During the launch of the fourth-generation iPad and iPad mini on 23 October 2012, I couldn’t help but prick up my ears when I heard Phil Schiller state, at least twice, that these new models had Wi-Fi that was “up to twice as fast” as earlier models.
My immediate supposition was that Apple had upgraded the 802.11n chips to models that could handle two “spatial streams.” One of the speed boosts in 802.11n over earlier versions was its capability, using multiple antennas (known as MIMO for “multiple in, multiple out”), to send unique streams of wireless data that take different paths through space. All iOS devices were single stream (really, SISO: “single in, single out”), because that requires a simpler radio system and simpler antenna architecture.
But, no, Apple actually chose a course that doesn’t truly provide “up to twice” the throughput except in limited circumstances. The company’s iPad features page trumpets “advanced Wi-Fi technology,” which is odd, as it would mean that Apple has upgraded to technology it has offered since 2006 in Macs! Apple writes, “With dual-band (2.4 GHz and 5 GHz) 802.11n Wi-Fi and support for channel bonding, download speeds can reach up to 150 Mbps.”
There are problems with that statement, as clear as it might seem. Channel bonding uses what Apple previously called “wide” channels: it takes two separate 802.11n channels and treats them as one. This feature has been in all Macs that were sold with 802.11n (whether initially or through a later firmware update in 2007), and is offered in all 802.11n Apple base stations, too. The raw single-stream data rate for 802.11n is 75 Mbps; double the channel width and you can get a raw rate of 150 Mbps. In practice, the improvement is never quite that much. In my testing, I have often seen 30 to 35 Mbps for an interference-free short-range connection using a single stream, and more like 50 Mbps with a wide channel.
But beyond the raw rate is where wide channels may be used. Although the IEEE 802.11 committee that approved 802.11n allows wide channels in 2.4 GHz, and the Wi-Fi Alliance that certifies gear with the Wi-Fi label offers it as well, it’s rarely used. The 2.4 GHz band is crowded with baby monitors, cordless phones, Bluetooth, and other Wi-Fi networks. The algorithm that enables wide channels has three separate tests in 2.4 GHz to make sure it’s not “talking over” another network by bonding channels.
In most cases, circumstances don’t allow wide channels to kick in with 2.4 GHz networks, and the Wi-Fi Alliance has been making it harder to use in 2.4 GHz. Further, Apple never enabled wide channels for 2.4 GHz in any 802.11n base station it has sold. (There used to be a switch to disable channel bonding in 5 GHz, but that has been removed in AirPort Utility 6.)
So the wide channels are typically available only for use on base stations in the 5 GHz band, where there are more channels in most countries than with 2.4 GHz, and where the channels are more spread out. But even though that frequency space is available, 802.11n still employs a few mechanisms in the 5 GHz band (less restrictively than in 2.4 GHz) that prevent bonding channels in many cases, too. So long as there are other 5 GHz networks nearby, it’s unlikely that wide channels will be available. (The ugly technical details are in a 2007 post to my defunct Wi-Fi Networking News site.)
In the end, the “up to 150 Mbps” or “2x” improvement is a sort of silly marketing point that should have been confined as a note in the tech specs and not trumpeted as “advanced,” “new” or, in fact, in any way special during the launch event. Adding two-stream or even three-stream support to future iOS devices would have a greater impact on improving throughput than this channel bonding tweak affords.