You may be overcharged for data in poor coverage areas

From:
http://www.extremetech.com/mobile/1...ill-youre-probably-being-overcharged-for-data

A recent study conducted by UCLA professor Chunyi Peng shows that carriers generally count data usage correctly, but those customers who commonly use their device in areas with weak signal strength or to stream audio or video are often overcharged. Peng and three other researchers used data gleaned from an app installed on Android smartphones on two different carriers.

The issue appears to be in how the system is set up to count data usage. Under the current scenario, data is charged as it is sent from the carrier’s network to the end user. What does not exist is a system to confirm whether the packets are received, and thus preventing charges for unreceived data.

Click to expand...

The study (PDF):
http://www.cs.ucla.edu/~chunyip/publications/mobicom12-peng-accounting.pdf

Removing unconfirmed packets from the bill would require counting at higher network layer. TCP is the lowest one with built-in secure delivery. But not all traffic is TCP. Ping for example, but many other applications don't use TCP. Streaming protocol do not offer secure delivery, it is a sort of broadcasting scheme. So, I guess, someone using such protocols accept risk of being overcharged. Other matter is education of the users about counting methods and this is what Cellphone companies don't do.
I am little bit concern about popularisation of DNA protocol for malicious use, intentionally disturbing network by bogus requests. Potential for new type of trojan horse loaded to the handset in the form of nice application - Cellular virus?

sajunky said:

Removing unconfirmed packets from the bill would require counting at higher network layer. TCP is the lowest one with built-in secure delivery. But not all traffic is TCP. Ping for example, but many other applications don't use TCP. Streaming protocol do not offer secure delivery, it is a sort of broadcasting scheme. So, I guess, someone using such protocols accept risk of being overcharged.

Click to expand...

Maybe it needs to be done at a lower level - it might be that only at the radio level is the network aware of lost radio packets.
It also may just be impossible to measure.

sajunky said:

Other matter is education of the users about counting methods and this is what Cellphone companies don't do.

Click to expand...

We will have to educate the users here.

If you are using a wireless internet connection from a fixed location you should invest in an antenna, especially if you are in a poor coverage area.

Which level of the OSI stack is being used for charging is less of an issue than WHERE its being charged in the network. Typically charging occurs on the GGSN, which is located close to the Internet breakout point. From here, to reach the consumer it must still go through the SGSN, base station controller and base station. This could be quite a distance with the ip network and transport systems linking these elements. Even if packets are lost along this path it was already charged.

Mier said:

Which level of the OSI stack is being used for charging is less of an issue than WHERE its being charged in the network.

Click to expand...

I missed a point or you didn't try to explain why. Yes, ginngs already pointed correctly that if counting were at radio station (which is the largest contributor to packet lost) it could improve accuracy. This is assuming Internet not losing packets, which is not true.
My point is that with counting at TCP layer, no matter WHERE is made, it would eliminate overcharging (at least for TCP packets, but not UDP or other protocols - thats a weak point).

My point was that typically you are charged at the point where the data enters the carriers network, not at the point where it's "sent from the carrier’s network to the end user" as the article put it. There are many systems in between and any packet loss in those systems will also contribute to over-charging, not just poor radio conditions.

I agree that TCP could potentially be more accurate, but that will only be the case IF the operators network is protocol aware and IF they exclude TCP retransmission from the packet count.