Atheros talks about WAPI, 802.11N, and standards: Atheros recently announced that they would be the Wi-Fi chip supplier for Lenovo, the Chinese firm that plans to take over IBM's personal computer division. While that takeover may be temporarily held up due to politicking, it's still likely to go through, and gives Atheros a foothold in a new market.
But this relationship reminded me of the WAPI (WLAN Authentication and Privacy Infrastructure) debate last year in which China was about to require this proprietary and closed security standard in all domestically sold Wi-Fi adapters and equipment. After high-level intervention by the U.S. government, China backed off from the requirement.
My objections to WAPI stemmed from its closed nature: many in the encryption world believe that closed crypto is made to be broken because it cannot be tested in a robust fashion. It's also extremely easy to provide back doors, which can be used for government monitoring as well as being themselves weak points of entry. The U.S. government wanted explicit backdoors in encryption in the 1990s; there's no reason China wouldn't want that right now.
I had the chance to speak to Atheros's Sheung Li, their product line manager, about both the WAPI standard and the progress in Task Group N (802.11n), the high throughput standard underway at IEEE as a successor to 802.11g.
The interview appears after the jump.
WAPI: Li clarified two points: first, WAPI isn't dead, but rather is an optional requirement for domestic gear. China may choose, for instance, to require hotspots to use WAPI--which includes authentication elements such as are found in 802.1X--but individual users and buyers aren't being forced to use it versus, say, WEP or WPA.
Second, the Lenovo deal is largely about export, and Atheros isn't required to implement WAPI as part of it. But they're not concerned, either, as they see it as just another technology standard they need to be involved in. Li pointed out that China has submitted WAPI to the ISO, an international standards body, and it's possible that they will open the specification to examination as part of that process, although that's not required nor guaranteed.
Li made a rejoinder to the issues around back doors in WAPI by pointing out that the AES standard that is part of the higher strength encryption available in WPA2 in the near future has been approved by the NSA.
This makes some in China nervous, as it may indicate that the NSA has the ability to break weaker AES keys. In fact, various analyses around the Internet support this concern in part: there are several key lengths for AES, and the NSA doesn't give the shortest key very enthusiastic support as it does for two longer versions.
Without Li or myself positing any particular conspiracy or collusion, this is clearly some of the thinking that has led the Chinese government to develop its own standards for many technologies. This doesn't obviate the issues of back doors or closed standards, but it gives me more insight into the industrial espionage and political intrigue that finds itself embedded in technology.
802.11n and TGn Sync: Atheros is part of the TGn Sync proposal group at the IEEE 802.11 Task Group N, and has the highest votes of any proposal yet. Before the most recent meeting, Qualcomm dropped its separate proposal and joined TGn Sync, while the Motorola proposal was taken off the table as well. This leaves WWiSE--arch-rival Broadcom, Texas Instruments, and several other firms--contending with TGn Sync.
Both proposals involve high throughputs, but TGn Sync pushes the envelope to the edge with a greater number of optional configurations, including 10, 20, and 40 MHz-wide bands and up to four antennas. WWiSE has similar options, but isn't stressing the highest speeds or antenna combinations in my reading of their proposal and intent. They appear to stress better signal properties and a greater compliance with existing international regulatory requirements.
TGn Sync has a larger alliance of members, now including Mitsubishi and Qualcomm, as well as companies like Cisco and Intel. There's no bets on it winning in the next round of balloting at which it needs to receive 50 percent in March to move ahead, but they may win on sheer numbers at this point.
Li explained that the 50 percent vote doesn't leave the losing proposal out in the cold. The meetings that follow a majority winner are taken up with revising a proposal until it receives at least 75 percent in balloting, which means that WWiSE or TGn Sync adherents can hold up the process in obtaining specific technical improvements or options that they believe are necessary.
I asked Li about the existence of so-called Pre-N equipment or equipment that just calls itself plainly MIMO without any standards reference. He said, in regard to this equipment--which Broadcom and Atheros are not offering chipsets for--"the only guarantee is that no pre-N product will talk to any other pre-N product in the pre-N mode."
(I should note that I and other reporters have tended to conflate Pre-N and MIMO: MIMO gear itself is a perfectly reasonable way to extend range at some additional cost. But anything identifying itself as Pre-N is writing a check it can't cash, unlike, for instance, pre-ratification 802.11g in January 2003 which was firmware upgradable to the final ratified specification.)
Li is concerned that there may be a number of dead ends in the early MIMO gear because MIMO is just one aspect of 802.11n, and is essentially a given right now that the ultimate solution will involve MIMO technology. "MIMO is one of 15 different technologies in 802.11n," Li said. "We spend basically zero time talking about MIMO because" the engineers in the group know how to implement MIMO.
The thinking at TGn Sync, Li said, is that the speed boost from 802.11g has to be significant. A leap from 21 Mbps of throughput to about 100 Mbps isn't very big, he said. "You need to have something usable in the one to two hundred [Mbps] range." TGn Sync could deliver well over 200 Mbps of throughput in its full implementation.
TGn Sync wants full support for lower speed flavors of 802.11n without sacrificing the very highest end. For instance, Li favors adoption of a 40 MHz encoding as part of 802.11n to avoid having to upgrade it from 20 MHz to 40 MHz after the standard is released to get the speeds that will then be desired. It's a bit of futureproofing.
I asked Li whether delays in 802.11e, the quality of service and multimedia extensions group, was provoking work in 802.11n. "E stands for endurance," Li said, referring to 802.11e's long lifetime without a ratified result, although some interim work has been released. But he agreed that having a bigger pool of bandwidth can reduce some of the need for elements of 802.11e that try to schedule or prioritize packets.
Finally, Li noted that 802.11n has a strange advantage that hasn't been mentioned: the faster a wireless networking standard operates, the less power it uses because the radio is active for less overall time. "Higher speeds mean lower battery usage," Li said. This might allow 802.11n to find its way into ever smaller devices because of power improvements.