[Asterisk-Dev] best approach for a new isdn driver mainly for asterisk.. ?!

[Steve Brown]

George Bean wrote:

>
>  
>
>I am well aware of this thread and have been following it since its
>inception as I do with all ISDN related threads. I spent several years
>coordinating ISDN services for a major institution and along the way
>collected several pieces of ISDN hardware that I would like to connect
>to the PSTN with a cost effective switching solution. Therefore I take a
>keen interest in all ISDN related threads on Asterisk lists.
>
>Modifying the PRI code to accept BRITE signaling has been discussed in
>this thread but unless I missed a message nobody has done so as of yet.
>Mr. Brown was putting forth a hardware solution by Paradyne that I was
>unaware of and unable to find detailed specifications regarding. I was
>merely pointing out that the hardware he was describing might not be an
>out of the box plug and play solution if its aggregate isn't a PRI. 
>
>ISDN PRI channel banks never made great in roads in the US; most LEC's
>and CLEC's utilized BRITE systems for the limited demand for ISDN that
>peaked half a decade ago. The Brite channel banks are strictly used to
>extend the range of ISDN BRI circuits in low density applications along
>with analog circuits. This includes extending them from a capable
>exchange to another non-capable exchange or to a neighborhood
>distribution hub. Therefore the odds of any channel bank purchased on
>Ebay supporting PRI/BRI are pretty slim. 
>
>In the end user market, the Adtran Atlas series boxes are the only
>widely available PRI/BRI device but they have a hefty price tag, $400
>per BRI port or more, which limits their cost effectiveness. This is
>especially true in light of Asterisk analog port options that are on the
>order of $30-$75 per FXS port. Utilizing channel banks with the Brite
>interface could drop the per BRI port cost below $100 as these are
>readily available on Ebay and less expensive than the Atlas series
>hardware to purchase new. Developing a multiport PCI card utilizing the
>HFC chipset could bring the BRI port costs in line with analog
>installations utilizing the TDM40B.
>
>I heartily support any efforts to incorporate Brite communications into
>Asterisk or develop a multiport HFC chipset interface card. I wish I had
>enough knowledge of Linux and programming to attempt the BRITE driver
>myself or the hardware background to develop the card but alas that
>isn't the case. I will therefore have to sit on the sidelines and await
>the efforts of other Asterisk developers.
>
>Regards,
>George Bean
>Puwaba Technologies
>
>
>_______________________________________________
>Asterisk-Dev mailing list
>Asterisk-Dev at lists.digium.com
>http://lists.digium.com/mailman/listinfo/asterisk-dev
>
>  
>
At the risk of abusing bandwidth:

I was refering to a channel bank using BRITE cards. These seem to 
support either 3DS0 (per T1X1.4 TR7) w/ the d channel using a full ds0 
channel and a scheme of multiplexing 4 d's on one ds0 channel. Hdlc 
encode/decode, the q.921 and q.931 protocols together with the NI-1 
(US), 1TR6 (German) and EDSS1 (Euro) bri TE-side national flavors are 
already supported by the i4l hisax driver. It's not clear that the 
existing asterisk PRI code is applicable. I am attempting an experiment 
get a BRITE port to work with i4l via the zaptel channel interface. What 
I think needs to be done is:

 1. get the existing NI-1 protocol working w/ my local 5ESS switch. 
Karsten Keil of SUSE and Guy Ellis of Traverse, the developer of NI-1, 
have been very helpful.
 2. either extend zaptel fcshdlc or the existing hisax hdlc code to 
encode/decode a brite d channel.
 3. write a hisax hardware-level driver to talk to the 3 zaptel channels.

So far, (1) can now place calls and shortly should be able to accept 
calls through a currently supported hisax card.
           (2) fasthdlc has been hacked to decode a brite d channel. I'm 
using this and 2 BRITE ports (one as TE and one NT) along with q931.c as 
a d channel protocol analyzer. It really helps to see why the isdn phone 
works and hisax doesn't.
           (3) still reading the O'Reilly book on linux device drivers.

This is meant as an experiment, not an implementation.

But, I do hope to convince the asterisk developer(s) that a real, 
supported brite interface

 - is not too hard
 - is really useful
 - regardless of cost vs a multi-port pci card, is an immediate and 
scalable solution for lots of bri ports
 - may be readily extendable to support the NT-side (isdn phones) based 
on the hisax patches of Andreas Eversberg at isdn.jolly.de
 - is mostly the same software needed for a multi-port isdn pci card anyway

Steve