June 2014
36
I
NTERNET
-
OF
-T
HINGS
Energy release can occur either in short bursts
or as a continuous trickle. In both cases, it typ-
ically needs to be accumulated and often con-
verted (to higher voltage levels) to be usable.
This places significant challenges on the design
of energy harvesting wireless sensor nodes.
Specifically, such devices need to have a very
energy-efficient system design using a very
low duty cycle (devices are sleeping most of
the time) and requiring only extremely low
standby currents while sleeping. The commu-
nication protocol used by such devices needs
to be optimised for energy efficiency to min-
imise their active time.
Since most energy harvesters deliver only very
small amounts of power, it is necessary to ac-
cumulate it over time while the system is sleep-
ing and to lose only a small fraction of it in
the process. Therefore, the most fundamental
requirement for such energy-efficient systems
is that they have an extremely low idle current.
This means that only a very tiny amount of en-
ergy is consumed while the system is sleeping.
Standard consumer electronics devices today
have a standby current in the range of a few
milliamperes (mA), whereas power-optimised
embedded designs typically achieve standby
currents in the range of a few microamperes
(μA), an improvement factor of 1,000. In
comparison, the latest generation of EnOcean
energy harvesting wireless sensors require stand-
by currents of 100 nanoamperes (nA) or less,
an improvement factor of more than 10,000.
Achieving this level of performance requires
very advanced design techniques and extensive
optimisation of each individual component.
The second requirement is that the accumu-
lated energy has to be used as efficiently as
possible when the system is in active mode.
For wireless sensor devices, the two main tasks
in active state are to measure an external quan-
tity and to wirelessly transmit information
about its value. Both tasks need to be optimised
for minimal power consumption. Specifically
for the case of the wireless transmission this
means that the chosen protocol must be as ef-
fective as possible. The payload associated with
sensors is often small (a few bytes), therefore
the protocol overhead must be limited as much
as possible. This last requirement is difficult
to achieve while using IPv6 as communication
protocol, even on the individual sensor level,
because it incurs significant overhead as the
IPv6 header alone requires 40 bytes of protocol
data (figure 1).
In addition to that, UDP – probably the simplest
communication protocol on top of IPv6 – would
require an additional 8 bytes of protocol data
(figure 2). Based on the IPv6 and UDP header
structure, the transmission of 1 byte sensor data
would require an additional 48 bytes of low
level protocol data. IPv6/UDP is therefore not
well suited for energy-efficient communication
on sensor level in a network. In comparison, the
industry-leading EnOcean protocol for energy
harvesting wireless applications in accordance
with ISO/IEC 14543-3-10 would incur only 7
bytes of protocol overhead for the transmission
of 1 byte of sensor data (figure 3).
Translation between such an energy-efficient
sensor protocol and IPv6 is provided by dedi-
cated IP gateways that represent the state of
each connected sensor node and act as their
representative within the IPv6 network. This
approach allows exchanging data with individ-
ual sensors even while they are sleeping and
therefore are unavailable for direct communi-
cation. Upon wake-up, sensors will then update
their state information in the gateway and re-
trieve messages/commands intended for them.
This integrated approach of protocol transla-
tion enables all parties to communicate with
energy harvesting wireless sensor and actuator
networks via IPv6. That way, a protocol such
as ISO/IEC 14543-3-10, which is optimised
for ultra-low power and energy harvesting
wireless applications, can be used for the com-
munication between the sensor and the gateway.
This allows the deployment of a broad range
of maintenance-free and cost-effective devices
which are wirelessly connected. In conjunction
with IPv6 gateways, these nodes will form the
foundation for the Internet of Things.
Figure 1. IPv6 header structure
Figure 2. UDP header structure
Figure 3. ISO/IEC 14543-3-10 protocol structure
