When I speak to someone over my cell phone, there is a tiny but noticeable delay (compared to when I use a land line), either in the sending of my words to my partner, or in the sending of their words to me. It’s just a tiny fraction of a second, but often it’s enough to screw up the cadence of a conversation.
What is the source of this delay? A/D conversion within the phone? Encoding for transmission? Something else?
What of these sources is not present in conventional landlines?
I can’t speculate as to the answer, but I’ve noticed that too and hate having a conversation on a cell phone because of it!
It’s all of the above.
Modern cell phones use packetized data. This implies a small but unavoidable latency in the audio. Add to this the delays in every step along the way - digitizing, compression, transmission, routing, reception, decompression, buffering, D/A and you end up with a delay of around 1/3 of a second or so.
This. It’s DSP- Digital Signal Processing. They have to convert your voice to 1s and 0s based on the appropriate codec in the vocoder, and then have to deconvert it from 1s and 0s back to an analog signal for the speaker.
"Give it a second! It’s going into SPACE!!"
But it doesn’t goo into space. Not even inter-continental calls go into space anymore, they go through submarine cables. That’s why the time lag on inter-continental calls is much shorter than it used to be.
Why did they use satellites at all given that the cables have been there for a lot longer? Was it a capacity thing, now resolved by fibre or whatever?
So, you are saying I can’t trust a stand up comedian for the truth about digital electronics?
I don’t care if the delay is shorter. They had submarine cables in Victorian times, dammit. I want my phone call to go into SPACE!
Yea, they only had limited capacity before (when the cables looked like this).
Fiber means that you can put an amazing amount of data through, so telephone calls are a no-biggie.
Cellphones are packet-switched (like your Internet connection, your USB bus, your cable TV-- well pretty much all data is packet-switched now), which means as the “phone lines” approach capacity, the latency increases.
Basically, the cell tower you (or your callee) was using was overloaded.
Don’t landlines have this delay as well?
I’m thinking of my boss and my former boss, both of whom have offices across from my cube (triangular).
When they’re both on a conference call, both on speakerphone - there’s a stadium echo effect when current boss (who has a big booming voice) speaks and there’s a small fractional second delay from when he speaks to when I hear it from the speaker of former boss.
EDIT: or is that just the delay of the conference call “server” or whatever telco calls it?
I know I am late to the conversation. The bottom line is the transmission technology used for cell phone connectivity. A lady (arguably the most important woman in U.S. History) named Heddy Lamar and another scientist (forgot his name) co-invented spread-spectrum technology in the 1940’s. Initially intended for anti-jamming systems for guided missiles. Technology went into servicein the 1960’s.
“Cell” phone refers to the grid layout of the network area.
Assume a flat plane and an tx antenna in the center. (Coverage area affected by geography/location)
Given the radius of the transmit area, the coverage area would be divided into hexagonal “cells” utilizing a specific set of frequencies so neighboring cells do not overlap. (You may have heard a voice on a phone line, this is explained when a frequency is intersected by a neighboring frequency. Since the carrier frequency changes constantly, these are always rare and short lived.)
As far as delay goes, land lines are always faster transmissions than cellphones because the cellphone transmission paths from end-to-end have more nodes (repeaters) than hard lines. Each time the signal has to be rerouted that costs time. Geography and atmospheric conditions affect the wireless transmissions.
The phone function of your cell phone is relatively unchanged since the invention of the radio.
Intelligence (voice) + Carrier freq (spread
Spectrum technology) + modulate intel onto carrier signal (AM/FM) + amplification = rf transmission. The rest of the phone is a computer that needs high bandwidth, so it utilizes a different transmission technology (4G LTE/etc). The latest versions are able to multiplex large amounts of data using a much higher freq range. (The higher the freq, transmit distance shortens and repeaters are needed.)
When it comes to voice on mobile services there are a range of competing factors, and the end to end delay is involved in these.
First up the voice is sampled in blocks of 20ms. Internal to the handset these blocks are compressed and otherwise processed to fit the modulation scheme in use for the call. So straight up, just the one block creates a delay of 20ms, since until the block is complete it can’t be processed and transmitted.
Internal to the mobile network the links have a limited bandwidth. Sending each block of 20ms on its own would create significant overhead - the payload would actually be smaller than the header. In order to get best use of the limited bandwidth things like statistical multiplexing are used. Here many voice packets from many conversations are aggregated into larger packets. Statistical techniques can allow the carrier to overuse the available bandwidth by relying upon the fact that in voice conversations there are pauses, so there are blank times when no packet needs to be transmitted. But of course sometimes, by pure chance, the conversations all line up together and there is too great a demand, and packets get dropped. Dropping packets leads to drops in intelligibility. But if you buffer up packets you can even up the flow and loss of packets can be reduced considerably. But each packet you buffer up is an additional 20ms of delay. And delay adds to loss of intelligibility once it gets much above 160ms.
So, there is a clear win to be had by allowing buffering of up to 8 packets in ordinary operation. A sixth of a second. In the face of other problems it may be reasonable to allow even more delay, take the hit on intelligibility in order to keep the packets flowing.
Buffering and packet aggregation (and disaggregation) are occurring all along the network. Depending upon the age of the network the actual packets may flow along packet switched or circuit switched lines. Newer networks are based upon IP, and that brings with it another host of compromises.
Once you get to inter-network links things can get more out of hand. You may be bridging between networks that use fundamentally different modes, and that is going to add even more buffering, with a 20ms delay for each time you buffer a single voice packet, more if you are buffering an aggregate of more than one for the conversation. Getting to the point where you exceed 160ms and get to the point of noticeable delay isn’t hard under these circumstances.
In principle holding the delay down can and is done, especially if you stay within the network of a single carrier, but it isn’t always easy.
Are you saying that cell networks pass along phone transmissions to other cell towers, creating a network of transmitting repeaters? Why wouldn’t they just jack every cell tower into the land line network?
No, cell towers do not communicate with each other. Each tower is connected to a backbone.
Well, kind of. Not every site can be set up in a place where sufficient land line backbone capacity is available. In those cases, they can use microwave shots to get to another tower location that has a good landline backbone connection.
Sure, but the point is, the cell towers are not some sort of “mesh” network.
I think the important point is that the cells don’t use the same frequency band or protocols to reticulate back the calls, data or control.
Cells can use cables or point to point microwave links to build the control and backbone network. This is pretty much independent of the operation of the cell for communication with handsets. In principle the cells could build a mesh network with point to point microwave. For cells that are using IP for communication this might not be a bad idea. But cells using ATM based infrastructure and protocols will be building what amounts to a whole set of virtual circuits reticulated back to a central system.
Each cell needs to be able to work out whether any handset that comes into range is able to connect - it needs to find out who is responsible for the handset and most importantly - who will pay for the call 
Consider the remarkable chaos that needs resolving when you first step off a plane on the other side of the planet and turn on your phone.
Nowadays the biggest question is cellular data anyway. The integration of voice with that, to the point where voice becomes a secondary question is well underway. Hence the focus on using IP end to end, even for pure voice. The advent of high quality voice is predicated on IP. Indeed if you discover that your carrier supports high quality voice, you may well find that it only does for calls that stay within its own network, because the interconnections between carriers is lagging technologically.
The OP question was perceived transmit/receive delay which in turn raises the issue of whether cell phones are full duplex or half duplex.
A traditional landline phone is full duplex – you can talk and receive at the same time, which facilitates natural conversational interaction. There is no transmit/receive delay since it’s both transmitting and receiving audio continuously.
Previous statements about packet delay don’t fully explain poor cellular audio quality. 3G cellular systems were NOT packet switched but circuit switched for voice and they also had poor audio.
Whether cell phones are full duplex or half duplex is complex. From an RF standpoint, a 4G cell phone is functionally full duplex. Using either Frequency Division Multiplexing or Time Division Multiplexing, there are effectively two separate RF channels available for transmit and receive. However how this is presented to the user at the audio level can vary.
In an attempt to conserve RF and system bandwidth, it appears that cellular carriers often create a half duplex or “walkie talkie” behavior at the handset. Maintaining a full time bidirectional or full duplex link would consume two separate audio channels with resultant RF and network bandwidth consumption. Since only one party is talking most of the time, they apparently often don’t continuously maintain the receive channel for the talking party. It’s a cheap way to cut bandwidth consumption but unfortunately also produces a cheap walkie-talkie feel.
They also apparently have some kind of cutoff threshold for transmitting non-voice audio. IOW if the DSP and vocoder logic doesn’t detect voice-type audio, why spend the bandwidth sending background sounds. However this further degrades the natural conversational feel, since when one party is quiet it feels like the call went dead. People often ask “are you still there?”. You never had to ask that on a plain old telephone landline.
On top of this there is vocoder compression, which further degrades audio quality. The vocoder (voice codec) itself in a cellular application is designed to shoe horn the voice audio bandwidth into the smallest possible data bandwidth, which in turn consumes less RF and network bandwidth. IOW it is highly compressed, and apparently they crank up the compression when the system is over subscribed, which is most of the time.
Yet VOIP can sound very good given enough bandwidth and network resources, e.g, Computer-to-computer Skype calling on a broadband network with good equipment. So it’s not inevitable that cellular calls have poor audio because of being packet switched or using vocoders. But the decision of cellular carriers to sacrifice quality for more in-use channels has resulted in this.
Here are several good discussions of the overall issues:
“Why Mobile Voice Quality Still Stinks—and How to Fix It” – IEEE Spectrum, 2014: Why Mobile Voice Quality Still Stinks—and How to Fix It - IEEE Spectrum
“What happened to the DUPLEX telephone call???” – Electronic Design, 2013: http://electronicdesign.com/forums/analog-mixed-signal/what-happened-duplex-telephone-call
“Why Is Cell Phone Call Quality So Terrible?” – Scientific American, 2015: https://www.scientificamerican.com/article/why-is-cell-phone-call-quality-so-terrible/
