Image Processing Reference
In-Depth Information
TABLE .
Empirical Comparison of Wireless Technologies in Discrete Factory Automation for Short Data
Packet Applications (Sensors)
Energy/Bit
with
Figure of Merit
Networks per
Cycle Time Air
Real-Time
Retransmits
(Nodes per ms a.
Cell
a
Limit
b
(ms)
I
Nodes/Network
Interface (ms)
(mWs)
mWs)
WISA
.
BT [,]
.
.
c
WLAN []
()
()
(.)
Zigbee []
()
.
(.)
()
c
Enocean []
()
.
(.)
II Nodes per Range
r
in Realizable Avg.
Cell Machine in m Node Density
Per
m
d
WISA .
BT .
WLAN .
Zigbee .
Enocean .
Source
:Scheible,G.,Dzung,D.,Endresen,J.,andFrey,J.-E.,
IEEE Ind. Electron. Mag
., , , . With permission.
a
Without significant performance change.
b
Air interface (error rate
-, at max. node density, and at five events per second and device).
c
Not defined, value assumed for comparison.
d
∼
Usable volume: assumed radius of range
r
,heightm.
can, after start-up, be transmitting again already after few ms, including internal secondary power
supply start-up procedures.
To quantify the achieved performance, another figure of merit was defined (see right column in
Table ./I) as number of nodes divided by real-time limit and by power consumption. he empirical
quantitative comparison of the technologies mentioned in Chapter was done using the figures as
giveninTable..Assumedwasthewirelessconnectionofsimpleproximityswitches(bit),the
most widespread sensor in discrete factory automation.
Data as listed in Table . are often a subject of controversy and consistent practical data cannot
be found in papers. Therefore, typical estimated values have been collected and taken to enable a
comparison. Where values cannot be really guaranteed due to the technology, the figures in paren-
theses have been used to conduct the performance comparison. In Table ., “Enocean” refers to a
proprietary technology known for its low power consumption, developed for building automation.
The real-time limit in Table ./I is the time delay, which can be “practically” guaranteed (very
small failure probability, goal here at the mentioned number of nodes in operation:
−
).
As an example, consider a Bluetooth pico-net having up to seven slaves. Ten Bluetooth networks
have been reported to operate directly in parallel [], without a significant decrease of performance.
Assuming the typical range in industrial environment, e.g., m in for Bluetooth ( mW class
version) and assuming a height of typical applications of m, a realizable average node density figure
per m
can be calculated (Table ./II). This average figure is quite small for Bluetooth due to the
longer range, compared to the requirement (.- wireless devices per m
in factory automation
applications, see Chapter ). Bluetooth can operate with a cycle time of
. ms, but under realistic
conditions at high node densities the maximum latency that can be practically guaranteed is much
higher, figures of ms have been reported by various groups. For Bluetooth, a power consumption
of mW has been estimated, assuming a certain duty cycle in use.
The resulting figure of merit (last column in Table ./I) for WISA is (nodes per ms and mWs),
roughly a factor of hundred higher than for any other technology. his relation changes somewhat on
the energy per bit scale, with assuming a bit payload, but will still be around a factor of higher.
Looking at Table ./II, it can be clearly seen from the achievable average node density in a factory
hall, that only WISA can achieve the requirements for a whole factory.
∼