Can't hear you no more

April 26, 2007 8:22 AM

What causes a cell phone call in which only one party can hear the other? Phone issue or carrier issue?

I recently switched to Sprint with a Samsung A900M. The service has been fine, with one exception: in about 1 out of 10 calls I make or receive, the person placing the call can hear perfectly, but the call answerer can hear nothing. It has happened both when I call someone and when someone (various people) calls me, so I don't think it's a microphone issue. It also doesn't seem to be related to my location; it's happened in places where I frequently make calls and where signal strength seems to be high at the time, including outdoors.

Also noteworthy is that if one party immediately calls the other back after this happens, the new call is fine. Other calls that go through properly are clear and I haven't yet had one dropped.

So what causes this? Would it be likely to stop if I had a different phone, or is it a Sprint issue? Or might it be specific to the connection of Sprint with another carrier? I'm interested in the how-it-works.

posted by zadermatermorts to technology (17 comments total)

There are four things going on in a cell phone call.

Tower sends radio to you (you hear your friend). This is strong.

You send radio to the tower (your friend hears you). This is weak.

Your friend sends to his tower (you hear him). This is weak.

Your friends tower sends to him (he hears you). This is strong.

If one of these four things is having a problem, for whatever reason, only one part of the call could be having problems. The tower sends a much stronger signal than your handheld device is capable of. So there is a disparity there, as well as in your location.
posted by popechunk at 8:37 AM on April 26, 2007

Popechunk, in J-STD-008 if the tower or the phone loses 3 frames in a row it drops the call. (A frame is sent every 20 milliseconds.) In other words, if there is a problem with any of the four radio links, you'll know immediately because the call will end.

The described behavior cannot be caused by a malfunction of the radio link.
posted by Steven C. Den Beste at 8:51 AM on April 26, 2007

I don't know what causes this but it also happens occasionally on my Verizon phone, and it occurs on cell to landline calls as well as cell to cell.
posted by pombe at 9:38 AM on April 26, 2007

Would it be likely to stop if I had a different phone, or is it a Sprint issue?

LG phone on Cingular here. Happens to me all the time just as you describe. No clue what's going on.
posted by Opposite George at 9:47 AM on April 26, 2007

Happens to me only when I call my mother (I've got a Verizon Razr and she's had a Cingular Blackberry and a Nextel something-er-other, and it happened with both). Perhaps the cell phone gods are telling me something....
posted by nursegracer at 11:31 AM on April 26, 2007

The described behavior cannot be caused by a malfunction of the radio link.

I know that you probably know more about CDMA in your pinky than I ever will (due to your job, and the fact that I worked at a TDMA carrier BITD), but I am having a hard time understanding what kinds of things would cause these issues that were not radio-related. Are you asserting that if the involved phones were connected to the towers by wires, these call-quality problems would persist?

What are your guesses? I am eager to know (not just because my Cingular phone drives me bananas with these same problems).
posted by popechunk at 11:44 AM on April 26, 2007

This happens when I call a particular friend who uses Cingular. I can't think of it ever occuring to other calls to any other cingular users that I talk to.
posted by winks007 at 12:48 PM on April 26, 2007

The described behavior cannot be caused by a malfunction of the radio link.

Sure it can. You're losing a full duplex transmission, which is sending and receiving either on two different frequencies, or dividing the signal by time.

Unfortunately, there's not really a way to tell if this is the phone or something peculiar about the carrier, although if it were the carrier, it would likely be specific to a tower or switch problem. I'm guessing that because of the consistency and the likelihood that the company would repair the issue relatively quickly, it's the phone.
posted by Cool Papa Bell at 1:09 PM on April 26, 2007

there's not really a way to tell if this is the phone or something peculiar about the carrier

The carriers maintain extremely detailed data (maps) about where all the crappy coverage spots in their networks are. They also know this on a per-phone basis. I have been friends with the guys who drive around in mini-vans full of cell phones charting signal strength all day.

My cellular days are a decade past. What I've said is for sure true of those days, but maybe not today, though.
posted by popechunk at 1:20 PM on April 26, 2007

J-STD-008 (and CDMA2K, which is what Sprint is upgrading to) is packetized digital. Frames (i.e. packets) are transmitted 50 times per second by the phone and by the cell. Each packet has a digital header and a checksum irrespective of what sound it represents.

Dead silence would create an eighth-rate packet, but it isn't a straight stretch of zeros. It's a real packet with a real checksum, and it either gets through or it doesn't.

I'm leaving out some details, like the function of the Viterbi decoder. The point is that the receiver can always tell whether it has successfully received a packet.

If conditions are marginal, it's possible for packets to get dropped. If they're wrong they don't get used, and the usual fallback is for the codec to repeat the previous packet, making a kind of strange stutter sound very briefly.

There are heuristics used by the phone and the cell to decide if conditions are too marginal to permit the call to continue. A certain percentage of bad frames received as a running average, or three bad frames in a row, or certain other conditions, lead to a decision to give up and drop the call. When that happens, the call ends. If, for instance, the forward link (from cell to phone) is poor and the phone receives three bad frames in a row, it drops the call and it ceases to transmit on the reverse link. That, in turn, means that the cell no longer receives good frames from the phone, which makes it drop the call as well. When that happens, it sends a signal to the other end of the call to terminate it.

So it is not possible for a call to continue for perceptibly long periods of time with one radio link out. If a given radio link is out the call will end after about 60 milliseconds.

But in fact the call won't even begin. If the conditions are that poor, it wouldn't even be possible to set up the traffic channel in the first place.
posted by Steven C. Den Beste at 1:26 PM on April 26, 2007

As to my guesses, I didn't make any because there isn't enough information. My first suspect would have been a marginal microphone in the phone. My second suspect would have been a flakey card in the cell. But the OP specifically excluded his mic or a particular cell as being culprits.
posted by Steven C. Den Beste at 1:28 PM on April 26, 2007

When a phone originates a call, the first thing it does is to register. That means that it sends a message to the cell on the paging channel announcing, "Here I am!" The cell has to respond with "There you are!" for the registration to complete successfully, and if it doesn't do so the phone will try a couple more times.

If the registration succeeds, then the phone sends a message on the paging channel requesting a traffic channel. There are three or four messages that bounce back and forth in order to set it up. Note that the paging channel is on the same carrier frequency as the traffic channels. (Also as the pilot and sync channel, but let's not get into that.)

If one of the radio links is out, this cannot happen. And if it fails (timeouts and retries exhausted) the phone goes out-of-service and starts searching. (I.e. it starts to hunt for a decent carrier using its preferred roaming list.)

It isn't possible to set up a traffic channel unless both radio links are working at least somewhat reliably. And if, after the traffic channel has been set up, one or both radio links stop working, the call will drop after 60 milliseconds, three sequential bad frames.
posted by Steven C. Den Beste at 1:43 PM on April 26, 2007

Steven, from what you're saying it sounds like it's impossible for the cell/phone link to drop one side of the conversation for any appreciable length of time. And we've already ruled out phone mic and cell radio-side problems (or at least determined they're not likely.)

So I think the obvious question is whether the non-radio part of the network between the ends of the call can result in one side being dead. Bad interface between the cell and the landline network? Something else? The whole redialing-making-the-problem-disappear thing just reminds me so much of the "got a bad line" issue familiar from the good old days of wired service that it seems like that's the place to look.

Any idea what could go wrong in between the ends or am I just on crack here?*

*The two aren't mutually exclusive, btw. And go easy on my mangling the lingo - I'm making most of these terms up.
posted by Opposite George at 7:47 PM on April 26, 2007

Steven, from what you're saying it sounds like it's impossible for the cell/phone link to drop one side of the conversation for any appreciable length of time.

That's correct. The observed behavior for the loss of a radio link is "unexpected call termination", not "I can't hear you".

The problem is that we don't have anything like enough information to figure it out. We don't even know if it's Sprint's problem; it could be a problem with the OP's friend's carrier. It could also be a POTS problem.

Taking a hypothetical case of the OP being a Sprint customer and her friend being on Cingular, then the number of potential sources of error is tremendous -- because Sprint and Cingular probably don't have a direct connection running between them.

What's more likely is that each of them is connected to an old-style landline service provider.

The POTS (Plain Old Telephone System) doesn't use the same codecs as cell phones use. And Cingular's GSM system doesn't use the same codecs as Sprint. If it were a call between two Sprint users, the voice data would stay encoded from one end to the other, but when going through the POTS to get to Cingular it has to be recoded at least twice, and maybe even more.

Moreover, the voice traffic from the OP to her friend doesn't necessarily follow the same path overall as the voice traffic coming back. Papa above referred to this as "full duplex" but that's not correct; "full duplex" means that traffic goes both ways on the same wire, and that's not how the modern phone system works. A better way to think of it is as two simplex lines which work in tandem, but even that is deceptive because they aren't wires. They are virtual circuits, and they really are separate. Probably they run through the same cabinets, but they go through different hardware inside. They may run through the same cables, but won't be on the same fibers. (Fibers are simplex.)

An outage or failure could happen anywhere in there and there's absolutely no way for us to know where it is. It could be a problem in a POTS switch. It could be a multiplexing problem feeding a fiber. It could be a bad interface between one of the cell systems and the POTS. It could be a bad card in a cell, or a bad card in the cell system office. The system is amazingly complex, and if the only symptom you have to work with is "It don't woik" there's not much to go on unless you have access to system debug logs and tracking information -- and we don't.

About the only thing the OP can do is to complain to Sprint, and provide them with lots of details, and hope that Sprint cares enough to try to chase the problem down.
posted by Steven C. Den Beste at 9:23 PM on April 26, 2007

By the way, it isn't even the case that the radio link is "full duplex". The radio link is also a pair of simplex connections, because the forward link (from cell to phone) is a different carrier frequency than the reverse link (from phone to cell).
posted by Steven C. Den Beste at 9:28 PM on April 26, 2007

Papa above referred to this as "full duplex" but that's not correct; "full duplex" means that traffic goes both ways on the same wire, and that's not how the modern phone system works.

Not quite, and thanks for the backhanded slap. Duplex transmissions for mobile work either by frequency division, which is what the TDMA standard does; time division (the CDMA standard supports both), or a few other ways for 3G standards. This is why I said exactly this in my post.

I'm thinking the answer is still very simple. The OP dropped their phone and damaged the antenna, and now only gets spotty reception on an FDD signal, resulting in missing half the conversation.
posted by Cool Papa Bell at 6:29 PM on April 27, 2007

I have more information on this subject. I experience this consistently when calling an associate in California from NJ. At the time I was on Verizon with a Samsung A670. His audio would drop out, I couldn't hear him but he could hear me. I'd hang up and call back and everything would be fine, until the next drop off. I upgraded to a new phone to solve the issue, LG 9900. I experience the exact behavior with the new handset. I can tell you for certain that it is not handset related (no damaged antenna, no bad mike). I have this happen periodically (though not as often), with calls to other lines, both mobile and land, so it's not only his line.

Since both of us are engineers, we've experimented to try to come up with an explanation. Here's the closest we've gotten. We believe that the signal processing drops one side of the conversation if the level gets too low. It's like it's being clipped out. There is an experiment you can do to verify this, which we've done. When my partner drops out and I can no longer hear him, he yells into the phone and the clipping ends. He immediately comes back at full volume. This consistently works; it doesn't solve the problem but it gives a clue.

The problem is most consistent with him, speaking to him on his analog phone line in his 100 year old house (up a mountain in Ca.). I believe that his land line is putting out a signal that is on the hairy edge of the signal processing software. It clips him out when it thinks it lost him but he can wake up the audio stream by raising the volume of the transmission.

I'd be interested in knowing if other people who experience this can duplicate this result. Next time you can't hear someone and you know the call hasn't dropped, tell them to speak very loudly and see if they reappear.

My 2 cents for what it's worth.
posted by gk5591 at 4:56 PM on June 29, 2007