What is the difference between the voice and date services. I have read that for voice services T1/E1 frame is used and for data services a frame with a standard header and payload nd tail is used defined by a protocol. Moreover voice services use circuit switched and data services use packet switched technique. But why? Can somebody explain this? and is multimedia services like video calling date or voice services?

  • 1
    \$\begingroup\$ As a sidenote: Today both voice and data services use both circuit switched and packet switched techniques. For example in DSL it's not uncommon to see PPP over ATM used for data transfer between say DSLAM/authentication/aggregation system and the ISP, when ISP doesn't control the "last mile" infrastructure. \$\endgroup\$ – AndrejaKo Jan 20 '13 at 1:38

The most important reason for the distinction is quality of service, which has very different requirements for both modes of operation.

In voice transmissions, you need to have the packets to arrive in order and they have to take the same short time to transmit. In voice chat, a few dozen milliseconds sound like a long time: "are you still there?". Circuit switched networks can usually avoid "store and forward" and directly transmit the packets. The data rates are generally low, a few kbit/s, which makes this cheap enough. The connections tend to last a relatively long time, a few minutes at least. It is know that a certain bandwidth is going to be moved between the two endpoints for some while. Downside: Connections get broken when the network has to rearrange. Another aspect is that usually the phone calls are somehow measured and paid for separately, especially international calls. This would be hard to do with TCP/IP.

Data stream packets generally do not have to arrive in sequence, or in similar time differences. Most high level protocols allow for loss of packets, arranging retransmission of packets, also rearranging out-of-order packets. Here you strive for maximum data rate, rerouting depending on traffic etc, which tends to drop packets or bring them out of order. The datarate to be transmitted is generally not known in advance, it comes in bursts, and can vary a lot over time, and the time of the connection is usually unknown. Also, cost only is calculated at the generating endpoint ignoring the route.

These differences in the modes of transmission required separate networks to be set up, each being the optimum for the different expectations, but the distinction is not as important any more as it used to be.

Now that the internet is reasonably fast and reliable, it becomes viable to route voice and video traffic through it. Until relatively recently, this was not possible in most places. With faster CPUs you can have better and faster compression and decompression of data in the end devices, allowing for a reasonably low latency in the end devices. Also, these protocols are adaptive with regard to bandwidth and latency and change the voice/video quality depending on the transmission.

With telephony your QOS (quality of service) terms basically guarantee in term, short latency delivery. With VOIP or Skype over internet protocols, you can expect it to usually work, but the quality of delivery will vary over time.

Most phone international phone traffic is IP based by now, being moved between the respective circuit switched networks. Over time, more and more circuit switched systems will move over to IP, because the services have become so reliable and cheap.

Some multimedia contents, say youtube, do not count in the same category as telephony. There is client side buffering, and usually you dont care if the video plays a second or so later. The main reason is: the content is predetermined. The content of the video is the same no matter what time you play it.

Some other content, say livestreams, while similar at first, are entirely different: The content is not predetermined, but is recorded concurrently with transmission. In this case, you cannot buffer a lot on the client side, simply because the packets for the next second are not known yet. And you want to be as close to real time as possible.

  • \$\begingroup\$ Actually buffering long time works with livestreams as well for certain values of "live". Often the need to be as close to real time doesn't exist. If you're watching a broadcast of some event over, it doesn't matter if you have one second delay or five seconds delay from what you actually see as long as the image appears to be in motion. Of course, if the system is somehow imagined to be interactive, then I do agree that significant buffering is a problem. \$\endgroup\$ – AndrejaKo Jan 20 '13 at 1:34
  • \$\begingroup\$ @Posopiet I did get an understanding of transmission of voice and data. But whats the basic definition of these two? In video calling may I say it is both? And similarly in what caegory should VoiP fall? \$\endgroup\$ – sk1 Jan 20 '13 at 10:27
  • 2
    \$\begingroup\$ Short: Every information transmitted is data. Just for some data like Voice and VoiP there are additional requirements which make it better suited for transmission along other types of networks. Requirements like: in-order, low-latency, bandwidth, connection-specific. \$\endgroup\$ – Posipiet Jan 20 '13 at 11:03

There are fundamental differences between the transmission requirements of voice and data. These differences make it difficult to design a network to support both voice and data services. Some of the differences are:

  1. Voice transmissions require a steady bit rate while data rate requirements can be very variable with large idle periods followed by big file downloads.

  2. Voice transmission can tolerate errors in the received bit stream while data networks have very strict error rate requirements.

  3. Voice communications is sensitive to delay while data is more robust in this regard.


Some say 'today it doesn't make a difference', meaning that everyone might as well be on data instead of voice, so therefore, for example, Apple and the data providers have colluded to prevent you from buying or using an iPhone without a 'data plan'. Now what is the value of a lesser capability if you must pay for the greater capability in order to use it? The difference between 'voice' and 'data' is $30 per month in the USA, which is $12,000 per customer NPV at 3%. Therefore, in the United States you cannot use the cellular voice service on an iPhone (for example) without purchasing a 'data service' wether you need the cellular data service or not.


From your tags, I'm guessing your context is wireless phone networks - 2G/3G/4G, GSM/LTE etc.
The answer then is: it depends... For 2G and 3G, there are basically two networks, one for voice, one for 'data'. The voice network is 'circuit switched' as it connects to the (circuit switched) PSTN voice network. The data network is generally packet-switched - basically IP. Note that the distinction between 'voice' and 'data' is very much dependent on the network infrastructure. You mention E1/T1 - note that although these are 'data' transport protocols (at 2Mbit/1.5Mbit respectively), they can transport both 'voice' and 'data' and have traditionally been designed for circuit switched traffic (basically ISDN traffic). 'Data' in this context doesn't automatically mean 'packed switched' or 'not voice'. It just means 'binary' and not 'analog'.
Nowadays, the connection that goes from the radio tower into the 'network' can be anything - E1/T1 (often used for GSM backhaul), ATM (for early 3G networks) and Ethernet. It depends mostly on the age of the mobile equipment.
When it comes to 4G networks things become a bit different - 4G is 'packet only'. As such, a 4G voice call is a Voice Over IP call and as such, the transport mechanism is the same for both data and voice. However, individual IP packets might have different ToS and QoS parameters to tell the network to treat the packets differently - low latency, low packet loss, etc. whichever is most suitable for the type of traffic (or even the type of voice codec that's being used for the VoIP call). The packets will also have a different destination depending on what you are doing: if you're browsing the web, they'll go to an internet gateway router. If you are calling your grandma on her old PSTN phone, they'll go to a gateway that 'translates' VoIP to PSTN and vice-versa. If you are calling another 4G phone, they'll go straight to that phone. In summary, there's a lot to it, and with the onset of 'all-IP', the distinction in actual network architecture and protocols becomes less and less as everything is made to support IP. You end up with 'old' E1 networks transporting IP, or even IP over ATM (which is what xDSL is basically).


protected by Nick Alexeev Jul 7 '14 at 5:45

Thank you for your interest in this question. Because it has attracted low-quality or spam answers that had to be removed, posting an answer now requires 10 reputation on this site (the association bonus does not count).

Would you like to answer one of these unanswered questions instead?

Not the answer you're looking for? Browse other questions tagged or ask your own question.