Can toggling airplane mode actually improve your 3G data speeds?
by Andrew J. Shepherd
Sprint 4G Rollout Updates
Friday, August 17, 2012 - 1:14 PM MDT
CDMA1X and EV-DO carrier channels are shared resources. In CDMA1X, many subscribers share the same carrier channel, their individual traffic kept theoretically orthogonal by code division. Likewise, in EV-DO, individual traffic is separated by time division. But what happens when Sprint (or any other CDMA2000 network provider) has deployed greater than one CDMA1X and/or EV-DO carrier channel on a given cell site? How does your handset determine which carrier channel to utilize?
You might like to think that your handset would automatically choose the least loaded CDMA1X and/or EV-DO carrier channel. But that is not really the case. Instead, when multiple carrier channels are available, each cell site broadcasts a channel list message of the available carrier channels on that site. Upon receiving this list of multiple carrier channels, each handset then invokes a hashing algorithm to select which carrier channel to use. Think of it like a multi lane highway, but each car must choose a particular lane based on the car's license plate number.
For CDMA1X, the hashing algorithm -- which is a kind of pseudo random number generator -- is seeded with the handset's ESN or the subscriber's MDN/MSID (i.e. phone number). Unless the subscriber changes devices or phone numbers, these values remain static, hence the carrier channel hash is quite predictable. And Sprint, for reference, seems to use MDN/MSID based hashing. Nearly a decade ago, I built a spreadsheet that emulates the CDMA1X hashing algorithm, downloadable as an XLS file.
However, for EV-DO, the carrier channel hash is not quite so outwardly predictable. To seed the hashing algorithm, EV-DO uses a session number, which obviously varies from data session to data session. Each time that a handset powers up, crosses a SID/NID boundary, or even toggles airplane mode, for example, generates a new EV-DO data session, hence a new session number. And it is this session number that determines the output of the hashing algorithm.
To demonstrate this process, I positioned myself in one location about a quarter of a mile distant from the north sector of a local cell site. Over the course of several minutes, I grabbed three screen caps of the EV-DO engineering screen on one of my handsets. In between each screen cap, I cycled airplane mode at least once, each cycle generating a new data session. In the span of four minutes, I was able to get my handset to hash to each of the three EV-DO carrier channels deployed on this site. When I arrived at the site, my handset hashed to PCS 0175, which is the third EV-DO carrier channel (F3) in the channel list message. The second and third hashes after toggling airplane mode several times were to PCS 0150 (F2) and to PCS 0100 (F1). See the Channel Number field depicted in the screen caps:
In addition, here is a raw RF look with a spectrum analyzer at the seven CDMA2000 carrier channels deployed on this cell site sector:
The four CDMA1X carrier channels are PCS 0050, PCS 0075, PCS 0125, and PCS 0200. As is oft the case, the three aforementioned EV-DO carrier channels -- PCS 0100, PCS 0150, PCS 0175 -- are distinguishable by their slightly higher RF power output. Furthermore, for those curious, PCS 0025 (at the far left of the graph) and PCS 0225, PCS 0250, and PCS 0275 (at the right of the graph) are fallow spectrum on this site. If deployed, PCS 0025 would be the next EV-DO carrier channel (F4), PCS 0275 the final EV-DO carrier channel (F5), while PCS 0225 and PCS 0250 would be additional CDMA1X carrier channels.
Back to the hashing algorithm, while it attempts to distribute users more or less evenly among available EV-DO carrier channels, it does not take into account several other factors, such as loading and backhaul. For example, if you are stuck on a carrier channel and sector with a few data hogs who have stronger signal than you do, your data speeds will likely suffer as the "fair and proportional" scheduler integral to the EV-DO airlink attempts to maximize total throughput by allocating greater time slots to the users with better signal quality. Additionally, backhaul may not be distributed evenly among deployed carrier channels, so it is possible that some carrier channels may have inherently greater data capacity than others do.
Another benefit of rehashing to a different carrier channel is that you may be able to connect to a closer cell site. Because not all cell sites have the same number of deployed EV-DO carrier channels, carrier channel hashing is an imperfect process. To illustrate, the cell site (call it cell site "A") that I detailed above for this trial has three EV-DO carrier channels (F1, F2, F3), as duly noted. But the adjacent cell site to the north (call it cell site "B") has only two EV-DO carrier channels (F1, F2). A handset that hashes to F3 on cell site "A" will cling to carrier channel PCS 0175 even as it moves north well into the coverage area of cell site "B." Interference will not be a problem, as cell site "B" does not transmit PCS 0175, but signal strength (and data speeds) will diminish until cell site "A" drops below a network defined threshold, at which point the handset will handoff to cell site "B" and hash to PCS 0150. This can require substantial movement and/or time. So, if you always want the most crisp EV-DO handoffs, you can try to ensure that your handset always hashes to F1, the EV-DO carrier deployed on essentially every site in the market.
To conclude, by no means is airplane mode a panacea for slow 3G data ills. EV-DO carrier channel deployment and backhaul can vary from site to site, while loading can also vary from site to site, even from minute to minute. And EV-DO networks in some cities are just generally overloaded. But if you are at work, in a restaurant, at a park, etc., and find yourself with unbearably slow 3G data or lower than usual signal strength for that location, try toggling airplane mode. A 30 second on/off cycle of airplane mode will start a new data session and could get your handset to rehash to another EV-DO carrier channel that is on a closer site, has better backhaul, and/or is currently less loaded.
Sources: Qualcomm, author's field data
- 21
12 Comments
Recommended Comments