WhichVoIP Blog

An IP packet is simply a the data you want transferred encapsulated in a frame or packet. The packet is typically broken up into smaller chunks to be transmitted over a T1 line. It’s kind of like writing a letter, you could take one line at a time and mail it in separate envelopes, if you lose one envelope then no big deal you could probably understand the message or at least decipher it after you received the rest.

The IP packet can be a rather complex entity with header information such as IP address sent from, destination IP address, header checksums, Identification field, protocol used and many others. Below is pictorial view of a packet. The data itself may also have sending and destination addresses, checksums, flags, etc so deciphering a packet can be like peeling an onion, revealing layer after layer.

Version – static 4 bit value indicating which version of IP is being used.

IHL IP Header Length – 4 bit value indicating the number of 32-bit words that make up the header. 

Type of Service – 8 bit value that is used tell the network how to treat the IP packet. These bits are generally used to indicate the Quality of Service (QoS) for the IP Packet.

Packet Length – 16 bit value indicating the size of the IP Packet in terms of bytes. This gives a maximum packet size of 65536 bytes.

Identification – 16 bit field used for reassembling the packet at the destination.

Flags – 3 bits indicating to network equipment if the IP packet can be further fragmented or not and if the packet is the last fragment or not of a larger transfer.

Fragment offset – 13 bit value used in the reassembly process at the destination.

Time to Live – 8 bit value telling the network how long an IP packet can exist in a network before it is destroyed.

Protocol – 8 bit value used to indicate the type of protocol being used (TCP, UDP, etc).

Header checksum – 16 bit 1’s compliment value designed to indicate errors in the header only. Every node in the network has to check and re-insert a new checksum as the header changes at every node (TTL value is decremented)

Source address – 32 bit value representing the IP address of the sender of the IP packet.

Destination address – 32 bit value representing the IP address of the packets final destination.

Options – variable bit field representing options The actual options requested are not optional in that there are well defined options that can be requested, what is optional is whether to request theses options or not.

Padding -  Variable size bit field. These bits are used to ensure a 32 bit boundary for the header is achieved.

Next time we’ll highlight the main difference between a data transfer and a VoIP call over a network.

WhichVoIP Team   

http://www.whichvoip.com

First let me state there are no physical differences between a T1 for Voice and a T1 for data. They use the same physical medium, they have the same bandwidth, the same signal coding and the same frame format. The only difference is that it is now data instead of voice.

We’ll confine our discussion to data sent via the internet.

First lets recap on the T1 frame. One frame every 8KHz, 24 DSO’s in a frame and 8 bits in a DS0. This can be seen in the diagram below. If we are talking about a T1 to a PBX then each DS0 carries one voice call, with that call occupying the same DS0 every frame until the call is terminated. 

 

 

 

 

 

 

With data transfers the DS0’s in a frame can be grouped together. If for example we grouped 6 DS0’s together then we would have a bandwidth of 384Kbits per second. This kind of generates it’s own frame inside the T1 frame 

 

 

 

 

 

 

Now we have created a second sub frame how do we know when one of these sub frames starts or stops? This is where we need to discuss IP Packets.

Next time round we’ll look at these IP packets and how they are formatted.

WhichVoIP Team 

http://www.whichvoip.com

 

The previous blog explained why we have 8 bits in a DS0, here we’ll explain how we get these 8 bits.

When  your voice is digitally sampled the amplitude or how loud you are speaking is captured, this happens 8000 times a second. However this digital sample is not what is sent. For a T1 line an algorithm called u-law is used to compand the data. This is a logarithmic type of algorithm that has the effect of giving greater resolution to lower value samples i.e. when you are talking quietly, and less resolution to higher amplitude samples. This intuitively makes sense, if someone is shouting then you can afford to loose a little differentiation between sounds but when someone is whispering you want to hear or accentuate the different levels so you can make out what they say. Typically when your voice is sampled, an Analog to Digital converter that has a 13 bit output is used, the u-law algorithm as well as performing this logarithmic compression has the effect of reducing the voice sample to the 8 bits that is sent.

Up until now we have confined our discussions to voice T1’s. Next time we’ll take a look at how data is transmitted via a T1 line.

WhichVoIP team

http://www.whichvoip.com

How did the 8KHz frame rate and the 8 bits per DS0 come about?

I’ll take the easy one first - Why 8 bits per DS0? For digital or telecom engineers 8 or any multiple of 2 is a perfect number. It’s very easy in digital logic to count in multiples of two thus it is very easy to extract these DS0’s, generate the frequencies and clocks required to receive and transmit at these rates and multiplex these T1 streams into T3 streams or higher rates. The trickier question is how do we get our voice to be transferred from one point to another in only 8 bits every 125us(this is the equivalent time of an 8KHz frame rate).

There is a theorem called Shannons theorem or sometimes referred to as Nyquist-Shannons theorem that drives the fundametal basics of all digital communications.  First remember that T1’s were invented long before the advent of the internet or VoIP and they were designed with the specific goal of transferring voice in a digital format. The result of Shannons theorem is that to sample an analog signal(your voice in this case), convert it to a digital format and then reconstitute it as an analog signal(your voice at  the other end) that has lost no or little information(the person you are talking to can understand you) the original analog signal must be sampled at a rate of at least twice the highest frequency you want to transmit.  The typical human voice has a frequency range of about 50Hz to 3500Hz. So if you want to sample that voice and digitize it you have to sample it at a rate of at least 7000Hz. As in all good designs there has to be some design margin, often referred to as a guard band in communication theory, a rate of 8000Hz was chosen. The frame rate of 8KHz became the standard.

Now we know why 8 bits is used in a DS0 and the rate it must be sent at. In the next blog we’ll go into a little more detail on how the 8 bits are generated and what they represent.

WhichVoIP Team

http://www.whichvoip.com