Wednesday, October 14, 2009

Is Wi-Fi Gunning for Bluetooth? Not Precisely

Wi-Fi Direct is both parallel to and complementary of Bluetooth. Discuss: Today's announcement of Wi-Fi Direct, a peer-to-peer Wi-Fi transfer method, might seem to be firing across Bluetooth's bow. But it isn't quite. Intel's My WiFi is a much more direct threat, and even then may not materialize in quite the way that's being predicted. (Read my coverage, "Wi-Fi Alliance Peers into the Future with Ad Hoc Replacement.")

To review, Bluetooth is a PAN (personal area networking) technology in which devices under the control of the same person or computer communicate over short ranges and relatively low speeds. Bluetooth can create peer-to-peer connections or piconet (host plus up to seven client) networks. In a very standard configuration, a cell phone might use Bluetooth to communicate with a laptop, sharing its 3G mobile broadband connection, while at the same time a Bluetooth earpiece is paired with the phone to handle audio.

Bluetooth requires a pairing process, in which devices authenticate to each other and agree through a handshake (with optional encryption) to talk to one another. The SIG, device makers, and desktop and mobile OS developers have done a great job of simplifying this process down to typically entering a PIN instead of having 20 to 25 steps as it used to be (Secure Simple Pairing).

Bluetooth's current release (2.1+HDR [high data rate]) encompasses a wireless spec for 3 Mbps data transfer (raw) using the 2.4 GHz band. The spec also includes application-layer elements, which are called profiles, and which define a host of specific behavior. This allows any manufacturer to make a Bluetooth keyboard that talks the HID (human interface device) profile and is tested and certified as such can talk to any other Bluetooth device that supports input via HID.

The Bluetooth SIG, which maintains and develops the spec, isn't tied to its physical medium. It's tried to partner with other specs in process to extend itself, notably tying its cart at one point to both major ultrawideband (UWB) encodings, and then picking WiMedia, which was the "winner" in UWB. WiMedia disbanded, but handed off the Bluetooth component to the SIG; there may still be life in it. (Originally, Intel et al. wanted to stick one UWB radio in computers and devices and overlay many specs on top, like Bluetooth, IP, Wireless USB, and video.)

While UWB fiddled and burned, however, the SIG worked on Bluetooth 3.0+HS (High Speed), which incorporates a high-speed transfer mode that allows a Bluetooth device to coordinate a switch with a peer to 802.11 (802.11g, but not Wi-Fi, notably) for a bulk transfer for a large file or high-speed streaming. The session is still within the structure of a Bluetooth PAN, and the use of 802.11 is entirely under the control of the Bluetooth session. The devices don't suddenly become ad hoc nodes or soft access points.

Wi-Fi Direct is an outgrowth of the interest by Intel and others in reducing the number of radio technologies and the level of complexity in devices, which can correspondingly reduce battery usage, while also developing a spec that's to their liking. Intel has a board seat on the Wi-Fi Alliance and the Bluetooth SIG, but still enjoys charting its own course.

Wi-Fi Direct is a peer-to-peer technology, at least the way it's being described initially. Wi-Fi devices that have services to offer (like printing, file sharing, etc.) can advertise those in a way that other equipped devices can access directly. This new method offers the speed and security of an infrastructure Wi-Fi network with an access point at the center without the overhead of joining such a network or making such networks public to allow access to specific resources. That is, someone can print to your printer without you giving them a key to your network. Wi-Fi Direct is built on top of 802.11n, so it can work in both 2.4 and 5 GHz, too.

The simplicity of Wi-Fi Direct is supposed to aid in devices without keyboards or easy data entry methods, much as Wi-Fi Protected Setup (WPS) was supposed to offer a one-click secure connection. With a peer-to-peer approach, a camcorder could hook up with a laptop to transfer data directly without you needing to enter a WPA2 Personal passphrase or even connect at all to an existing Wi-Fi network.

Beyond speed and security, Wi-Fi Direct will allow an adapter to be scanning and accessing peers while also maintaining a full infrastructure connection to a network. It's this feature that allows devices to ostensibly cut the Bluetooth "cord," although I'm still dubious about that as a general element, as I'll explain.

The My WiFi technology that Intel developed (apparently at least in part with Ozmo Devices) emphasizes more of the PAN aspect, talking about having eight devices associated with a laptop, for instance.

So, the question at the outset was whether Wi-Fi Direct is a competitor to Bluetooth?

Bluetooth and Wi-Fi Direct definitely compete head to head on trying to make the simplest network connection between two devices for a variety of straightforward purposes.

However, Wi-Fi Direct won't be backward compatible to the hundreds of millions of devices on the market that already have Bluetooth 1.x or 2.x. Bluetooth's later flavors (2.x and 3.x) are backwards compatible with those older devices.

And while Wi-Fi with a PAN mode could reduce circuit counts, most Wi-Fi chips that are being sold in the mobile market, and I believe in the desktop/laptop market, are integrated Bluetooth/Wi-Fi modules that often throw in other radios and circuitry as well.

Wi-Fi may eventually be appropriate to build into keyboards, mice, wireless headsets, earpieces, and other low-battery peripherals, but that's not really the case today. Bluetooth dominates there in hundred of millions of installed devices.

Bluetooth's profiles also seem like an advantage to me. Kelly Davis-Felner, the Wi-Fi Alliance's marketing director, said that Wi-Fi Direct would not have application or task overlays, but would be focused on the networking and communication level, as with other Wi-Fi certifications.

Which means that if I connect my mobile phone with my computer to transfer music over, I still need an application on both sides that handles the file transfer. With Bluetooth, the profiles still need an interface on top, but a universally supported file-transfer method already exists. I can use a Bluetooth program under Windows and on the Mac and within various mobile phones to transfer files today.

If I want a method that synchronizes stored files and handles it automatically, then OS makers or third-party developers still do have to build an application on top of that. But with Bluetooth, they can rely on having the mechanism in place. It's asymmetric, in that a desktop OS program for syncing MP3 files or photos doesn't require a corresponding program to be installed on a mobile phone that allows access to its storage via the Bluetooth profile.

Now, of course, I'm being a little disingenuous about profiles, because Wi-Fi Direct will create an IP-based network between the two parties, so existing service discovery methods available (like Apple's Bonjour and whatever the current name of Microsoft's technology is) will work just as they do today. But none of these methods aren't supported across mobile operating systems, gadgets (like cameras), and multiple operating system platforms. That's going to be the challenge for Wi-Fi Direct.

In the end, I certainly see Wi-Fi Direct as provoking additional industry efforts to figure out precisely what's useful about PANs and how to package this better for consumers.

The best thing about Wi-Fi Direct is that it enables a secure, high-speed ad hoc mode that will actually work among different devices, something that's long been needed.

Wi-Fi Alliance Peers into the Future with Ad Hoc ReplacementThe Dub Pistols to headline ICMP Graduation