text stringlengths 1 112 |
|---|
combinations of ft and nse parameters the location of the flush points can be calculated |
from the equation s in the core vol part b |
section connected isochronous streams |
returning to figure we see the initiator sending packet p in the first subevent which |
is acknowledged so the controller immediately moves on to transmit packet p transmitting |
it three times before it is acknowledged |
in the second isochronous interval packet p is transmitted but the acceptor sends nack s |
to indicate error s in the received packet as the flush timeout is set to and bn the |
flush point for packet p is in the same isochronous interval coming after two further |
transmission attempts neither of which have been successfully received by the acceptor |
at that point the initiator starts transmitting p although once again in this example t he |
acceptor responds with nacks to indicate a problem with the packets it received after two |
attempts p is flushed by the initiator |
in the final isochronous interval of figure p is successfully transmitted leaving three |
opportunities for p all of which are unsuccessful in each case the initiator will attempt to |
transmit the pdus in every available subevent before that pdus flush point after each |
acknowledged transmission it will move on to the next available packet or if there are no |
further packets available close the isochronous event |
in this example p p and p would be discarded in real life wed expect a much better |
rate of acknowledgement this is just an example to illustrate the principle |
as a final example figure shows the eff ect of increasing the flush timeout to with a |
burst number of which gives the opportunity to transmit in two consecutive isochronous |
intervals in this example no packets are flushed |
figure example of nse bn and ft |
using flush timeout and burst number can be very useful to provide more retransmission |
opportunities span ning multiple isochronous intervals they are particularly useful if youre |
chapter isochronous streams |
in a no isy environment however they have an effect on latency every increment of flush |
timeout increases latency as the retransmissions are spread across more isochronous |
intervals whilst burst number increases the duration of the isochronous intervals note that |
burst number is confined to multiple payloads arriving in an isochronous interval it cant |
be sued to solve the problem of a single payload hogging transmission opportunities which |
we saw in figure |
these parameters cannot by set directly by the host it is limited to setting values for the |
maximum transport latency the maximum sdu size and the sdu interval bn nse and |
ft are then calculated in the contr oller which takes into account any other radio requirements |
in the chip it is however very useful to have an understanding of the potential effects these |
parameters have on the isochronous channel structure table of tmap provides |
examples of how a controller might interpret the host values for a cis to suit different |
operating conditions such as prioritising airtime for coexistence or minimising latency the |
exact allocation of parameters is always d own the algorithms in the scheduler which will be |
set by the chip supplier |
framing |
the other parameter in a cis pdu header which needs to be understood is the pair of llid |
link layer id bits which indicates whethe r the cis is framed or unframed unframed refers |
to the case where a pdu consists of one or more complete codec frames it is used where |
the isochronous interval is an integer multiple of the codec frame length in contrast framed |
is where you have a mi smatch between the codec frame length and the isochronous interval |
which results in a codec frame being segmented across multiple sdus this starts to get |
complex but is important where an initiator may need to support bluetooth connections |
which have d ifferent timings well revisit that when we look at a feature called isoal which |
is the isochronous adaptation layer which has been designed to cope with this mismatch |
the llid bits also indicate when t here is no iso pdu which is diff erent from the npi in |
the iso pdu header |
multiple cises |
having cover ed all of the features of a single unidirectional cis the next step is to add more |
of them the most common application for this is when an initiator is sending audio to a left |
and a right earbud in this case the initiator will set up separate isochronous stream s with |
two different acceptors in figure we can see that its a straightforward extension of |
what weve seen before the initiator transmits and receives an acknowledgment from the |
first acceptor then repeats this for the second ac ceptor |
an important point to note is that although the left and right audio channels are sampled at |
the same time the iso pdus are sent serially |
section connected isochrono us streams |
figure timings for cises to two acceptors |
note that where we have more than one cis they always have the same isochronous interval |
their anchor points are different as data is sent serially but for each cis their anchor points |
are the same iso interval apart each anchor poin t represents the point of the first |
transmission from an initiator to an acceptor for that cis as figure shows each cis |
has an associated acl li nk the acl link always needs to be present because it is used to set |
up the cis and control it if there are multiple cises between an initiator and an acceptor |
they can share the same acl if the acl is lost for any reason any associated cises are |
terminated the application then needs to decide what to do with any remaining cises |
established with other acceptors within that cig |
figure multiple cises and associated acl links |
when an initiator is scheduling two or more audio channels there are two options for how |
they are transmitted regardless of whether they are connected to one or multiple acceptors |
the obvious approach is to tra nsmit them sequentially so that you send all of the data on cis |
chapter isochronous streams |
and then all the data on cis continuing until the initiator has worked through all of the |
packets for both cises this is illustrated in figure where we have an nse of it |
shows all of the subevent s within cis being transmitted before the subevent s for cis |
figure sequential arrangement of two cises |
the disadvantage of this approach is that if you have a reliable connection where the first |
transmission of the packet is acknowledged you end up with gaps between each cis in |
devices like phones which are often sharing the bluetooth and wi fi radios thats wasted |
airtime that could be used for something else to address this the controller can choose to |
interleave cises as an alternative approach as shown in figure |
figure interleaved arrangement of two cises |
in this case each subevent for cis is followed by a subevent for cis and so on for the |
remaining cises the diagram repeats the example of figure having an nse of and |
shows cis transmitting and receiving its first subevent followed by cis if both |
acceptors receive those first transmissions and acknowledge them they can close their cis |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.