Packet Change for a 3 SCO Links Device
Original Post: A question about 3 SCO links (eGroups Msg.
) Date: 2000-09-05
Let us assume that three SCO connections have been
setup from one device to another , so HV3 packets are in use (see 'Effect
of HV1 Packets on Bluetooth Device Capacity'). However all the
bandwidth is being used as three SCO connections use all the slots for SCO
information.
Question
What happens if LMP PDUs or other data needs to be
transmitted, does this mean the SCO data must be interrupted and so SCO
data will be lost?. Is there any way to get around this?. What is the
standard method at the audio side to cope with this loss.
Answer
There are a number of opinions, whether to use the
existing underlying ACL link for signalling , or switching the link to
just use DV packets. Let us look at the facts:
Option 1 : Using Underlying ACL Link : Recommended
A HV3 packet is sent every 6th timeslot. However,
since BT uses TDD full duplex, every other time slot is for reverse
direction. So, there are no 3 timeslots still free, because these are
reserved for reverse direction. So, 1 x HV1 is full load, 2 x HV2 is full
load, 3 x HV3 is full load.
ACL DM1 packets (or another DM/DH packet type) are
always setup first, (before any SCO link is setup afterwards). The
signalling (link control/link management) is therefore taken care of by
the ACL link. When SCO is using "full load", the ACL signalling
must use time stealing, i.e. some of the SCO HVx packets are dropped, and
used for ACL signalling instead. Therefore DM1 packets are always used for
signalling. There is no need to change the link.
Option 2 : Switching Link to Use DV Packets
However,the possibly exists of just using DV packets for
combined voice/signalling .This does means that the voice quality may
drop, since DV is unprotected voice (no FEC). Also 57.6 Kbps is however
way over budget of what is needed for pure signalling, so this may not be
a good solution. The DV packet are a solution when using SCO full load,
and you need to maintain the voice link(s) while supporting data
packet transmission simultaneously (file transfer e.g.). Also if you use
DV packets, a DV packet can only
contain 10 bytes of voice data which is the equivalent of one HV1 packet,
but only 1/2 of an HV2 packet and only 1/3 of a HV3 packet - so there will
still be some voice data loss, so I don't believe it is useful to switch
to a DV packet.
|
