sundance-communications.com/forum Forums Wiring - Telecom Carriers - Trunks & Services T1's, CSU's and DSU's I need someone to help me with the specifics on T1 framing/etc

Let me start with where my thinking is coming from. VoIP's biggest trump card is that it allows WAN networking of telecommunications. TDM's biggest limitation to this is that it's based on 64K DS0 channels, or more commonly for business a T1's 1.544mbps. While using G729a over a full-T1 will yield around 50 channels (plus/minus), it is using almost 15-20kb per stream as pure overhead for the Ethernet framing. So each 8kb/s G729 stream is really 24-30kb/s.

Now this is where I start to go off the reservation with my idea.

A DS0 is 64kbs. G729a is 8kbs. Is it possible to frame multiple G729a frames (or any kind of data) inside a frame on a T1 channel? Like encapsulate a frame into a frame. I'm sure it wont be 8 streams into a frame (if it can be done) because of overhead.

I'm not 100% sure how this could be done as I dont know and haven't looked into the real specifics of how T1 frames are transmitted/constructed/etc.

Also i'm not 100% sure you would call it a T1 frame anyways

I believe that G729a after it adds it's own header info to the stream it's around 11kb/s. This would mean you could squeeze nearly 5x the channels into a standard T1.

I'm not really trying to debate how G729 will yield a reduction in voice quality, i'm just interested in figuring out if it's even theoretically possible.

If you happen to have some good links for T1 designs send 'em to me. I'm all for digging through to find the answer.

And on a side-note, if you could do a single Point-to-Point T1 and get 120 channels of G729a quality over it, would it even interest you?

Thanks in advance for any feedback

Kumba,

As I do not speak "geek", I would say you just re-invented the wheel.

So, welcome to the "Real World".

I'm not real sure what you're asking. If you're asking if a compress signal can be run on a T-1 than yes it can. As Ken pointed out with VOIP you're running a whole like more than voice.

Like Ken and Bill have mentioned, it sounds like you are inquiring the possibility of applying the Ethernet framing format (data structure 'encapulation', etc.) to a T1 circuit, to achieve some sort of compressed, higher level of optimization in regards to more channels, higher bandwidth, etc.. I'm not very knowledged with Ethernet or Ethernet framing, but I'll provide a detailed description of the T1 Frame, and different T1 framing format principals below. Hopefully this information will assist you in finding the answer to your question. :thumb:
I believe in keeping it simple. It's best to try to explain T1 framing 'from the ground up'. You'll hear terms like D4 (or SF) and ESF. Although we normally associate the term T1 with a voice or data pathway with a 1.544 mbps data rate, the actual term 'T1' (T Carrier Level 1) applies to the standard T1 Frame. The T1 frame consists of 24 timeslots. Each timeslot is 8 bits. Therefore, the T1 frame is 24 timeslots x eight bits = 192 bits + 1 extra synchronization bit (located at the begining of each T1 frame) equaling a total of 193 bits. Each of the 24 timeslots has a data rate of 64,000 bits per second. 64,000 (bits/second)devided by each 8 bit timeslot equals 8000 frames a second. A T1 frame will arrive every 1 second/8000 frames/sec = 125 microseconds. The data rate is (24 channels x 8 bits/channel), devided by 1 frame, x 8000 frames/second = 1536000 bits/second. The total line rate is (24 channels 'times' 8 bits/channel + 1 synchronization bit)/1 frame* 8000 frames/second = 1544000 bits/second. (standard 1.544 mbps "T-Carrier Level 1"). So when you say T1, you have also been referring to the T1 frame.

Here's a diagram of a T1 Frame:



A standard T1 framing format known as D4 or SF (super-frame) is comprised of twelve T1 frames. There are two types of framing bits, the terminal framing bits (Ft) and the signaling framing bits (Fs). D4 uses two signaling bits in the 6th and 12th frames. These are called A and B bits. The A and B bits are also used in T1 channel associated signaling (CAS), applying to 'robbed bit' signaling, and only when all 24 channels are used for voice.
Here's a diagram of D4 or SF (superframe) T1 framing format:


With ESF framing (Extended Super-frame), the T1 frames have been doubled from D4/SF's 12 frames to 24 frames:

KLD & Bill: I'm not trying to run VoIP exactly. In a manner somewhat I am looking at better utilization of the wheel and not really reinventing it. I want to try and take 5 or 6 8K G729 packets and pack them together into one 64K channel without IP overhead. It would be solely a point to point link. Similar to how TDM works but done with data stacking and a compressed codec on top of it. I want to try and reduce the overhead of the codec stream to nothing more then a CRC value, start/stop, and ID.

5-Etek: Thanks for the info, that's the bulk majority of what I was looking for! Were the tables pulled from a book or did you google them?

My goal with what I am thinking is to be able to get a higher call volume through existing links. Stacking different bits together to make one larger chunk is old-hat and has been done for years. The IP standard being one example.

Kumba -

An excellent reference on T-Carrier is William Flanagan's Guide to T-1 Networking. Check it out on Amazon:
https://www.amazon.com/Guide-T-1-Networking-Install-Desktop/dp/1578200210
Years ago I taught a class on T-Carrier and I used this book as a textbook.

Sam

Well I figured out how to send 96 voice channels over a single T1. It turned out that I needed to step back a little and approach it from a different perspective.

I wont go into exactly how I did it in here since it has less to do with T1 structure and operation then I thought. The end result is that 96-channels (taken as 4 RBS T1's) come in/out of the box on the client side, it gets compressed/decompressed with G729a, and streams in/out at 1.536mbps on the network side using ethernet. You can either jump it up onto the internet at this point or have a point-to-point data T1 carry it. The two gotcha's are a reduction of voice quality (Somewhere between a cell-phone and u-law) and it is incompatible with fax.

I still thank you for the great input.

Silver: That book is ordered! Thanks for the suggestion.

What you are suggesting is similar to AAL2 subcell multiplexing. In Voice over ATM they put chunks of each voice channel into the same ATM cell to cut down on overhead.

Similar but I did it with the data-side of things as opposed to the T1 framing/transmission side. All the transmission/network protocols are standard.

It takes 4 T1 PRI/RBS on the client side and spits out UDP Ethernet packets on the network side. Each T1 can be handled independently of each other and with mixed PRI/RBS setups. The data protocol it uses is NAT friendly and only uses 1-port for all traffic. It basically acts as a transparent high-density bridge.

Currently i'm writing a web-based administration to this so that it can be deployed and configured fairly easy. Hopefully the only thing you have to configure is IP Settings, remote-side IP, L/P, and which T1's are PRI or RBS.

It can also scale in groups of 96-channels. How many groups depends on the system.

This solution sounds like a big VoIP ATA box. We have used a box from a company called VegaStream, to link standard RBS T1's from Dialogic cards to a dedicated data pipe pointed at a Phillipenes agent center from a CoLo in LA.

I also was looking into integrated T1's from an ISP/phone carrier, where the T1 could actually carry up to 40 voice channels. The data bandwidth scaled as the need for voice channels changed. This would use an Integrated Access Device like those from Adtran. It wasn't VoIP.