Image Processing Reference
In-Depth Information
offered by this approach, as for larger numbers, the distances to some of the interferers are likely to
be too large to be harmful.
In [ZSMH,Zür,MZTM], thus a more sophisticated approach has been chosen and
Bluetooth-Bluetooth coexistence results have been obtained with help of detailed radio network sim-
ulations that include traffic, distribution, and fading models as well as adjacent channel effects. All
results have been obtained for an office of
m
,assuminganaveragemaster-slavedistanceof
m. Naturally a factory floor is likely to be significantly larger than
×
m
and to contain more
metal objects. Still, the range Bluetooth is comparably small, so that it is less impacted by multipath
propagation than WLAN technologies (see Section ..). On the factory floor it is thus possible
toplacetheBluetoothunitswithalmostthesamedensityasintheinvestigatedoicescenariowith-
out loss in performance. As a factory floor is larger than the investigated office, the overall number
of piconets that can be used simultaneously on the factory floor is larger too. Additionally, location
and traffic of the factory floor units are likely to be more predictable. Directive antennas also help
to improve the performance. he results of the aforementioned publications thus give a good idea of
what performance is achievable:
•
A
×
m
room supports HV simultaneous speech connections with an average
packet loss rate of % (a limit that still allows for acceptable quality).
•
HVpackettypesarepreferabletoHVandHV.hesubjectivequalitywillnotincrease
with additional payload coding. Using a coded HV packet just increases (unnecessarily)
the interference in the network and the power consumption.
•
(!) simultaneous WWW-sessions (bursty traic with an
average
data rate of . kbps
each) in the
×
m
size room result in a degradation of the aggregate throughput of
×
only %.
•
Maximum aggregate throughput in the room is Mbps (at fully loaded piconets).
These piconets then transmit at a unidirectional data rate of kbps.
•
Long and uncoded packets are preferable to shorter and/or coded ones. It takes inter-
fering piconets using the same packet type, interfering HV connections (worst case),
or a link distance of m (which is far beyond the envisioned range of m) before
another packet type yields a larger throughput than DH [MZTM]. It is advisable not
tousetheoptionalFEC(DM-packettypes).Asthecodingisnotappropriatetohandlethe
almost binary character of the Bluetooth (ad hoc) transmission channel,
∗
the additional
power (and bandwidth) that would be needed for the coding can be saved.
Bluetooth is inexpensive and consumes significantly less power than IEEE . systems. The
ACL-link is reliable with best-effort traffic (with a maximum throughput of kbps). he SCO-link
has reserved bandwidth though the packets might contain residual bit errors (even when using the
enhanced SCO-link). In principle, Bluetooth is very robust against other Bluetooth interference and
good performance can be achieved even in very dense environments. Note that customized imple-
mentations, which cannot be based on existing profiles, might be difficult to realize, as the regulations
do not allow the implementation of proprietary solutions. he specification of new profiles though
canbequitetimeconsuming.
∗
∗The reasons are manifold. Without interference, the channel varies already due to hopping over relatively narrowband
channels. Additionally, with the wavelength used in Bluetooth even small changes in position can cause large changes in the
received signal strength. When there is interference the effect becomes more pronounced. he existence or nonexistence
of a close co-channel interferer can make the channel change from very good to very bad within the fraction of a moment
(see also Figure .).
