A Little More on 802.11ac
The key factor defining the capabilities these products will have when they hit the market is the chip sets that are at the heart of each vendor's access point.
A couple of weeks back I posted a piece about the buzz around the developing 802.11ac Wi-Fi standard. For those who missed it, 802.11ac is the next-generation Wi-Fi radio link that will eventually replace 802.11n and could potentially push Wi-Fi data rates to almost 7 Gbps through a combination of more efficient signal coding (256QAM vs. 64QAM in 802.11n), wider channels (up to 160 MHz vs. up to 40 MHz in 802.11n), and better MIMO (up to 8 spatial streams vs. a max of 4 in 802.11n). I noted that we're not getting all of that on day one, so I thought I should be a little more specific about what we are getting and when.
The key factor defining the capabilities these products will have when they hit the market is the chip sets that are at the heart of each vendor's access point (AP). While you might be buying APs from Cisco, Aruba, Motorola, Meru, Aerohive, Ruckus, HP, Juniper, or whoever, they are all buying their chip sets from a handful of suppliers that include Broadcom, Qualcomm/Atheros, Marvell, and a few specialized companies.
As I noted in my earlier post, most enterprises won't be jumping to ac in the near term. The problem is the shortage of ac-compatible clients and the possible requirement for significant backbone upgrades to support the increased traffic load. Regardless of that, 802.11ac should absolutely be taken into account in your planning going forward, both with regard to the wired infrastructure and the frequency planning (Note: 802.11ac will only operate in the 5 GHz band).
The 802.11ac products are being rolling out in a "phased approach", and none of the first-wave products will be offering the full set of 802.11ac's defined capabilities. This shouldn't be particularly surprising as we've seen very few products that delivered the full range of 802.11n capabilities, and enterprise-grade 802.11n devices first appeared in 2009!
The "typical" spec for a Phase 1 802.11ac product includes 256QAM in an 80 MHz channel and support for 3 spatial streams. The difference between the Phase 1 products and the full 802.11ac spec is shown in Table 1. Given a maximum bit rate of 433.3 Mbps per stream, that adds up to 1.7 Gbps, almost 4 times the maximum 450 Mbps data rate of a typical 3-stream 802.11n device (in a 40 MHz channel).
Probably the most ingenious development in 802.11ac is Multi-User MIMO (MU-MIMO). MIMO provides the ability to use multiple spatial streams, effectively multiplying the bit rate in a given swath of radio bandwidth. The 802.11n spec specifies a maximum of 4 spatial streams, which means you can send 4 times as many bits as you can with 1 spatial stream. In practice, most enterprise APs top out at 3 spatial streams, for a maximum bit rate of 450 Mbps in a 40 MHz channel. However, in those legacy MIMO implementations, that higher-capacity channel is connected between the AP and one client.
With MU-MIMO, the AP can actually send the four separate streams to four different clients simultaneously. This is particularly important in working with small form-factor devices like smartphone and tablets. One key challenge with small form factors is battery life. Generating 4 spatial streams requires a battery that can drive 4 different radios. The big advance with 802.11ac is that you can use use 1 spatial stream over a wider channel, the most power-efficient way to increase capacity.
With MU-MIMO, the AP can be sending to 4 clients simultaneously. When you consider that most Web traffic is downstream, MU-MIMO can produce an enormous improvement in network efficiency--despite the limitations of the small form factor device. Two of the smaller chip manufacturers, Redpipe and Quantenna, do support MU-MIMO in their current chipsets. Redpipe supports only 3-stream MIMO while Quantenna supports 4 streams.
Unfortunately, support for MU-MIMO requires an upgrade to the client as well as the AP, so we're still waiting for that to occur. Marvell and Quantenna have included beamforming in their chipsets, that should improve overall rate, range, and reliability.
Given the impact that MU-MIMO will have, it is probably wise to wait for the wave-2 802.11ac products that should arrive in mid-2014. Of course, it takes two to tango, so we will also need to monitor availability of client devices with MU-MIMO support--unfortunately, those types of important details are often buried away in the specifications. This next iteration of Wi-Fi will be a major step forward, so it's time to start getting ready.