Wireless sensor networks in
autonomic environments
Shuping Liu
Networking Lab
HUT
Agenda
 What
is sensor?
 What is WSN?
 Communication topology
 Why we need it?
 WSN special characters
 Some filed interesting
 A programmable routing for autonomic WSN
 Data dissemination in autonomic WSN
What is sensor?
What is WSN?
Habitat monitoring on the Great Duck Island (USA Maine.
2002/2003. UCB & Intel & Atlantic Univ.)
Communication topology
Why we need it?
Seamless and Ubiquitous communication with the
real world.
 Wide usages
Military application
Environmental application
Health application
Home application
Traffic Surveillance
Other commercial applications
……

WSN special characters
Person unattended, inaccessible  autonomic
 Limited resource: power, memory, MPU…
 Topology changes / breaks frequently (war field,
etc.)
 High density employed, broadcast communication
paradigm (normal ad-hoc networks uses point-topoint communications)
 Must be self-organized, self-maintaining and
operate at low duty cycle.

Some fields interesting
Efficient routing
 Data dissemination
 Low power
 Security
 Programming the Ensemble (configuration)

 WSN is a new field, especially in autonomic
environment. I will introduce some work in the first
two topics in autonomic WSN
A programmable routing for
autonomic WSN (1/19)
 The
goal of a WSN is to collect, process,
and forward sensed data to other sensor
nodes and/or base stations.
 Therefore, the proper routing algorithm is
essential to WSN applications, which must
be lightweight, due to limited available
resources.
A programmable routing for
autonomic WSN (2/19)

Several existing routing for WSN
GPSR: Greedy Perimeter Stateless Routing
GEAR: Geographical Energy Aware Routing
TBF: Trajectory Based Routing
DD:
Directed-Diffusion
TTDD: Two-Tier Data-Dissemination
CBM: Content-Based Multicast
RR:
Rumor-Routing
……
A programmable routing for
autonomic WSN (3/19)



These routings have distinct properties,
Try to meet the resource-limited requirements
Different from traditional routing such as OSPF,RIP,BGP
(see table 1 for difference in details)
Each of them is designed to meet specific goals and
therefore is not efficient for all applications.
e.g. DD is more energy-efficient than TTDD when the
number of sink nodes is large, while TTDD is better
when the number of sink nodes is small
There is a need to have a routing for WSN that can
adapt to different applications and different network
conditions  autonomic WSN
A programmable routing for
autonomic WSN (4/19)
Table 1. Comparison of routing services in WSN and Traditional Networks
A programmable routing for
autonomic WSN (5/19)



Currently, it is very difficult, if not impossible, to change a
routing service in a large WSN because the service is
statically pre-configured into each node, which is often
unattended.
Yu He et. al. in USC propose a programmable routing for
autonomic WSN.
Their work includes a universal routing service and an
autonomic deployment service.
A programmable routing for
autonomic WSN (6/19)




The universal routing service allows the introduction of different
services through its tunable parameters and programmable
components.
The deployment service completes the configuration of the universal
routing service throughout a WSN in an autonomic and energy-efficient
way.
Through this deployment service, a self-configuration ability is realized
for sensor routing service.
With the changeable parameters and programmable components of
the universal routing service, the self-optimizing as well as other
autonomic abilities can be explored.
A programmable routing for
autonomic WSN (7/19)
Sensor node
architecture
with
programmable
routing
A programmable routing for
autonomic WSN (8/19)
Table 2. shows the data-forwarding and state-collecting functions of the
existing routing is covered by the programmable structure
A programmable routing for
autonomic WSN (9/19)


The suggested architecture is proposed to cover all
existing routing services and to introduce new services
for WSN.
The state information is a list of neighbor entries, each of
which consists of four parts,
Neighbor description (id, location, direction, distance,
energy reading, etc.)
Neighbor interest (type, rate, duration, etc.)
Neighbor data availability (type, duration, etc.)
Neighbor’s latest data copy (data, timestamp, etc.)
 the above state involves only local information and
thus is scalable
A programmable routing for
autonomic WSN (10/19)

Different packets are used to collect each part of the state
information, (state collecting)
Neighbor description – hello / announcement / query
packets
Neighbor interest – query packets
Neighbor data availability – announcement / data packets
Neighbor latest data copy – data packets
A programmable routing for
autonomic WSN (11/19)


The deployment service receives deployment packets that
contain parameters or modules of the programmable
routing services and deploy services according to packet
content.
There are three levels of deployment,
(1) the deployment service only changes parameters to
the state-collecting and/or data-forwarding modules. (least
bandwidth requirement  relatively frequently)
(2) either of the two modules is replaced. (middle case)
(3) the entire routing service is changed.
(most overhead  only occasionally)
A programmable routing for
autonomic WSN (12/19)

Note that this deployment service allows different routing
services to reside in different parts of WSN. For example,
GPSR service and RR (Rumor-Routing) service can be
deployed in heterogeneous parts of a WSN.
A programmable routing for
autonomic WSN (13/19)




Now let us consider a case with complex routing service,
then we will deploy a large code.
Transferring the large routing code can be very expensive
in WSN where energy is a very scarce resource.
A. Boulis et. al. proposed a separate running environment
for deployment service. But it is computation inefficiency.
A deployment approach for routing services should be both
energy-efficient and computation-efficient.
A programmable routing for
autonomic WSN (14/19)


The approach proposed by Yu He. et. al. is to move a part
of routing service code into WSN, which contains common
routing services operations and is designed as a shared
library, before deploying routing service modules.
With shared library, the written routing modules have small
code size while keeping the computation efficiency.
A programmable routing for
autonomic WSN (15/19)
A sample
node
architectu
re with
shared
library
A programmable routing for
autonomic WSN (16/19)


The deployment discussed above assumes that all nodes
in a network can be reached at one time.
But this is generally not the case for WSN because,
Sensor node is prone to fail due to running out of energy
Communication failure due to lossy channel or obstacles
Sensor node sleep periodically or dynamically for some
time due to energy-saving mechanisms
 inconsistency among nodes for deployed services
A programmable routing for
autonomic WSN (17/19)



Yu He et. al. proposed a synchronization protocol that
enables a sensor node to make itself consistent with its
neighbors in an energy-efficient way.
Each node runs this protocol after waking up from sleeping
or after a period.
Each node maintains a version number for each deployed
component (a parameter or a module).
A programmable routing for
autonomic WSN (18/19)

A broadcasts an initial request among its neighbors
 each neighbors Ni with greater version number starts a
timer
after timeout, the neighbor sends an initial reply to A
A also starts a timer after sending initial request
A sends formal request to the node with higher version
number
 reply with formal reply
 complete synchronization
A programmable routing for
autonomic WSN (19/19)
Deployment
synchronizati
on from
neighbors
Data dissemination in
autonomic WSN (1/5)

In WSNs, data communication, from the point of
view of the communication entities, can be divided
into three cases,
From sensor to a monitoring node
Among neighboring sensors
From a monitoring node to sensors
Data dissemination in
autonomic WSN (2/5)
Data communication schemes in WSNs
Data dissemination in
autonomic WSN (3/5)

Reliable data dissemination is crucial to WSN since
a monitoring node has to perform some specific
activities, such as
Change the operational mode of part or entire WSN
Broadcast a new interest to the network
Activate / deactivate one or more sensors
Send queries to the network
……
Data dissemination in
autonomic WSN (4/5)
Max do Val Machado et. al. proposed a new data
dissemination algorithm, TEDD (Trajectory and
Energy-based Data Dissemination).
 The key idea is to combine concepts presented in
TBF (Trajectory-Based Forwarding) with the
information provided by the energy map of the
network to determine routes in a dynamic fashion,
according to the energy level of the sensor nodes.

Data dissemination in
autonomic WSN (5/5)
 Simulation
result revealed that the energy
spent with the data dissemination activity
can be concentrated on nodes with highenergy reserves, whereas low-energy node
can use their energy only to perform sensing
activity or to receive information addressed
to them.
Thanks!
Any comments and questions?