Transcript Where are the opportunities for ICN in Sensor Networks?

Where are the opportunities for ICN in Sensor
Networks?
Ralph Droms, Distinguished Engineer
Cisco, <insert acronym string here>
© 2015 Cisco and/or its affiliates. All rights reserved.
IRTF icnrg interim meeting, 1/13/2015
1
Why is ICN for Sensor Networks interesting?

Intuitively, data objects are a match for the requirements of
applications using sensor networks, as opposed to
communication endpoints and sessions
 The characteristics of sensors and related network technologies
may not match the requirements of IP very well
 ICN architecture is feasible for sensor networks; where are the
specific advantages in implementation, deployment, operation
and application support?
 If ICN is going to have an impact in the sensor network space, it
has to do something better than current technologies.
© 2015 Cisco and/or its affiliates. All rights reserved.
IRTF icnrg interim meeting, 1/13/2015
2
Opportunities in Wireless Mesh Networking

There are many standards required for an IPv6 wireless
mesh stack.
Does ICN require fewer resources (CPU, memory, energy)
for a stack/application implementation?
Are there opportunities to take advantage of specific
characteristics of the MAC/PHY transport to optimize the
ICN application/stack?
© 2015 Cisco and/or its affiliates. All rights reserved.
IRTF icnrg interim meeting, 1/13/2015
3
Routing and Addressing

Routing IP in a mesh network requires mesh establishment,
ongoing route management and other overhead.
Can ICN name-based routing and object forwarding
reduce overhead by using on-demand or geoloc routing?

There is overhead is associated with address assignment
and management.
Can ICN reduce overhead by using named objects rather
than host addresses?
© 2015 Cisco and/or its affiliates. All rights reserved.
IRTF icnrg interim meeting, 1/13/2015
4
Object Security…

The IP model requires session security, access controls and
other security features in a constrained device.
Can ICN reduce or refactor operational overhead by using
purely object security?
Application
© 2015 Cisco and/or its affiliates. All rights reserved.
IRTF icnrg interim meeting, 1/13/2015
5
Application Layer

Before ICN replaces IP, there will have to be a gateway function
somewhere between ICN sensor networks and IP networks
Can a “secure object” architecture bridge the gap?
IP network
ICN network
Application
Repo
© 2015 Cisco and/or its affiliates. All rights reserved.
IRTF icnrg interim meeting, 1/13/2015
6
Designing for Actuators

…or, more generally, a push model for data object delivery

Not directly related to “what can ICN do better”, but some way is needed
to address applications that currently don’t fit the Interest/Object
message paradigm
Is there an ICN solution for “push” model that is an improvement
over the existing IP models?
© 2015 Cisco and/or its affiliates. All rights reserved.
IRTF icnrg interim meeting, 1/13/2015
7
Backup…
© 2015 Cisco and/or its affiliates. All rights reserved.
IRTF icnrg interim meeting, 1/13/2015
8
Sensor Network characteristics

Constrained devices




Limited computing resources
Automated provisioning
Sleeping devices
Constrained networks

Limited bandwidth
 Dynamic network topology

Tend to generate time series of data, with various scheduling models

Periodic
 Exception
 Polling
© 2015 Cisco and/or its affiliates. All rights reserved.
IRTF icnrg interim meeting, 1/13/2015
9