 Bandwidth, delay jitter & delay
 Emergency services
 Network availability
 Group communications
 Battery life

 Generally the parameters that are important are:

 bandwidth
 delay jitter
 battery charge
 processing power
 buffer space
109
Challenges in QoS Routing
 Dynamically varying network topology
 Imprecise state information
 Lack of central coordination
 Hidden node problem
 Limited resource
 Insecure medium

110
 4.5.1
TBP (Ticket-Based Probing)
QoS of Bandwidth

111
Ticket-Based Probing
 Distributed multi path QoS routing scheme
 Bandwidth-constrained routing and delay-constrained routing
 There are numerous paths from source to destination, w
e shall not randomly pick several paths to search
 We shall not use any flooding path-discovery approach
es, which may send routing messages to the entire netw
ork
 Multipath search is tolerant to imprecise information
 We want to make an intelligent hop-by-hop path selecti
on to guide the search along the best candidate paths

112
Ticket-Based Probing (cont.)

113
Ticket-Based Probing (cont.)
 A ticket is the permission to search one path. The sour
ce node issues a number of tickets based on the availab
le state information
 Utilizes tickets to limit the number of paths searched d
uring route discovery
 A ticket is the permission to search a single path
 More tickets, more QoS constraints are required
 Probes (routing messages) are sent from the source to
ward the destination to search for a low-cost path that
satisfies the QoS requirement
 Each probe is required to carry at least one ticket

114
Ticket-Based Probing (cont.)

i
P1(1) P1
S (1)
P2(
2) )
P4(2
D

3)
j P

P3(
3(3
)
k

115
Ticket-Based Probing (cont.)
A
Demand = 3

3 3
S
D
3
C x2
3 5
2 2

6 E
B
116
 Ticket-Based Probing (cont.)
Demand = 4 A
(1.1,3) (1.1,3)
3 3
S
D
C
3 2
2 5
(1.2,1) 2 2 (1.2,1)

6 E
B
(1.2,1)
117
 Ticket-Based Probing (cont.)
A
Demand = 4

3 3
S (1.1,3) (1.1.1,2) D
C
3 2
(1.1.2,1)
2 (1.1.2,1) 5
(1.2,1) 2 2 (1.2,1)

6 E
B (1.2,1)
118
 Ticket-Based Probing (cont.)
T1 T1
D D
S S
T2

T2

T1
D
S T2
x

119
 Ticket-Based Probing (cont.)
Demand = 4 A
(1,4)
3
x
4
S
(2.1,3) D
C
3 2
x2 (2.1,3) 5
(2.1,3)
(2.2,1) 4 3 (2.2,1)

6 (2.1,3)
E
B (2.2,1)

120
Conclusion
 The routing overhead is controlled by the number of tickets, w
hich allows the dynamic tradeoff between the overhead and the
routing performance. Issuing more tickets means searching mo
re paths, which results in a better chance of finding a feasible p
ath at the cost of higher overhead.
 A distributed routing process is used to avoid any centralized p
ath computation that could be very expensive for QoS routing i
n large networks.
 This approach not only increases the chance of success but also
improves the ability to tolerate the information imprecision bec
ause the intermediate nodes may gradually correct a wrong dec
ision made by the source.

121
Conclusion (cont.)
 Ticket-based probing scheme achieves a balance between the si
ngle-path routing algorithms and the flooding algorithms. It do
es multipath routing without flooding.
 The basic idea is to achieve a near-optimal performance with m
odest overhead by using a limited number of tickets and makin
g intelligent hop-by- hop path selection.

122
References
 S. Chen and K. Nahrstedt, “On finding multi-constrained paths,” in Pro
c. IEEE ICC’98, pp. 874-879.
 R. Guerin and A. Orda, “QoS-based routing in networks with inaccurate
information: Theory and algorithms,” in Proc. IEEE INFOCOM’97, Jap
an, pp. 75-83.
 Q. Ma and P. Steenkiste, “Quality-of-service routing with performance g
uarantees,” in Proc. 4th Int. IFIP Workshop Quality of Service, May 199
7, pp. 115-126.
 Z. Wang and J. Crowcroft, “QoS routing for supporting resource reserva
tion,” IEEE J. Select. Areas Commun., Sept. 1996.
 S. Chen and K Nahrstedt, “Distributed Quality-of-Service Routing in A
d Hoc Networks,” IEEE J. Select. Areas Commun, vol.17, no. 8, pp. 14
88-1505, Aug. 1999.

123
References
 T. Hea, J. A Stankovic, C. Lu, and T. Abdelzaher, “SPEED: a st
ateless protocol for real-time communication in sensor network
s,” in Proc. IEEE International Conference on Distributed Com
puting Systems, pp. 46-55, May 2003.
 G. S. Ahn, A. T. Campbell, A. Veres, and L.H. Sun. “SWAN: S
ervice Differentiation in Stateless Wireless Ad Hoc Networks,”
In Proc. IEEE INFOCOM'2002, June 2002.

124