Title: Next Century Challenges: Scalable Coordination in Sensor Networks
1Next Century Challenges Scalable Coordination in
Sensor Networks
- Deborah Estrin, Ramesh Govindan, John Heidemann,
Satish Kumar - (Some images and slides adopted from Santhosh R
Thampuran - CMU)
2Outline
- Characteristics of sensor devices.
- Motivating applications.
- Key requirements of a sensor network and
differences with current networks. - Localized algorithms for coordination.
- Directed Diffusion a model for describing
localized algorithms.
3Characteristics of Sensor Devices
- Ability to monitor a wide variety of ambient
conditions - temperature,
- pressure,
- mechanical stress level on attached objects
- Will be equipped with significant processing,
memory, and wireless communication capabilities.
4Applications Environmental Analysis
5Applications Contaminant Flow Monitoring
6Applications Traffic Control
- Sensor attached to every vehicle.
- Capable of detecting their location, vehicle
sizes, speeds and densities road conditions - Alternate routes, estimate trip times
7Applications Biological Systems
8Key Requirements
- These futuristic scenarios bring out two key
requirements of sensor networks - support for very large numbers of unattended
autonomous nodes. - adaptivity to environment and task dynamics.
9Differences with Current Networks
- Sensor Networks ratio of communicating nodes to
users is much greater. - extremely difficult to pay special attention to
any individual node. - Sensors may be inaccessible
- embedded in physical structures.
- thrown into inhospitable terrain.
10Differences with Current Networks
- There are large scale unattended systems, today.
- Automated factories are deployed with very
careful planning and react to very few external
events.
11Differences with Current Networks
- Sensor networks deployed in very ad hoc manner.
- They will suffer substantial changes as nodes
fail battery exhaustion, accidents new nodes
are added nodes move. - User and environmental demands also contribute to
dynamics.
12Overall Design of Sensor Networks
- Is it sufficient to design sensor network
applications using Internet technologies coupled
with ad-hoc routing mechanisms? - Data-Centric Application-Specific.
- Sensor network coordination applications are
better realized using localized algorithms
distributed as opposed to centralized. - scales with increase in network size, robust to
network partitions and node failures.
13Localized Algorithms for Coordination
- Clustering efficient coordination.
14Localized Clustering Algorithm
- For every sensor, level ? radius
- Advertisement hierarchical level, parent ID,
remaining energy
C
D
B
E
A
wait time
15Localized Clustering Algorithm
- Start promotion timer if no parent.
- Promotion timer inv prop (remaining energy,
number of other sensors from whom level 0 adv was
received)
C
D
B
E
A
promotion timer
16Localized Clustering Algorithm
- Periodic advertisements at the level 1 radius.
- Advertisement B,C,E
C
D
B
E
A
level 1 sensor
17Localized Clustering Algorithm
- Two key design constraints
- asymmetric communication in the network.
- limited energy of sensors.
18Application of Clustering Algorithm
- Aim To pinpoint in an energy-efficient manner,
the exact location of objects. - Accuracy widest possible measurement baseline.
- Energy efficiency fewest number of sensors
participating in the triangulation.
19Triangulation
Z
A
- Determine position in space.
- Can specify approx direction of object relative
to its own location.
20Base-line Estimation
21Advantages of Cluster-based Approach
- Sensor algorithms only use local information.
- generally lower energy consumption in comparison
to global communication. - Robust to link or node failures and network
partitions - mechanisms for self-configuration can be simpler.
22Advantages of Cluster-based Approach
- Local communication and per-hop data filtering
- avoid transmitting large amounts of data over
long distances. - preserving node energy resources.
- Node energy resources are better utilized
- cluster-heads adapt to changing energy levels.
23Disadvantage of Cluster-based Approach
- Non-optimal under certain terrain conditions.
24Several Sensors Electing Themselves
Obstacle
Allow a cluster-head to switch on some number of
child sensors in its cluster to do object
location.
25Adaptive Fidelity Algorithms
Z
Y
A
quality of the answer can be traded against
battery lifetime, network bandwidth, or number of
active sensors.
26Tradeoffs
- Localized algorithms exhibit good robustness and
scaling properties.
- May sacrifice resource utilization or sensing
fidelity, responsiveness, or immunity to
cascading failures.
27Directed Diffusion
- A novel data-centric, data disemmination paradigm
for sensor networks. - Data generated by sensor node is named using
attribute-value pairs. -
- A sensing task is disseminated throughout the
sensor network as an interest for named data.
28Directed Diffusion
- This dissemination sets up gradients within the
network designed to "draw" data matching the
interest. - Events start flowing towards the originators of
interests along multiple paths. The sensor
network reinforces one, or a small number of
these paths.
29Directed Diffusion
30Directed Diffusion
- Allows intermediate nodes to cache or locally
transform data. - leverages the application-specificity that is
possible in sensor networks. - The diffusion models data naming and local data
transformation features capture the
data-centricity and application-specificity
inherent in sensor networks.
31Related Work
- Ad-hoc Networks
- Proactive vs. reactive routing protocols
- Energy-efficiency issues
- Distributed Robotics
- Robots cooperate to discover entire map
- Internet Multicast and web caching
- Lightweight session
32Current Developments
- Smartdust project
- cubic millimeter sensors
- Sensors float in air like dust
- WINS (wireless integrated wireless Sensors)
- WSN (Wireless Sensing Network)
- Odyssey
- Habitat monitoring
- The Cricket Indoor Location System