Sie sind auf Seite 1von 4

INSIGHT# 2007-24M

MAY 10, 2007

High Hopes for "Low PAN"

By Harry Forbes

Keywords
Wireless Sensor Networks, 6LowPAN, IETF, SP100

Summary

Wireless sensing technology for manufacturing is converging in some areas


while other areas struggle to avoid further fragmentation. At the physical,
or radio layer (layer 1 in the OSI model) all of the major standards groups
have converged on the IEEE 802.15.4 standard. At the data link and net-
work layers (OSI layers 2-3), the picture is very
The Internet Engineering Task Force has
developed and will soon issue a standard different. Now a new technology for using IP
method for using Internet Protocol on on sensor networks, named "6LowPAN" is ap-
wireless sensor networks. Several proaching standardization by the IETF. There is
companies as well as the ISA SP100 group some support within the SP100 for using
are now wrestling with the implications. 6LowPAN as a standard for sensor networks.

Analysis

Wireless sensing technology for manufacturing is converging in some


areas. At the physical, or radio layer (layer 1 in the OSI model) all of the
major standards groups have converged on the IEEE 802.15.4 standard.
ZigBee was the first to adopt this radio in 2003, followed by the HART
Wireless working group in 2006. Last month the ISA SP100 working group
voted to use the latest update to this standard as the sensor network in its
initial standard for wireless sensing in manufacturing.

At the data link and network layers (OSI layers 2-3), the picture is very dif-
ferent. The ZigBee network layer was designed to be a general purpose
network, but it has been consistently outperformed in specific applications
by one or more of the many proprietary sensor network layers in the mar-
ket. Yet the ISA SP100 working group needs to choose and specify a sensor
network layer. For end users of wireless devices, this network-level frag-
mentation is unsatisfying, but it is less disruptive than fragmenting radio

THOUGHT LEADERS FOR MANUFACTURING & SUPPLY CHAIN


ARC Insights, Page 2

technologies because radio interference is a physical layer phenomenon.


Numerous 802.15.4 radio systems could share their medium even if they
did not share the same network protocol.

Will IP Make for a True “Internet of Things”?


Given this fragmentation of wireless sensor networks, there is a delicious
irony in some pundits loudly announcing the arrival of an “Internet of
Things”. The sole feature shared by all the “things” that have been at-
tached to the Internet since its DARPA days is the Internet Protocol (IP).
However, IP has not been suitable for wireless sensors and today virtually
none of them, commercial or academic, use IP.
It is ironic for pundits to proclaim the era
of an "Internet of Things" when the things As of today, the “Internet of Things” begins and
spoken of do not support the Internet ends at the “thing” that translates to/from IP for
Protocol – the sole common element of all the wireless sensor network – the sensor net-
the things now on the global Internet. work gateway.

In March 2007, the Internet Engineering Task Force (IETF) designated a


proposed standard that defines a method for the Internet Protocol (IP) to
work as a network layer for these same sensor radios. The IETF work is
novel because the idea of using IP in sensor networks has been considered
but dismissed since their earliest days. Its use remains controversial.

The IETF working group is dubbed “6LowPAN”. The group has proposed
a method for running Internet Protocol Version 6 (IPV6) on low-power,
low-data rate Personal Area Networks (PAN) – hence the Working Group’s
exotic name. However, if IP version 4, which is ubiquitous, is unsuited for
sensor networks, and general-purpose solutions like ZigBee still cannot
compete with proprietary sensor network technology, why should a proto-
col like IPV6 be likely to succeed?

Skeptics argue that IPV6 has so much overhead that there would be hardly
any data capacity remaining on a sensor network for the information payl-
oad. Furthermore, they say IP in any form requires too much memory for
sensor nodes, and the supporting code base and networking applications
are far too large to reside in sensor nodes. Deterministic message delivery
of such a network is also questioned. Finally, there is the question of how
much energy the wireless sensor will consume during service – a critical
question for networks that consist primarily of battery-powered devices, as
would be found in future process manufacturing plants.

©2007 • ARC • 3 Allied Drive • Dedham, MA 02026 USA • 781-471-1000 • ARCweb.com


ARC Insights, Page 3

That is a fairly long list of potential liabilities. However, there is also a set
of counter arguments for the use of IP technology. One strong argument is
Moore’s Law. Wireless sensor nodes will certainly have far more memory
and processing capacity as time progresses. The IETF working group has
also worked to use available properties of IPV6 such as header compression
in order to reduce the protocol’s overhead in the smaller sensor networks
that people are deploying today. Another advantage is the ability of sen-
sors to directly address messages to anywhere on the global Internet.

The more questionable difficulties are performance and power consump-


tion. Improvements in power consumption and batteries have dramatically
improved the lifetime specifications of the most recent designs of wireless
field devices, but these remain exclusively to devices that operate on a
power budget. In theory, one could use IP as a common wireless carrier for
any number of different application layers, as is
Despite great improvements over the past
now the case with Ethernet in factory automa-
3 years, wireless field devices remain on a
tion. However, the performance levels of such
very tight power budget.
as design remain unproven.

Advocates of 6LowPAN claim that it can be implemented in a “Pay-as-you-


go” fashion. They say that smaller and simpler networks will not use sig-
nificantly more power than with proprietary solutions. This is a result, they
claim, of properties in IPV6 that make it possible to minimize node
processing and protocol overhead in comparison with IPV4.

There are companies and academics working in this area now. Two ven-
ture-stage firms have added IPV6 options to their sensor network products.
One is Arch Rock Corporation founded by technologists from Cal Berkeley.
Arch Rock contributed to 6LowPAN and is the primary advocate for using
6LowPAN in sensor networks. The second is Augusta Systems, whose
work is primarily for the US Departments of Defense and Homeland Secu-
rity. There are other sensor network firms whose offering seems to be tilted
toward using IP as a future sensor networking solution. The sensor net-
work technology hothouse of South Korea is a center of academic research
into 6LowPAN. The ISA SP100 working group has been given a presenta-
tion on the IETF standardization of 6LowPAN.

Integration Options
There are two key questions concerning integrating 6LowPAN devices with
existing systems. First is the translation to/from IPV6 and IPV4, since

©2007 • ARC • 3 Allied Drive • Dedham, MA 02026 USA • 781-471-1000 • ARCweb.com


ARC Insights, Page 4

many existing TCP/IP networks run IP version 4. This area is really a ge-
neric question that will have many commercial solutions that can be
incorporated in a straightforward fashion, since portions of the global In-
ternet now run IPV6 and almost all of today’s network infrastructure is
already capable of supporting IPV6.

Routing is another question area. Existing IP


IP V4 IP V6 routing protocols are not suitable for power-
Network Network
constrained sensor networks. The IETF has char-
tered a working group to develop routing
protocols specifically for sensor networks, but
Sensor Network
IPV4-IPV6 Routing Gateway this will take some time, and many solutions to
IPV6 Services Layer the sensor network routing problem have already
Sensor Network Data Link
been developed. The most intriguing answer is
to use the existing sensor network routing proto-
cols at the data-link layer in 6LowPAN sensor
networks. This way the IP network could ab-
stract the entire sensor network as a slow but
Sensor Network
(Mesh, Star, or Tree) reliable shared link. The data-link layer would
6LowPAN Networks Can Use Existing manage the path used to reach a particular node
Sensor Network Routing Technologies
in a sensor network. This type of solution could
incorporate any of the existing network solutions for sensor networks and
still address each node using IPV6. This type of solution is called “meshing
under”, in that sensor network meshing occurs underneath and out of the
view of the Internet Protocol.

Recommendations

• Process manufacturers should plan to eventually use IEEE 802.15.4 as


the radio for sensor networking applications.

• Suppliers and solution providers should evaluate the performance pe-


nalties of 6LowPAN versus the values of IP standardization.

For further information or to provide feedback on this Insight, please contact your
account manager or the author at hforbes@arcweb.com. ARC Insights are pub-
lished and copyrighted by ARC Advisory Group. The information is proprietary to
ARC and no part of it may be reproduced without prior permission from ARC.

©2007 • ARC • 3 Allied Drive • Dedham, MA 02026 USA • 781-471-1000 • ARCweb.com

Das könnte Ihnen auch gefallen